Volume 2 • Number 5 • 1999

VALUE IN HEALTH

© ISPOR 1098-3015/99/$14.00/350 350–408

350

ISPOR Second Annual European Conference Contributed Abstracts

Contributed Podium Presentations

ECONOMIC & OUTCOMES STUDY RESULTS OF CARDIOVASCULAR DISEASE

CV1

LONG TERM COMPLIANCE WITH ANTIHYPERTENSIVE DRUG THERAPY IN A MANAGED CARE ORGANIZATION

Simons R

Millennium Biostatistics & Health Economics, Millburn, NJ, USA

Evidence is accumulating that long term compliance (LTC)may be better with newer antihypertensive agents that havebetter safety profiles.

OBJECTIVES:

To evaluate whether newer, better tolerateddrugs improve LTC with antihypertensive medications.

METHODS:

We used pharmacy and healthcare adminis-trative claims data from United Healthcare (1994–1997)to compare retrospectively LTC by antihypertensive drugclass. The study population consisted of hypertensive pa-tients on monotherapy from six U.S. HMOs, and enrolledin the plan for the duration of the 18 months of follow up.We assessed patients on diuretics, beta-blockers (BB),CCBs, ACEs, and ARBs. We estimated a two-stage modelthat measures LTC by drug class while adjusting for physi-cian’s decisions to select a particular drug due to the pa-tient’s medical history. Logistic regressions were used firstto estimate physician’s prescribing decisions based on thepatient’s medical profile. Second, Cox regression was usedto estimate LTC for each drug class while controlling forphysician’s choices, plan type, HMO geographic location,and patient demography.

RESULTS:

The study entry criteria provided for analysisof 9,002 patients with mean age of 54 years and 52%male. By drug, patients were distributed as diuretics 342(3.7%), BB 316 (3.5%), ACEs 3516 (39.0), CCBs 4514(50.1%), and ARBs 314 (3.4%). LTC was significantlybetter for patients treated with ACEs and CCBs comparedto diuretics and BB (61% vs 37% at 1 yr). ARBs showedthe most favorable LTC (90% at 1 yr). The log rank testsindicated that LTC were statistically different (better) forpatients treated with ACEs and CCBs compared to diuret-

ics and BB (p

,

0.001) and statistically different (better)for patients treated with ARBs compared to ACEs andCCBs (p

,

0.001).

CONCLUSION:

These results indicate that choice of anti-

hypertensives plays an important role in LTC and neweragents with improved tolerability like ARBs provided thebest LTC.

CV2

AN ECONOMIC ASSESSMENT OF DALTEPARIN VERSUS WARFARIN IN TOTAL HIP REPLACEMENT

Pleil A

1

, Calais M

1

, Vreeland M

2

, Reinhart S

2

, Mummert L

2

, Trotter J

2

1

Pharmacia & Upjohn, Stockholm, Sweden;

2

Ovation Research Group, Highland Park, IL, USA

OJECTIVES:

To assess the relative cost-efficacy of low-molecular-weight heparin thromboprophylaxis using dalte-parin continued beyond hospitalization and warfarin duringhospitalization in patients undergoing total hip replace-ment (THR) surgery.

METHODS:

Patients in this double blind, multi-centerclinical trial received either initial (P1) or extended (P2)thromboprophylaxis post arthroplasty. P1 patients wererandomized to receive dalteparin pre-operatively (Dpre),dalteparin post-operatively (Dpost), or warfarin (Warf) for6

6

2 days. In P2, patients continued with dalteparin (Dpre,Dpost) or placebo (Warf) through 35

6

2 days. Economicoutcomes were based on medical interventions associatedprimarily with suspected deep vein thrombosis (DVT) and/or symptoms of pulmonary embolism (PE) with onset dur-ing the study. Total inpatient length of stay (LOS) was in-cluded in calculations of total costs. The thromboprophy-laxis treatment regimen and a composite intervention forall major events were valued based on local resource costs.These costs were then applied to specific patient cases.

RESULTS:

A total of 425 patients in the United States(U.S.) and 705 patients in Canada were in P1; 93 U.S. and532 Canadian patients continued to P2. The incidence ofsuspected DVT was higher in the Warf group in P1 and P2(p

5

.002; p

5

.001). No differences were detected inLOS. In P1 and P2 respectively, more dalteparin patientsreceived blood transfusions (p

5

.001; p

5

.001). Overallper patient costs (U.S. dollars) were similar in P1 (Dpre

5

$6667, Dpost

5

$6467, Warf

5

$6339) and in P2 (Dpre

5

$6060, Dpost

5

$5976, Warf

5

$5718). Incremental cost-efficacy for dalteparin ranged (in U.S. dollars) from $1799to $4661. An analysis by country yielded similar results.

Abstracts

351

CONCLUSIONS:

Despite the higher acquisition cost,thromboprophylaxis with dalteparin is cost-efficacious ver-sus warfarin in preventing DVT after THR.

CV3

A COST COMPARISON STUDY OF AMLODIPINE AND ENALAPRIL AS TREATMENT FOR HYPERTENSION IN THE UNITED STATES

Doyle J, Arikian S, Casciano J, Amsel A, Casciano R

The Analytica Group, New York, NY, USA

OBJECTIVES:

To compare the cost-effectiveness for treat-ment of mild and moderate hypertension using either am-lodipine or enalapril in the United States. The study isbased on data from a randomized controlled clinical trial,conducted with 461 patients.

METHODS:

We used the raw data from a one-year,double blind clinical trial of amlodipine versus enalaprilto quantify the treatment dosages associated with eachpatient group. The amlodipine group consisted of 231patients, and the enalapril group included 230 patients.Within the clinical trial, there were no between group sig-nificant differences for age, gender, concomitant medica-tions and current illness at randomization. Mean dosagesof amlodipine and enalapril, as well as the frequency anddosage of diuretic use, were calculated between groupsbased on age, gender and study phase, and were used toestimate the cost of treatment. Efficacy and adverse eventrates were also calculated for each group to compare theoutcomes of the therapy.

RESULTS:

Amlodipine ($551.62 per patient annual drugtreatment cost) was less expensive in treating the hyper-tensive patients within the study as compared to enalapril($663.48 per patient annual drug treatment cost). Fur-thermore, a drug price sensitivity test found that with asmuch as a 17% reduction in the cost of enalapril, amlo-dipine would remain a lower or equivalent treatment interms of cost. The mean final visit dose amounts of drugper patient were 7.2 mg/day for amlodipine, and 28 mg/day enalapril. The total reduction in sitting DBP was16.9 mmHg with amlodipine and 16.2 mmHg with enal-april. However, significantly (p

,

0.05) more patients inthe enalapril group (n

5

46, 20%) required the use of adiuretic (HCTZ) to attain control of DBP than in the am-lodipine group (n

5

27, 11.7%). Finally, there were nosignificant differences (p

,

0.05) in adverse events be-tween groups (amlodipine

5

21.2%; enalapril

5

17.4%).Yet, the type of adverse event differed by drug where asignificant effect (p

,

0.001) of amlodipine and edema(15.2%) was found, and a significant effect (p

,

0.001)for enalapril and cough (7.4%) was found.

CONCLUSION:

This study suggests that amlodipine is aless costly therapy compared to enalapril, and hence apotentially more cost-effective treatment for mild andmoderate hypertension.

CV4

THE COST OF HYPERTENSION AND ITS CORRELATES IN EMILIA ROMAGNA REGION (ITALY): RESULTS FROM THE GREAT STUDY

Ambrosioni E

1

, Belisari A

2

, Bustacchini S

3

, Costa FV

1

, Greco S

2

, Ruffo P

3

, Mantovani LG

2

and the GREAT (GRoup on Economic Assessment of hyperTension) investigators

1

Clinica Medica III, Policlinico S. Orsola, Universita di Bologna, Bologna, Italy;

2

Centro di Farmacoeconomia, Instituto di Scienze Farmacologiche, Universita di Milano, Milan, Italy;

3

Pfizer Italiana, Rome, Italy

OBJECTIVE:

Evaluation of the economic impact of hy-pertension and its correlates.

METHODS:

Twenty-seven GPs and 9 specialist centersparticipated in this longitudinal study. Information onlifestyle, SBP, DBP, comorbidities, diagnostic and lab ex-ams, hospitalizations, physician’s visits, drug and medicaltherapies was collected. We report on healthcare costs(hospitalizations excluded), quantified in the perspectiveof the Italian National Health Service (NHS), by meansof tariffs expressed in Italian Lire 1998, and referred to athree-month period. We used multivariate linear regres-sion to investigate the association between healthcarecosts, drug cost (or proper transformations) and the levelof SBP and DBP. P

,

0.05 were considered statisticallysignificant.

RESULTS:

416 patients were assessed, 210 women (meanage 63

6

15) and 206 men (61

6

12). The mean levels ofSBP and DBP were 149

6

11 and 90

6

9 mmHg for menand 149

6

15 and 88

6

9 mmHg for women. The totalthree months cost of hypertension care was Lit 294.221for men and Lit 253.938 for women (NS, Mann WhitneyU test). Drugs represented the largest part of costs (58%),possibly due to the short time span. We found that newpatients tend to be less costly (P

5

0.006). Patients en-rolled by specialists centers (P

,

0.001); patients with co-morbidities (P

,

0.001); smokers (P

5

0.002) and previ-ously hospitalized patients were more costly (P

,

0.001).No influence of age and sex was detected. The health carecost of hypertension was associated with the level of SBP(P

5

0.007) and DBP (P

,

0.001). The cost of drug ther-apy was significantly higher in patients with higher SBPand DBP levels (P

,

0.001 and P

5

0.005 respectively).

DISCUSSION:

Hypertension is a quite costly disease.Healthcare costs of hypertension and the costs of anti-hypertensive drug therapy are associated with the levelboth of SBP and DBP. Interventions effective in controllingSBP and DBP are likely to affect costs.

CV5

SIMULATED LIFETIME COSTS TO NATIONAL HEALTH SERVICE (NHS) OF BIOLOGICAL AND MECHANICAL HEART VALVE REPLACEMENT IN YOUNGER VERSUS OLDER PATIENTS

Griffiths R, Duff S, Prosek B, Llana T

Covance Health Economics and Outcomes Services Inc., Washington, DC, USA

352

Abstracts

BACKGROUND:

Selection of biological versus mechani-cal heart valve replacement entails tradeoffs in the risk andexpected cost of post-operative bleeding, embolic eventsand reoperation. Additionally, the risk of these events var-ies by age at implant.

OBJECTIVE:

To identify the direct medical costs to theNHS of biological and mechanical valve replacement inyounger versus older patients.

METHODS:

A Markov decision-analytic model was con-structed to identify the cumulative lifetime costs of valvereplacement and related events in a simulated cohort of10,000 patients followed from valve implantation untildeath. Events included bleeding, embolism, endocarditis,structural valve deterioration, reoperation and death. Eventrates were modeled using linear, and non-linear statisticalhazard functions based on clinical series reported in theliterature. Medical resource use related to events was es-timated based on clinical expert opinion. Costs were as-signed to each event using standard lists of NHS costs.

RESULTS:

For aortic valve replacement, the expected life-time costs were £6,812 (biological) and £8,873 (mechani-cal) for persons aged 60, versus £6,281 (biological) and£8,137 (mechanical) for persons aged 70 at implant, re-spectively. In the mitral position, costs were £6,968 (biolog-ical) and £8,760 (mechanical) versus £6,299 (biological)and £7,989 (mechanical) in persons aged 60 versus 70 atimplant respectively. Results were most sensitive to bleed-ing, embolic and reoperation event rates, but less sensitiveto the cost per event.

CONCLUSION:

The expected lifetime cost of biologicvalve replacement was lower than mechanical valve re-placement for both age groups and valve positions. Thissuggests the economic impact of anticoagulation therapy,bleeding and embolic events, which occur at higher rates inthe mechanical valve, is greater than the economic impactof structural valve deterioration leading to reoperation,which is greater in the biological valve.

CV6

COST-EFFECTIVENESS OF STATINS: MOVING BEYOND THE PRIMARY AND SECONDARY PREVENTION DISTINCTION

Caro J, for the CORE Study Group

Caro Research Boston, MA, USA

BACKGROUND:

Economic analyses of cardiovascular dis-ease (CVD) prevention with statins have generated contro-versy on the most efficient allocation of health care funds:primary prevention, secondary or both? Previous analyseshave focused on one setting or the other. Comparing thesetwo oversimplifies the task of allocating health care re-sources and may lead to unjustified decisions concerning“appropriate” statin use. Instead, an integrated view acrossthe continuum of risk is required.

METHODS:

An economic model of CVD prevention withpravastatin—Continuum of Risk Evaluation (CORE)—based on West of Scotland Coronary Prevention Study (pri-mary prevention) and Cholesterol and Recurrent Events

(secondary prevention) study data is detailed. The modelsimulates 10,000 individuals at various stages of CVD(prior manifestation of CVD through multiple events). Allevents are tallied in monthly cycles with costs and life ex-pectancy implications applied appropriately.

ANALYSES:

Analyses were completed for various popula-tions and treatment strategies to help determine the mostcost-effective scenarios. For the purpose of these analyses,populations were described in terms of the proportion ofindividuals at various disease stages at the start of follow-up. Treatment strategies were defined on the basis of therisk cut-off at which treatment is initiated for individualswithout pre-existing disease. Analyses were conducted fol-lowing the NCEP and Canadian treatment guidelines.

CONCLUSION:

An integrated approach to prevention ofCVD is an area that has not been explored in term of itseconomic impact. CORE permits realistic analysis of pol-icy decisions which involve the entire continuum of riskrather then isolated consideration of specific, but arbi-trary, “stages” of disease.

CV7

AMLODIPINE REDUCES HOSPITALIZATION ASSOCIATED WITH TREATMENT OF CARDIOVASCULAR DISEASES IN PATIENTS WITH CLINICAL CORONARY ARTERY DISEASE

Chen G, Byington R, Moran M

Department of Public Health Sciences and Internal Medicine, Wake Forest University School of Medicine, Winston-Salem, NC, USA

OBJECTIVE:

To examine whether amlodipine can reducehospitalization associated with treatment of cardiovascu-lar disease (CVD) in patients with angiographic evidenceof coronary artery disease. Amlodipine is a long acting cal-cium channel antagonist that has been proven to be effec-tive in treating cardiovascular diseases.

METHOD:

We used clinical data derived from the Pro-spective Randomized Evaluation of Vascular Effects ofNorvasc Trial (PREVENT). PREVENT was a 3-year, ran-domized, masked, placebo-controlled, multicenter, clinicaltrial originally designed to test the antiatherogenic effect ofthe calcium channel blocker amlodipine in 825 patientswith coronary artery disease (417 patients for amlodipineand 408 patients for placebo). The outcome measures wereclinical CVD events associated with hospital care which in-cluded congestive heart failure (CHF), myocardial infarc-tion (MI), stroke, angina, coronary artery bypass graft(CABG), percutaneous coronary angioplasty (PTCA), stents,athrectomy, valve replacement, and catheterization.

RESULTS:

Overall, the net number of hospitalization asso-ciated with CVD averted was 27.9 per 100 patients(38.51% reduction) in the amlodipine patient group overthree years of the study. Comparing with the placebo group,the treatment group had fewer hospitalizations related toPTCA (

2

11.75 per 100 patients, 52.69% reduction),CABG (

2

3.03, 42.62% reduction), stent (

2

2.02, 41.22%reduction), angina (

2

9.27, 31.52% reduction), and CHF

Abstracts

353

(

2

2.46, 91.11% reduction). There were no significant im-pacts on hospital utilization related to stroke (

2

0.03 per100 patients), valve replacement (0.48) and fatal MI (0.13).

CONCLUSION:

In this study, amlodipine is found to bean effective agent in reducing hospital utilization related toCVD, especially CHF, PTCA, CABG, stent, and angina, inpatients with clinical coronary artery disease relative to theplacebo. Further analysis will examine whether amlodipinecan reduce hospital care costs associated with CVD.

CV8

DETERMINANTS OF DISCONTINUATIONOF NEW COURSES OFANTIHYPERTENSIVE MEDICATIONS

Gregoire JP

1

, Moisan J

1

, Guibert R

2

, Milot A

1

, Ciampi A

3

1

Equipe de pharmaco-epidemiologie, Universite Laval, Quebec, Qc, Canada;

2

Monash University, West Heidelberg, Australia;

3

McGill University, Montreal, Quebec, Canada

Discontinuation of drug use constitutes a major barrier toadequate control of high blood pressure. Although variousfactors may be associated with discontinuation of treat-ment, so far, very few causal associations have been studied.

OBJECTIVES:

To examine the effect of an array of poten-tial predisposing, enabling and reinforcing factors on dis-continuation of initial antihypertensive medication.

METHODS:

We conducted a prospective cohort studythrough a network of 173 pharmacies across Canada whoidentified individuals starting a new antihypertensive medi-cation as a monotherapy. We excluded pregnant women,individuals who were taking other antihypertensive medi-cations at the same time, and those who took medicationsfor chronic heart failure, or angina. We interviewed sub-jects by telephone four times over a 36 month period. Weanalyzed data using a Cox’s proportional hazard model.

RESULTS:

Out of 682 eligible subjects, 43% had discon-tinued their initial medication at the end of the observationperiod. Individuals more likely to discontinue their initialmedication were those who perceived side effects with theinitial medication (Adjusted Hazard Ratio (AHR)

5

1.91;95% Confidence interval (CI)

5

1.47–2.48) and those whobelieved the antihypertensive medication had no effect(AHR

5

1.33; 95% CI 5 1.03–1.72). Insurance coveragefor antihypertensive medication had a protective effect(AHR 5 0.74; 95% CI 5 0.55–0.99).CONCLUSIONS: Persistence with newly prescribed medi-cations could be improved by selecting antihypertensivemedications with fewer side effects, by changing misleadingperceptions of patients about their treatment, and by re-moving economical barriers.

ECONOMIC & OUTCOMES STUDY RESULTS OF DISEASES OF THE CENTRAL NERVOUS SYSTEM

CN1

COSTS OF DEMENTIA IN GERMANY—AN ANALYSIS OF 7490 HEALTH INSURANCE FILES

Schwarz U I1, Krappweis J1, Rentsch A1, Pirk O2, Kirch W1

1Institute of Clinical Pharmacology, Medical Faculty, University of Technology, Dresden, Germany; 2Novartis Pharma GmbH, Department of Health Economics, Nuremberg, Germany

OBJECTIVE: The economic burden of dementia combinedwith financial crisis in public health systems makes study-ing costs of dementia very important. In Germany only one“bottom-up”-study (150 patient records) of office basedphysicians demonstrated the burden of dementia disease.Our study is the first one in Germany using social health in-surance data to demonstrate the costs of dementia for thesocial health insurance.METHOD: Direct costs incurred in the outpatient sectorfor diagnosis and treatment of dementia were calculated bymeans of analysis of health insurance data from the statu-tory health insurance fund (AOK) of the region Dresden/Riesa. At a defined date (1st July 1993) a random sample of7490 insured persons were selected from the AOK masterfile. The following data was covered: personal data, pre-scription sheets, claim forms with diagnosis, diagnostic andtherapeutic procedures. Information about hospital treat-ment was available from 1st and 2nd quarter of 1994. Plain-text diagnosis was transferred into ICD 10 code. Dementedpatients were identified by ICD items F00 to F09 and I60to I69. A control group of non-demented patients identicalto the dementia group in terms of age and sex was selectedfrom the AOK sample using the matched pair method (twocontrols for each case).RESULTS: 327 of 7490 insured persons were diagnosed asdemented (one-year prevalence: 4.4%). Diagnostic proce-dures and non-pharmacological care interventions were or-dered significantly more often in cases than in controls.79.2% cases vs. 44.2% controls were prescribed at least oneof the drugs associated with dementia. Office based phy-sicians are more often visited by demented than by non-demented patients (23 vs. 90 visits per year). Hospitaliza-tion rate in cases was higher than in controls (11,0 vs.5,1%). This data allowed calculation of the total directcosts covered by social health insurance in one year for a de-mented patient: EUR 4080,64 (diagnostics EUR 1128,52;medication EUR 1906,94; non-pharmacological care inter-vention EUR 107,18; hospital EUR 977,13). Comparedwith controls the following attributed costs per case peryear were found: Total EUR 823,80 (diagnostics EUR247,22; medication EUR 387,65; non-pharmacological careinterventions EUR 93,30; hospital EUR104,64).CONCLUSION: The data indicated, that demented pa-tients are not treated properly. Recommendations for ratio-nal treatment of dementia should be implemented in theoutpatient setting.

CN2

AN ECONOMIC COMPARISON OF OLANZAPINE VERSUS HALOPERIDOL IN THE TREATMENT OF SCHIZOPHRENIA IN FRANCEGregor KJ1, Allicar MP2, Lilliu H3, Bouhassira M2, Le Pen C3,4

1Eli Lilly and Company, Windlesham, UK; 2Eli Lilly and

354 Abstracts

Company, Paris, France; 3CLP-Sante, Paris, France; 4LEGOS, Dauphine University, Paris, France

OBJECTIVES: The purpose of this study is to provideeconomic information comparing olanzapine- and halo-peridol-treated patients from the subset of French patientsthat participated in a large, international, randomised clin-ical trial in schizophrenia.METHODS: Patients were evaluated from initiation untildiscontinuation of treatment or completion of the 54-weekstudy. The primary clinical measure was ‘marked clinicalresponse’ (derived from BPRS total scores). The secondarymeasure was ‘marked clinical improvement’ (derived fromCGI severity of illness scores). The primary economic mea-sure was mean per diem, per patient total direct medicalcosts.RESULTS: A total of 275 French patients were included inthe study. Demographics and other baseline differences be-tween olanzapine- and haloperidol-treated patients were notstatistically significant. Olanzapine-treated patients (205 6142 days) experienced significantly (p , 0.001) longer val-uation periods than did haloperidol-treated patients (132 6129 days). Olanzapine-treated patients (54%) were signifi-cantly (p 5 0.03) more likely to experience a marked clini-cal response than were haloperidol-treated patients (40%).Olanzapine-treated patients (69%) were significantly (p 50.02) more likely to experience a marked clinical improve-ment than were haloperidol-treated patients (54%). Themean per diem, per patient total direct medical costs werestatistically lower (p 5 0.033) for the olanzapine-treatedpatients (F619 6 509) compared with the haloperiodal-treated patients (F756 6 478).CONCLUSION: Olanzapine treatment was associatedwith significantly better clinical outcomes and significantlylower per diem total direct medical costs than was halo-peridol treatment. The findings in this study indicate thatolanzapine is cost-effective compared with haloperidol forthe treatment of schizophrenia. These findings are of in-creased relevance in France and add to the existing, largebody of evidence supporting olanzapine’s cost-effective-ness relative to typical antipsychotics.

CN3

COST-EFFECTIVENESS OF RT-PA IN TREATMENT OF ACUTE ISCHAEMIC STROKE IN UKChambers M1, Kalra L2, Hutton J11MEDTAP International, London, UK; 2King’s College Hospital, London, UK

OBJECTIVE: To estimate the cost-effectiveness of usingrt-PA in treating acute ischaemic stroke in the UK.METHOD: We used a decision-analytic model (StrokeOutcome Model) to estimate the long term health and costconsequences of administering rt-PA to patients with acuteischaemic stroke, based on the pooled results of three ma-jor trials: ECASS-1, ECASS-2 and NINDS (2044 patientsin total). Long-term consequences of stroke were based on

the Oxfordshire Community Stroke Project. Disability wasdefined using the modified Rankin scale (disabled: catego-ries 3–5). Utilities for disability states were taken frompublished studies of stroke survivors. Resource use mea-sured from a health and social service perspective, wasbased on published studies and a panel of clinical experts.Unit costs (1996) were obtained from national sources.Costs and health outcomes were discounted at 6% pa.RESULTS: For patients treated within 6 hours of strokeonset the probability of symptomatic intracranial haemor-rhage (SICH) was 12.6% (rt-PA group) and 3.7% (pla-cebo group). In the absence of SICH the probability ofdeath within 90 days was 10.1% and 14.1% and the pro-portion of survivors who were disabled was 38.2% and49.6% respectively. The model predicted that over 5 yearsfrom the incident stroke patients in the rt-PA group on av-erage lost 0.04 life-years but gained 0.31 disability-freelife-years and 0.13 QALYs. This resulted in cost savings of£2,500–£3,000 per treated patient, depending on howcosts of initial acute care were included. Improvements inhealth outcome and cost savings were greater for thegroup treated with rt-PA within 3 hours of stroke onset.CONCLUSIONS: Trial results suggest that rt-PA is likelyto be cost saving in patients treated within 6 hours of onsetof acute stroke. A comprehensive analysis needs to con-sider the fixed costs and wider policy implications of intro-ducing and maintaining a local thrombolysis service.

CN4

ECONOMIC APPROACH FOR ASSESSING THE OVERALL COST OF MANAGING ISCHEMIC EVENTS (MYOCARDIAL INFARCTION STROKE) IN ATHEROTHROMBOTIC PATIENTS:A EUROPEAN OVERVIEWLevy E1, Gabriel S2, Dinet J2, Carita P2

1Université Paris IX, Dauphine, Paris, France; 2Health Economics & Outcome Research Department, Sanofi-Synthelabo, Gentilly, France

OBJECTIVES: The management of atherothrombotic pa-tients is a heavy burden on the Health budget. As MI andIS patients need acute as well as chronic care, a 2-year pe-riod was considered for studying local treatment patternsand the resulting costs (hospital & ambulatory costs). Theoverall cost of managing an ischemic event (MI,IS) was di-vided in acute (Ac) and follow-up (Fup) costs over 2 years.METHODS: France(F), Belgium(B), Switzerland(CH), Swe-den(SW), Italy(IT), Spain(SP), Portugal(P) and Austria(A)were involved in the study. The management of ischemicevents was analysed during the acute phase and the sub-sequent 6-month periods. A decision tree was used; as-sumptions concerning patient management and resourcesutilization were based on currently available local and in-ternational literature, official national statistics and localexpert opinions (Delphi panel). The costing was per-formed by using Diagnosis Related Groups (IT,SW,P,A),hospital data bases and national tariffs (F, B, CH, SP).

Abstracts 355

RESULTS: Differences between countries about the averageclinical management patterns for a 2-year period are shown.The economic impact concerns the overall cost (Ac 1 Fup)as well as the breakdown between Ac and Fup costs. ForMI, overall costs range from 9513 Euros in Belgium to18294 Euros in Austria. Weights of the follow-up costsrange from 33% in Portugal to 53% in Austria. For IS,overall costs range from 5607 Euros in Austria to 56370Euros in Switzerland. Differences also concern the weight ofthe Fup costs (from 17% in Portugal to 75% in France).CONCLUSION: Differences in overall cost and costbreakdown can be explained by local treatment patternspecificities and by the availability of specific and well-adapted structures for patients rehabilitation. The follow-up of an ischemic event should not be neglected in the glo-bal economic burden assessment.

CN5

THE COST OF INPATIENT TREATMENT OF SCHIZOPHRENIA: A STUDY OF TWO LEADING ATYPICAL ANTIPSYCHOTICSDuchesne I1, Shavit O1, Jones M2

1Janssen Research Foundation, Beerse, Belgium; 2Jones & Just Pty Ltd, Sydney, Australia

OBJECTIVES: To i) compare the relative costs of twoleading atypical antipsychotics and ii) examine the contri-bution of cost drivers of inpatient schizophrenia care.METHODS: Data were pooled from a series of retrospec-tive, single centre studies across 11 centers within 5 coun-tries. Only patients with a diagnosis of schizophrenia orschizoaffective disorder were considered.RESULTS: A total of 601 patients received either risperi-done (RIS; n 5 290) or olanzapine (OLA; n 5 311) as firstline therapy after admission. There were no baseline differ-ences between both groups. The two products showedequivalent efficacy rates (RIS: 78%, OLA: 77%; p 5 0.8)but RIS patients achieved efficacy sooner (median 5 10days) than OLA patients (median 5 18 days; p 5 0.001).The mean modal daily doses were 4.9 mg (RIS) and 14.9mg (OLA). OLA patients were more expensive (USD) thanRIS patients in both daily treatment costs (RIS: mean 53.3; OLA: mean 5 6.5; p 5 0.0001) and daily all medica-tion costs (RIS: mean 5 4.2; OLA: mean 5 7.3; p 50.0001). The contribution of cost drivers was examinedthrough regression on the two key inpatient cost outcomes,daily all medications costs (log transformed) and length ofstay (LOS; censored Cox regression). Daily all medicationcosts were independently affected by prior hospitalizations(1), choice of atypical (OLA higher than RIS) and dose ofatypical (1) and these findings were over and above certainintermediate factors. Length of stay was affected by priorhospitalizations (-) and dose of atypical (1) again over andabove certain intermediate factors. Despite being counter-intuitive, the data clearly indicate that 1st episode patientsstayed longest and the duration of stay continued to reducewith further admissions.

CONCLUSION: The choice and dose of atypical is a ma-jor driver of inpatient therapy cost. In addition the priordisease span and severity, as measured by previous hospi-talizations, is an important predictor of inpatient carecosts.

CN6

EFFECTIVENESS OF PHARMACOLOGICAL TREATMENT OF HYPERTENSION UNDER EVERYDAY CIRCUMSTANCES WITH REGARD TO THE REDUCTION OF STROKE INCIDENCEKlungel OH1,2,3, Stricker BHC2, Breteler MBB2, Seidell JC3, Psaty BM4, de Boer A1

1Department of Pharmacoepidemiology & Pharmacotherapy, Utrecht University, Utrecht, The Netherlands; 2Department of Epidemiology & Biostatistics, Erasmus University Medical School, Rotterdam, The Netherlands; 3Department of Chronic Diseases and Environmental Epidemiology, National Institute of Public Health and the Environment, Bilthoven, The Netherlands; 4Cardiovascular Health Research Unit, University of Washington, Seattle, WA, USA

BACKGROUND: Some observational studies have notconfirmed the beneficial effects of antihypertensive drugtreatment on the incidence of stroke and coronary heartdisease as demonstrated in randomized controlled clinicaltrials (RCTs). This contradiction is probably due to the bi-ased comparison in most observational studies betweentreated hypertensive patients and prognostically differentreference groups such as normotensives or untreated hyper-tensives regardless of their severity of hypertension and co-existence of other cardiovascular risk factors.OBJECTIVE: To assess the effect of the pharmacologictreatment of hypertension under everyday circumstanceson the incidence of stroke.METHODS: Approximately 45 000 men and women aged.20 years were examined in 2 population-based studies inthe Netherlands. A cohort of 2616 hypertensive subjectswho were either pharmacologically treated for hyperten-sion (n 5 1318) or untreated hypertensives who were “can-didates” for pharmacologic treatment on the basis of theseverity of their hypertension and the presence of other car-diovascular risk factors (n 5 983) was followed up for amean duration of 4.6 years and follow-up was complete for2369 (91%) subjects.RESULTS: In this observational study, compared to un-treated hypertensive subjects who were “candidates” forpharmacologic treatment, subjects who were pharmacolog-ically treated for hypertension had, after adjustment for dif-ferences in prognostic factors, a 39% [95% confidence in-terval (CI): 3–61%] reduced risk of stroke. About 46 [95%CI: 29–599] hypertensive patients need to be treated phar-macologically for 5 years to prevent one stroke in the gen-eral Dutch population.CONCLUSION: When a prognostically comparable refer-ence group is used, the pharmacological treatment of hy-pertension under everyday circumstances appears to be ef-fective in the reduction of the incidence of stroke. The

356 Abstracts

relative risk reduction that we found was similar to thosepublished on the basis of results from RCTs.

CN7

REDUCING ANTICHOLINERGIC ANTIDEPRESSANT USE IN THE ELDERLY: A RANDOMIZED TRIAL OF GROUP VERSUS INDIVIDUAL “ACADEMIC DETAILING”Van Eijk MEC1,2, Avorn J3, Porsius AJ1, de Boer A1

1Utrecht Institute of Pharmaceutical Sciences, Utrecht, The Netherlands; 2Health Insurance Company “OZ-zorgverzekeringen”, Breda, The Netherlands; 3Division of Pharmacoepidemiology, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA

OBJECTIVES: Dissemination of medical knowledge andimplementation of knowledge needs to be better under-stood. Academic detailing can effectively change prescrib-ing, but it is very resource demanding. This research com-pares the effect of an educational intervention throughindividual outreach visits and through a group approach,using academic detailing.METHODS: We conducted a randomized controlled trialto measure the effect of two educational programs on pre-scribing (individual intervention arm and group interven-tion arm versus control arm) to reduce anticholinergic anti-depressant (AAD) prescribing for the elderly. We includedall General Practitioners (GPs) and pharmacists (approx.230) in the southwest Netherlands health district and used adatabase covering all prescriptions to persons insured by thenational health insurance system in this area (approx.240,000).RESULTS: In the individual arm 43 of 70 GPs were visited;in the group arm 5 of 7 groups (41 of 52 GPs) agreed to jointhe program. Using an intention to treat analyses there wasa reduction in the incidence of AAD in elderly of 26% (95%confidence interval 24%–48%) in the individual arm andof 45% (8%–67%) in the group arm. The use of non-AADin the elderly increased by 40% (6%–83%) in the individualarm and by 29% (27%–79%) in the group arm.CONCLUSIONS: Both the individual and the group ap-proach significantly decreased the incidence of AAD and in-creased the incidence of non-AAD use in the elderly. Theseapproaches are practical means of improving the clinical ap-propriateness of medication use on a large scale among pri-mary care physicians.

CN8

PRODUCTIVITY COST MEASUREMENT: EMPIRICAL ANALYSIS OF QALY, HUMAN CAPITAL AND FRICTION COST METHODOLOGIESRothermich EA1, Pathak DS2, Salsberry PJ3, Schommer JC4

1The Ohio State University College of Pharmacy, Columbus, OH, USA; 2Ohio State University Center for Health Outcomes Policy and Evaluation Studies, Columbus, OH, USA; 3Ohio State University College of Nursing, Columbus, OH, USA; 4University of Minnesota College of Pharmacy, Minneapolis, MN, USA

Dissention exists among health economists regarding themeasurement and inclusion of productivity costs in a cost-utility analysis. Some health economists believe that produc-tivity costs are captured by quality-adjusted life years(QALYs) in the denominator of the cost-utility ratio. Othersargue that productivity costs should be measured mone-tarily via the friction cost method or the human capitalmethod, and included in the numerator of the ratio.OBJECTIVES: (1) To investigate the extent to whichchanges in productivity costs are explained by changes inQALYs. (2) To examine the difference between the hu-man capital method and the friction cost method.METHODS: An observational panel study with two sur-vey waves was used to collect data. Surveys were admin-istered through face-to-face interviews to a conveniencesample of 102 informal caregivers of the elderly.RESULTS: Changes in QALYs did not explain absencefrom paid labor, reduced productivity during paid labor,and lost leisure productivity costs. However, changes inQALYs did account for ten percent of the variance inchanges in lost unpaid labor (F 5 8.67, P , 0.01). Fric-tion cost estimates were significantly less than humancapital estimates in both the first and second waves of thestudy (t 5 6.96, P , 0.01, t 5 6.47, P , 0.01).CONCLUSIONS: Most productivity cost componentsshould be valued monetarily, and included in the numer-ator of a cost-utility ratio. If the friction cost method isused, rather than the human capital method, smaller esti-mates of productivity costs will be obtained.

ECONOMIC & OUTCOMES STUDY RESULTS OF INFECTIOUS AND RESPIRATORY DISEASES

IR1

THE HEALTH ECONOMICS OF RECLASSIFICATION OF DRUGS FROM PRESCRIPTION ONLY MEDICINES TO PHARMACY STATUSRowlands S1, Devalia H1, Bourke A1, Dean A1, Torgerson D2

1EPIC, London, UK; 2Centre for Health Economics, University of York, York, UK

OBJECTIVES: To examine the likely cost savings to theNHS of reclassification of a medicine from prescriptiononly medicine (POM) to pharmacy (P) status.METHODS: We used the General Practice Research Data-base to look at changes in GP prescriptions for acyclovircream 5% before and after reclassification. To control forthe effects of prescription charges we categorized the datainto three patient age groups: 0–19, 20–59 and 601.RESULTS: There was a 49% reduction in prescribingrate in the 20–59 age group which has no prescriptioncharge exemptions on age grounds. For an average sizedgeneral practice about 65 consultations would be savedannually. Annual savings to the National Health Servicedue to reclassification of one product are calculated to bemore than £5 million.

Abstracts 357

CONCLUSIONS: As well as resulting in cost savings, re-classification will tend to improve equity of access to themedicine. Given the potential for large cost savings to theNHS and society we recommend that the status of POMsbe reviewed on a more frequent basis than the current in-terval of five years.

IR2

THE COSTS OF THE INADEQUATE USE OF ANTIBIOTICSRovira J1, Figueras M1, Caminal J21SOIKOS, Barcelona, Spain; 2Institut de Salut Pública de Cataluny, Barcelona, Spain

OBJECTIVES: To assess the degree of inadequacy of anti-biotic utilisation and its associated costs in PHC centers inCatalonia.METHOD: We used an observational design. The uni-verse of the study was all the visits due to infectious dis-eases to PHC doctors of the 40 basic health care areas ofthe Region “Costa de Ponent”. 2.500 visits were includedin the sample, corresponding to 29 centers and 251 pre-scribers. The source of information was the clinicalrecords. The adequacy of the treatment prescribed was as-sessed with reference to the recommendations of the Cata-lan Society of Family Medicine, adjusted to the pattern ofantimicrobial resistance in the area. The efficiency of theprescription was assessed taking as a reference the mini-mum cost required by an adequate treatment that could beachieved by using the cheapest alternatives.RESULTS: The number of valid observations (visits due toinfectious diseases) was 2,470. In 1,349 cases (54.6%) anantibiotic was prescribed. The treatment prescribed was ad-equate in 56.3% of the cases. The treatment was assessed asinadequate in 43.7% of the cases. In 92 cases an antibioticwas required and not prescribed and in 680 cases an antibi-otic was not required but prescribed and in the remaining294 cases an antibiotic was required, but the antibiotic pre-scribed was not the recommended one. The total cost of theantibiotics prescribed was 3.5 million PTA. of which 2.4million (68.7%) could apparently have been avoided. 1.9million corresponded to clinically inadequate prescriptionsand the remaining 0.5 corresponded to clinically adequateprescriptions that used brands of higher price than the low-est priced alternatives available on the market.CONCLUSIONS: Inadequate use of antibiotic causes un-necessary costs to the health system and increases antibi-otic resistance. Action should be taken to improve the ade-quacy and efficiency of prescribing habits.

IR3

COST OF NOSOCOMIAL PNEUMONIA IN A LARGE TEACHING HOSPITALDietrich ES1, Demmler M1, Schulgen G2, Mast O3, Pelz K4, Fekec K2, Daschner F1

1Institute of Environmental Medicine and Hospital Epidemiology, University Hospital Freiburg, Freiburg, Germany;

2Institute of Biometry and Medical Informatics, University of Freiburg, Freiburg, Germany; 3Health Economics and Outcomes Research, Bayer Vital, Leverkusen, Germany; 4Institute for Medical Microbiology and Hygiene, University of Freiburg, Freiburg, Germany

OBJECTIVE: To investigate the incremental cost ofnosocomial pneumonia (NP) from the perspective of a1,800 bed University hospital and health insurance fundsby means of matched pair analysis of patients with andwithout NP (CDC criteria).METHODS: Analyses were based on patient data gath-ered between 2/97 and 3/99, a) prospectively in 5 ICUsand b) retrospectively in 2 ICUs and the general wards towhich patients were transferred.RESULTS: Prospectively, out of 114 patients, 29 pairscould be matched. The incremental length of stay (LOS)per NP-patient was 6.6 days (145%; 95% CI: 2.75–10.43) resulting in total incremental cost of DM 14,606per patient (95% CI: 5,285–23,927). Antibiotic cost wasDM500 higher (1150%, 95% CI: 105–896), and costfor microbiology DM486 higher (180%) for NP-patientsthan for controls. From the health insurance fund per-spective cost was DM7,988 (130%; 95% CI: 5,281–10,694) higher. Retrospectively, out of 89 patients 37pairs could be matched. LOS in ICU was 14 days longerfor NP-patients (180%; 95% CI: 11.2–16.9), the entireLOS 10.1 days (1190%; 95% CI: 6.6–13.7) respectivelywhich resulted in incremental cost of DM29,610 per pa-tient from the hospital’s perspective (1270%; 95% CI:23,054–36,174) and DM18,000 (1245%; 95% CI:14,885–21,020) from health insurance fund perspective.Incremental cost for antibiotics amounted to DM1,042(1870%; 95% CI: 718–1,367), cost for microbiology toDM825 (1384%).CONCLUSION: The study not only gives insight into thestructure of incremental cost caused by NP but alsoshows that based on a conservative cost calculation theincremental cost per NP-patient is higher for the hospitalthan for health insurance funds which indicates a signifi-cant financial deficit for the hospital. Antibiotics and mi-crobiology together only contribute 6.8% to incrementalcost. Therefore in a cost saving initiative their close rela-tionship to LOS must be considered.

IR4

BUDESONIDE GAVE LOWER HEALTH CARE COSTS IN ASTHMA TREATMENT THAN FLUTICASONE PROPIONATEEricsson Å1, Leuenberger P2, Perruchoud AP3, Cheung D4, Grassi C5

1AstraZeneca R&D, Lund, Sweden; 2CHUV, Lausanne, Switzerland; 3Kantonspital, Basel, Switzerland; 4Schieland Hospital, Schiedam, The Netherlands; 5Università degli Studi di Pavia, Pavia, Italy

OBJECTIVES: To measure cost-effectiveness of the twomost commonly used inhaled corticosteroids in asthma: Pul-

358 Abstracts

micort Turbuhaler® (budesonide) and Flixotide Diskhaler®(fluticasone propionate).METHODS: A prospective health economics analysis wasconducted in an 8-week, open, randomised, parallel-grouptrial comparing three treatment strategies: budesonide dry-powder inhaler 400 mg/day (BUD400), fluticasone propi-onate dry-powder inhaler 400 mg/day (FP400) and budes-onide dry-powder inhaler 1600 mg/day (BUD1600). 557patients from Switzerland (CH), the Netherlands (NL) andItaly (I) participated. Health care costs and effectivenesswere analysed in a cost-minimisation analysis with a third-party payer perspective. Effectiveness was measured assymptom-free days (SFD). Pooled resource consumptionwas analysed with the three countries’ prices in one analysiseach. The health economics analysis included BUD400 andFP400, while the BUD1600 alternative was viewed as a ref-erence arm in the clinical study.RESULTS: The percentage of SFD was 54.1 for BUD400and 58.9 for FP400 (NS). The 8-week health care costs perpatient were 60 (CH, p , 0.01), 38 (NL, p , 0.01), and 49(I, p , 0.01) percent lower for BUD400 than for FP400. Re-calculated to US Dollars, this is equivalent to a lower 8-week health care cost of USD 142.53 (CH), USD 30.31(NL), and USD 45.11 (I), respectively for the BUD400 treat-ment.CONCLUSIONS: Budesonide was a more cost-effectivetreatment alternative than fluticasone, since the cost wassignificantly lower for the budesonide treatment in all 3countries studied, with no significant difference in effect.

IR5

DECISION ANALYSIS OF ACUTE TONSILLOPHARYNGITIS MANAGEMENT IN RUSSIAN CHILDRENStratchounski L, Rozenson O, Bogdanovitch T, Sudilovskaya NSmolensk State Medical Academy, Smolensk, Russia

OBJECTIVES: Acute tonsillopharyngitis (AT) is causedby Group A beta-hemolytic streptococci (GAS) in fewerthan 30% of children. There is a risk-benefit tradeoffwhen considering whether to test and/or treat. On theone hand cost of testing (throat culture and express-tech-nique) is comparatively higher than cost of antibacterialtreatment; on the other hand overtreatment promotesdrug resistance and may produce unnecessary adverse re-actions.METHODS: Three strategies were considered: 1) Treat allpatients with penicillin V, 10 days (RxALL); 2) Use ex-press-test to detect GAS and treat patients with positive re-sults (ET1/Rx); 3) Culture to detect GAS and treat pa-tients with positive results (C1/Rx). Baseline estimatedvalues for variables included the probability of positiveculture in children with clinical symptoms indicating theGAS etiology (0.4—our own data), sensitivity of express-test (0.65)—Directigen, Becton Dickinson, USA (availablein Russia) and sensitivity of culture (0.9). Finally for eachstrategy a cost/effectiveness ratio (C/E) was calculated,where C was costs of tests, cultures and Rx from the per-

spective of a second–party payer; E was Correct TreatmentDecision (culture positive patients who were treated andculture negative patients who were not treated). Decisionanalysis was done with DPL software (ADA Inc). Sensitiv-ity analyses were applied to analyze the impact of varyingthe probability estimates and cost variables.RESULTS: C1/Rx was the most cost-effective strategy(C/E 5 229), followed by RxALL (C/E 5 248) and ET1/Rx (C/E 5 786). Sensitivity analysis comparing C1/Rxto RxALL showed that C1/Rx was sensitive to changesin several variables. Specifically, RxALL was preferred if:The cost of Rx decreased (<0.9); cost of culture in-creased (>2.02); sensitivity of culture decreased (<0.59).CONCLUSIONS: C1/Rx is the preferred strategy in thedecision-making process for antibiotic treatment of acutetonsillopharyngitis in Russian children.

IR6

DEVELOPMENT AND PRELIMINARY RESULTS OF THE HEALTH AND ECONOMIC CONSEQUENCES OF SMOKING MODELOrme M1, Paine A1, Hogue S2

1The Lewin Group, Bracknell, UK; 2Global Health Outcomes, Glaxo Wellcome, Research Triangle Park, NC, USA

OBJECTIVES: To determine the health and economicburden on health care systems of smoking-related illness.Additionally, to demonstrate the disease cases and costsaverted with successful smoking cessation programmes.METHODS: A number of different modelling techniqueswere considered. Discrete difference equations were cho-sen as the most appropriate for modelling smoking statusand health changes, as well as key smoking-related dis-ease costs and the effects of smoking cessation products,within six European countries. A user-friendly interfacewas constructed using Visual Basic with an Access data-base. The study was performed as part of the WHO Eu-ropean Partnership Project to Reduce Tobacco Depen-dence and the countries thus included the four of theWHO Partnership Project (the UK, France, Germany, Po-land) and additionally the Netherlands and Italy. Withineach of these countries, six smoking-related diseases wereconsidered, COPD, asthma exacerbations, stroke, heartdisease, lung cancer and low birth weight. From currentsmoking prevalence, disease prevalence and incidence,and local treatment patterns and costs of these smoking-related diseases, the model predicts disease outcomes(e.g., number of cases) and costs in the short term (2-, 5-years) and in the longer term (10-, 20-years). Assuming aone-off investment in smoking cessation products, themodel also considers the improved outcomes that mightbe seen from such a cessation program.RESULTS: The model demonstrates that over the next20 years in these European countries, between 33% and54% of current smokers will contract a smoking-relateddisease, resulting in 14 million preventable deaths. If theprevalence of smoking remains constant, more than US

Abstracts 359

$300 billion will be spent in direct medical costs in thesesix countries to treat smoking-related diseases over 20years.CONCLUSION: The model predictions are consistentwith previously published sources and the results demon-strate the importance of smoking cessation efforts to pre-vent deaths from smoking within Europe.

IR7

A SYSTEMATIC REVIEW OF SYSTEMIC AND TOPICAL ANTIMICROBIAL AGENTS USED IN THE PREVENTION AND TREATMENT OF CHRONIC WOUNDSO’Meara S, Cullum C, Majid M, Sheldon TANHS Centre for Reviews and Dissemination and Centre for Evidence Based Nursing, University of York, York, UK

OBJECTIVE: To determine the effectiveness and cost-effectiveness of antimicrobial agents in the managementof chronic wounds.METHODS: A systematic review of the literature wasundertaken. Eighteen electronic databases, relevant jour-nals, conference proceedings and bibliographies of re-trieved papers were searched, and an expert panel wasconsulted. Randomised and non-randomized trials, withpatients, limbs, or lesions as the units of allocation wereselected. People with diabetic foot ulcers, pressure ulcers,leg ulcers, pilonidal sinus, non-healing surgical wounds,chronic cavity wounds, and those at risk of developingpressure ulcers were included. Any systemic or topicalagents with antimicrobial properties were considered.The primary outcome was wound healing (incidence ofnew lesions for prevention studies) assessed with an ob-jective measurement. All included studies were assessedagainst a comprehensive checklist for methodological qual-ity. A narrative (qualitative) overview was conducted, withresults grouped according to wound type. It was not pos-sible to combine results by meta-analysis due to lack ofsimilarity across the studies.RESULTS: Twenty-seven trials were included, 23 of ran-domized design. There were 9 evaluations of systemic an-timicrobials and 18 of topical agents. Several method-ological problems were detected, the most common beinginadequate sample size. Results do not support the rou-tine use of systemic antibiotics for leg ulcers (various eti-ologies) or diabetic foot ulcers without acute infection,however they may be useful as an adjunct to surgery forpilonidal sinus. Several topical preparations may be help-ful, including dimethyl sulfoxide, silver sulphadiazine,benzoyl peroxide, oxyquinoline, and gentamicin. Hydro-gels, hydrocolloids, and silastic foam dressings appear toproduce equivalence in terms of clinical effectiveness ver-sus antimicrobial agents, and may be associated with costsavings.CONCLUSIONS: Most of this research requires replica-tion in larger, well-designed studies to establish both clin-ical and cost-effectiveness.

IR8

COST-EFFECTIVENESS OF COMBINED HEPATITIS A AND B VACCINATION AND HEPATITIS B MONOVACCINATION IN SWITZERLAND AND AUSTRIASchneller P1, Banz K1, Staginnus U2

1OUTCOMES International, Basel, Switzerland; 2SmithKline Beecham Biologicals, Rixensart, Belgium

OBJECTIVE: To analyze the cost-effectiveness of com-bined hepatitis A and B vaccination (Vacc A1B) and hepa-titis B monovaccination (Vacc B) in children and adoles-cents.METHODS: A decision tree combined with a Markov-model was used to predict clinical and economic outcomesfor three different age groups: 0–15, 1–15, and 11–15years. The model was designed to follow an initially vacci-nated age group over 30 years. The analysis took the per-spective of society, i.e., direct and indirect costs were con-sidered. Costs were discounted at 5% per year. Primaryoutcomes measure was the incremental cost per avoided in-fection of each vaccination strategy compared to no vacci-nation.RESULTS: The analysis indicated that for both Switzer-land and Austria, Vacc A1B is more cost-effective thanVacc B in all three compared age groups. The incrementalcost-effectiveness ratios of the Vacc A1B strategy for Swit-zerland ranged from CHF 11,070 to 12,210 (Austria ATS61,260 to 70,270) versus CHF 24,630 to 26,520 (AustriaATS 141,750 to 159,140) for Vacc B depending on theage group. For both vaccination strategies, the most favor-able cost-effectiveness ratio was obtained when vaccinating11–15 year olds. The superior cost-effectiveness of the VaccA1B strategy is due to a reduction in hepatitis A coupledwith a modest increase in vaccination costs. In view of evi-dence suggesting that hepatitis A and B cases are more fre-quent than officially reported, these results are likely to beconservative.CONCLUSIONS: Prevention of both hepatitis A and B ascompared to hepatitis B alone is cost-effective in Switzer-land and Austria. When taking into consideration the lowincidence of hepatitis A in both countries, combined hepati-tis A and B vaccination is likely to have even greater impactin countries with a higher incidence of hepatitis A infection.

ECONOMIC & OUTCOMES STUDY RESULTSOF CANCER

CA1

PROSPECTIVE ECONOMIC EVALUATION OF ANTIBIOTIC PROPHYLAXIS IN SMALL CELL LUNG CANCER (SCLC) PATIENTS RECEIVING CHEMOTHERAPYCaleo S1, Tjan-Heijnen VGC2, Burghouts J3, Giaccone G4

1EORTC Data Center, Brussels, Belgium; 2University Hospital, Nijmegen, The Netherlands; 3Groot Ziekenhasthuis, s’Hertogenbosch, The Netherlands; 4Academisch Ziekenhuis Der Vrije Universiteit, Amsterdam, The Netherlands

360 Abstracts

OBJECTIVE: To determine whether the incidence of fe-brile leucopenia (FL) is reduced in SCLC patients by ad-ministering prophylactic antibiotics thereby reducing re-source use and costs associated with treatment.METHODS: A cost-effectiveness analysis was conductedalongside a multicenter phase III randomized controlledtrial with a 2 3 2 factorial design. The trial was designed tocompare standard dose CDE (cyclophosphamide, dox-orubucin and etoposide) versus intensified CDE (125%dose with G-CSF), with or without prophylactic antibiot-ics. Patients in the verum arm received roxithromycin150 mg and ciprofloxacin 750 mg, both given twice dailyfor 10 days, commencing the day after completing eachcycle of chemotherapy. The economic evaluation examinesthe costs and effects of patients taking antibiotics versusplacebo. Results are presented for 33 patients in The Neth-erlands, representing 20.5% of the total sample (n 5 161).Resource utilization data were collected during the trialand included hospitalizations, concomitant medications,transfusions and medical procedures. The viewpoint of thehealth insurance system is taken, using 1998 prices ex-pressed in Dutch guilders (ppp of 2.07NLG/US$). The ef-fectiveness measures include the incidence of febrile leu-copenia and the number of infectious deaths.RESULTS: The average cost attributed to patients in theplacebo arm was 12,898 NLG compared to 6,999 NLG inthe verum arm, yielding a difference of 5,899 NLG (US$2850) (95% CI (2) 2081 2 13,897) in favor of administer-ing prophylactic antibiotics. The clinical trial (n 5 161)demonstrated a 50% decrease in the incidence of FL and anabsolute reduction in the risk of infectious death from 6%to 0% in patients taking antibiotics.CONCLUSION: The use of prophylactic antibiotics insmall cell lung cancer patients is cost-effective as it reducesthe incidence of FL and infectious deaths whilst reducingdirect medical costs by 45%. This study was partly spon-sored by Bayer BV.

CA2

COST COMPARISONS BETWEEN HOSPITALSIN DIFFERENT COUNTRIES BY MEANS OF A RESOURCE USE INDEXNeymark N, Adriaenssen IEORTC Health Economics Unit, Brussels, Belgium

OBJECTIVES: To compare the direct costs of managingpatients with advanced colorectal cancer in five Europeancountries by means of a resource use index, which avoidsthe confounding caused by differences in absolute and rela-tive prices.METHODS: In 10 hospitals in five countries, data on pa-tient management and resource use were collected from thehospital charts of 20 consecutive, newly diagnosed pa-tients. All available data on resource use for each patientfrom diagnosis until death or last seen alive were includedin the investigation. For only one country, Belgium, was itpossible to find a complete set of consistent unit prices,namely the tariffs according to which the providers are re-

munerated by the health insurance system. By using theseunit prices to value the use of resources in all countries, anindex for average total resource use per hospital has beenconstructed, which reflects only differences in real resourceutilization. The average across all the hospitals has been setequal to 100, so index values represent deviations from thisaverage.RESULTS: The estimated index values for average total re-source use vary from 75 to 131. Differences between thetwo hospitals in each country are just as big as those be-tween hospitals in different countries. When subdividingthe patients into four types according to site and stage ofdisease, the spread between the hospitals becomes evenwider. No hospital is consistently above or below the aver-age resource use for all patient categories.CONCLUSIONS: Superficial comparisons of the costs oftreating specific diseases in different countries will unavoid-ably be confounded by differences in prices. By using a sin-gle set of unit prices for weighting the use of resources, anindex reflecting only real differences in resource use may bedetermined and used for initial, aggregate comparisons ex-ploring the need for further analysis.

CA3

ESTIMATING SURVIVAL GAIN FOR COST-EFFECTIVENESS ANALYSIS: THE EXAMPLE OF EARLY HORMONAL TREATMENT IN LOCALLY ADVANCED PROSTATE CANCERNeymark N, Adriaenssen I, Caleo SEORTC Health Economics Unit, Brussels, Belgium

OBJECTIVES: To examine the impact of various meth-ods of estimating survival gain from censored survivaldata on the results of a cost-effectiveness analysis.METHODS: A cost-effectiveness analysis of the additionto radiotherapy of early hormonal therapy with a LHRHanalogue (goserelin) in patients with locally advancedprostate cancer has been carried out. The clinical resultsderive from a randomized controlled trial (EORTC22863, reported by Bolla et al. [1]), and resource utiliza-tion data were collected retrospectively in one Frenchhospital recruiting 90/415 patients in the trial. Costs arecalculated from the viewpoint of the French health insur-ance system. To estimate average survival, various meth-ods are available and will be compared, notably a re-stricted means analysis truncated at a median follow-upof 45 months and extrapolations based on fitting a para-metric model (e.g., Weibull or exponential) to the sur-vival curves.RESULTS: The clinical trial documented an overall sur-vival rate at 5 years of 79% in the group receiving radio-therapy plus hormonal therapy compared to 61% (p 50.001) in those receiving only radiotherapy. The averagecosts in the two groups were 81,700 FF and 69,900 FFrespectively. The survival improvement is thus associatedwith an increase of 11,800 FF in average cost, but to de-termine the incremental cost-effectiveness ratio (ICER)the exact gain in average survival must be known. The re-

Abstracts 361

sults of using the various estimation methods will be pre-sented and discussed at the meeting.CONCLUSIONS: Survival data from clinical trials arenormally not analyzed with the purpose of determiningaverage survival, but such estimates are necessary foreconomic evaluation. Due to censoring, mean survivalcan not be estimated without extrapolation or withouttruncation in a restricted means analysis. The results interms of ICER will vary according to the estimationmethod used. These variations should be examined insensitivity analyses

CA4

METASTATIC BREAST CANCER PATIENTS TREATED WITH DOCETAXELBurrell A1, Hall A2, Eggleton SPH1

1Rhône-Poulenc Rorer, Kent, UK; 2Centre for Medical and Dental Education, St Bartholomew’s Hospital, London, UK

OBJECTIVE: QuOLA-T was established as a real-life UKoutcomes audit to measure changes in Quality of Life (us-ing the EORTC QLQ-C30 questionnaire) in patients withmetastatic breast cancer treated with docetaxel (Taxo-tere®) outside the parameters of formal clinical trials.METHODS: QuOLA-T is an ongoing programme inwhich Quality of Life (QoL), clinical parameters of perfor-mance status (Karnofsky Index) and response data are col-lected at baseline and after every two cycles of treatmentand following chemotherapy. Demographic parametersare also collected at baseline. Evaluable patients were de-fined as those for whom a baseline registration and QoLform have been received, details of chemotherapy are com-pleted and at least one QoL form from the treatment phasehas been received. Non-evaluable patients were those stillreceiving treatment and those not meeting the criteriaabove. There were no significant differences between eval-uable and non-evaluable patients in any of the parametersat baseline, indicating that evaluable patients are a fairrepresentation of the entire data set.RESULTS: The median number of cycles of Taxotere ad-ministered: 5 (range 1–9). 103 patients completed plannedchemotherapy (49%), with a median of 6 cycles. Majorreason for withdrawal from treatment was progressive dis-ease (25%), median number of cycles: 4. There were sig-nificant detrimental intra-patient changes in two of theQLQ-C30 domains (Fatigue and Diarrhea) but there weresignificant beneficial intra-patient changes in four of thedomains (Emotional Functioning, Pain, Constipation andNausea and Vomiting).CONCLUSIONS: This is the first large prospective Qual-ity of Life monitoring programme conducted in an unse-lected population of patients with metastatic breast cancerusing a recognized instrument. The hypothesis that qualityof life is significantly decreased during the active phase ofchemotherapy with Taxotere is not supported. This studyprovides a much clearer picture of the real-life effects ofTaxotere on Quality of Life.

CA5

QUALITY OF LIFE IN A BREASTCANCER POPULATION UNDERGOING ADJUVANT TREATMENTNichol MB1, Venturini F1, Jones J1, Sherman G2

1University of Southern California, Los Angeles, CA, USA; 2Cancer Care Associates, Torrance, CA, USA

OBJECTIVE: To investigate the effect of adjuvant breastcancer therapy on quality of life, and to evaluate the oc-currence of dose delay or reduction by type of therapy.MATERIALS AND METHODS: The study populationwas drawn from patients treated by a major oncology cen-ter in Southern California. The inclusion criteria were: fe-males 18 years old or older treated with adjuvant therapy,with stage I, II or III non-metastatic breast cancer. Partici-pants were sent up to five surveys: one before treatment ini-tiation, two or three (depending on the number of cycles) atregular intervals between treatments, and a final one 4weeks after the end of the chemotherapy session. The surveycontained two quality of life instruments: the SF-36 and theBreast Cancer Chemotherapy Questionnaire (BCQ).RESULTS: At the beginning of the chemotherapy course,40 patients were prescribed CMF (48.2%), 24 AC(28.9%), 11 A-CMF sequential (13.3%), and 8 other regi-mens (9.6%). Average values for the physical componentsummary score (PCS) of the SF-36 did not vary over time,ranging from 42.1 to 46.1. The mental component sum-mary score (MCS) had a peak at mid-point (48.1), while itstabilized to the initial value at the end of the therapy(46.8 vs. 46.5). Comparisons of the difference between thebaseline quality of life scores and a summary measure ofthe subsequent scores among treatment groups were notsignificant. Correlation coefficients showed that, at eachperiod, MCS score was highly correlated with the disease-specific BCQ score (correlation coefficients from 0.69 to0.81, p 5 0.001). During the chemotherapy course, 32 pa-tients experienced a dose delay, 12 a dose reduction and 9prematurely terminated their chemotherapy regimen, withno difference among treatment groups.DISCUSSION: The MCS of the SF-36 was highly correlatedwith the disease specific instrument, while the PCS was not.

QUALITY OF LIFE STUDY RESULTS

QL1

THE QUALITY OF LIFE OF ESSENTIAL HYPERTENSIVES IN AN ECONOMICALLY DISADVANTAGED COMMUNITY:PREDICTORS AND RELEVANCE TO MANAGEMENT COMPLIANCEGolubev SA, Miliy MN, Shumann SGVitebsk State Medical University, Vitebsk, Belarus

OBJECTIVES: To evaluate the quality of life (QL) in es-sential hypertensives (EH) of Belarus, a European countrywith a unique combination of the important rise in cardio-vascular morbidity and mortality, highly restricted financ-

362 Abstracts

ing of the state health care system, and the society experi-encing socioeconomic hardships.METHODS: In 222 verified mild to moderate EH (meanage 49.4 6 0.9 years) as well as in 77 age-, gender-, bodymass- and social status-matched healthies (H) the QL as-sessment was performed using the well-validated, self-administered questionnaires: The Giessen Somatic Com-plaints Questionnaire (GSC) and The General Well-BeingAdjustment Scale (GWB).RESULTS: The QL tests revealed significantly lower QLscores on all GSC and GWB subscales in the investigatedBelarusian EH compared with H (Gastric Complaints,Pains, Cardiac Complaints, Exhaustion, Anxiety, Depres-sion, Well-Being, Self-Control, General Health, Vitality)with the total Complaints Index (mean 6 SEM) 52.4 6 2.6vs. 22.5 6 1.9 points (p , 0.001), and the total GWB in-dex 78.6 6 1.3 vs. 94.5 6 2.1 points (p , 0.01), respec-tively. QL scores did not correlate (Spearman) with officeblood pressure and age. Females demonstrated less favor-able scores on all subscales compared with males in bothgroups (Mann-Whitney, p , 0.01). Educational level, oc-cupation, physical activity had no impact on the QL(ANOVA). There was an association of the total QL withalcohol and coffee consumption as well as smoking (stan-dardized interview), while abstinents had the lowest QLscores (p , 0.05). The QL total indexes were significantly(p , 0.05) worse in the EH treated on regular basis com-pared with EH who did not treat their known hyperten-sion. The high QL score, smoking, drinking and male gen-der were revealed as factors of poor compliance withprevious routine medical care.CONCLUSIONS: Gender and certain life-style factors in-fluence the QL profile of EH living in Belarus and relate tocompliance, this requiring careful consideration during hy-pertension management.

QL2

PSYCHOMETRIC EQUIVALENCE EVALUATION OF FIVE NON-WESTERN-LANGUAGE TRANSLATED VERSIONS OF THE FUNCTIONAL ASSESSMENT OF CANCER THERAPY—GENERAL (FACT-G) VERSION 4Eremenco S, Chang C-H, Lent L, Cella DCenter on Outcomes, Research, and Education (CORE), Evanston Northwestern Healthcare, Evanston, IL, USA

OBJECTIVES: A rigorous translation methodology is es-sential in order to develop equivalent language versions ofhealth-related quality of life questionnaires for use in clini-cal research and trials. Using differential item functioning(DIF) within the item response theory framework to aug-ment traditional reliability analysis provides additional evi-dence to evaluate whether the translated versions are psy-chometrically equivalent.METHODS: The Functional Assessment of Chronic IllnessTherapy (FACIT) Measurement System translation meth-odology consists of 2 forward translations, 1 reconciledversion, a back-translation of the reconciled version, and 3

independent reviews by bilingual experts. The translatedversions are then pretested with at least 15 patients in therespective native countries. In this presentation, we focuson five translated versions of the FACT-G (Hungarian, Jap-anese, Polish, Portuguese, and Russian) that were eachtested with at least 40 patients and analyzed; the total sam-ple was 316 oncology patients. The internal consistency ofthe 27-item core FACT-G was evaluated using Cronbach’salpha. Rasch analysis was used to determine the item diffi-culty hierarchy of each language version, and then DIFanalysis was used to determine psychometric equivalence inrelation to the English version (n 5 131).RESULTS: The 27 core FACT-G items included in everyFACIT measure showed high reliability in these five lan-guage versions, with Cronbach’s alpha ranging from .87(Japanese) to .93 (Polish). Items exhibiting potential DIFand their possible causes were summarized.CONCLUSION: Reliability and DIF analyses confirm thatthe five translated versions are psychometrically valid andcomparable. The FACIT translation methodology has beeneffective in adapting a North American questionnaire tovastly different cultural contexts and creating valid versionsfor cross-cultural comparisons. Further research and devel-opment of new language versions is needed, especially innon-Western European languages to extend QoL researchpossibilities internationally.

QL3

ANTIPSYCHOTIC TREATMENT ANDCHANGES IN HEALTH RELATED QUALITYOF LIFE IN PATIENTS WITH SCHIZOPHRENIA USING EQ-5DBadia X1, Casado A1, Sacristán JA2, Gómez JC2, Gregor KJ31Public Health Institute of Catalonia, Barcelona, Spain; 2Eli Lilly and Company, Madrid, Spain; 3Eli Lilly and Company, Windlesham Surrey, UK

OBJECTIVES: The use of antipsychotics in the treatmentof schizophrenia has changed patient connections to theirsurroundings, placing them in greater contact with theirfamily and the environment in which they carry out dailyactivities. The purpose of this analysis of patients in thenaturalistic setting was to evaluate the changes in healthrelated quality of life (HRQL) associated with antipsy-chotic drug therapy and to evaluate the sensitivity of theEuroqol-5D (EQ-5D) as a tool for measuring HRQL inpatients with schizophrenia.METHODS: The study included 2,128 patients withschizophrenia who were treated with olanzapine, 821 whowere treated with risperidone or haloperidol, and smallnumbers of patients on other antipsychotics. Sociodemo-graphic, clinical (global clinical impression, global activityevaluation scale, presence of extrapyramidal symptoms)and the EQ-5D as a generic tool for the measurement ofthe HRQL were included as variables. The patients wereevaluated at treatment initiation and after 6 months.RESULTS: Olanzapine-treated patients showed the larg-est effect size on the Visual Analog Scale (VAS) (0.98)

Abstracts 363

and the EQ-5D Index (EQ-I) (1.13) compared with theremaining antipsychotic drugs (0.58-0.75 for the VASand 0.78-0.96 for the EQ-I). The olanzapine-treatedgroup had the highest proportion of patients with an im-provement, particularly in the daily activities and mooddomains, and the lowest proportion of patients whoworsened after 6 months of treatment.CONCLUSION: Treatment with antipsychotic drugs, es-pecially olanzapine, improves the HRQL of patients withschizophrenia. The study also indicates that the EQ-5D isuseful and sensitive tool for detecting changes in thehealth of patients with schizophrenia.

QL4

USING RASCH MODELS TO IMPROVE THE EVALUATIVE-SCALE PROPERTIES OF THE WOMEN’S HEALTH QUESTIONNAIRETurner R1, Hayes J1, Testa M2

1Phase V Technologies, Inc., Wellesley Hills, MA, USA; 2Harvard School of Public Health, Boston, MA, USA

OBJECTIVES: To improve the evaluative-scale character-istics of the WHQ. Comparative clinical trials require eval-uative scales that are sensitive and responsive as well asvalid and reliable. The Women’s Health Questionnaire(WHQ), a 36-item, 4-point Likert self-administered ratingscale, assesses subjective reports of emotional and physicalwell-being in mid-aged women using nine physical, emo-tional, and symptom-specific scales. We assessed theWHQ evaluative scale properties using both true score andRasch measurement models to determine whether thescoring techniques proposed by the scale developer opti-mized the evaluative properties.METHODS: We analyzed WHQ baseline and six-weekchange score data from 651 mid-aged women treatedwith HRT in two US clinical trials. The Multitrait/Multi-item Analysis Program—Revised provided true score psy-chometric information. The BIGSTEPS program pro-duced the Rasch analysis.RESULTS: The original nine-factor structure was not sup-ported by confirmatory factor analysis. Rasch analysisidentified scale-level category response problems for sev-eral scales in the revised six-factor structure. The problem-atic scales failed to detect differences between “knowngroups” formed by clinical criteria for menopausal symp-toms, and were not responsive to change within groupsthat changed significantly on clinical criteria. Revised scalescoring based on Rasch conversion to standardized logitsresulted in improved sensitivity and responsiveness.CONCLUSIONS: Rasch models provide important psy-chometric information for scale development and revi-sion. They resolve interval data assumptions, are moresensitive to information contained in the tails of the re-sponse distributions, and yield useful response categoryanalyses that guide scaling decisions independent of thesample characteristics. Based on the Rasch analysis, wemade scoring and interpretation changes to the WHQthat increased its sensitivity and responsiveness.

QL5

ASTHMA BOTHER PROFILE VERSUS AIRWAY QUESTIONNAIRE 20 AS SIMPLE MEASURES TO EVALUATE THE HEALTH STATUS IN ASTHMANishimura K, Hajiro T, Tsukino M, Oga TDepartment of Respiratory Medicine, Graduate School of Medicine, Kyoto University, Kyoto, Japan

OBJECTIVES: Measurements of the patient health statusshould be simple and concise in clinical practice. TheAsthma Bother Profile (ABP), which consists of 23 items,has been developed to assess how much asthma bothers pa-tients. The Airway Questionnaire 20 (AQ20) is a simple in-strument which consists of 20 items regarding the healthstatus of asthma and COPD patients. The purpose of thisstudy was to investigate how the ABP and AQ20 evaluatethe health status of patients with asthma, and to examinethe relationships between these two measures and clinicalparameters, generic and asthma-specific instruments.METHODS: A total of 193 patients with chronic asthma(age: 48 6 8 yr, 95 males) completed a pulmonary functiontesting, a measurement of airway hyperresponsiveness, thedyspnea rating (MRC; Medical Research Council dyspneascale), and assessments of the anxiety and depression(HADS; Hospital Anxiety and Depression Scale), and as-sessments of their health status. The health status was as-sessed using the ABP, AQ20, SF-36, Living with AsthmaQuestionnaire (LWAQ) and the Asthma Quality of LifeQuestionnaire (AQLQ). The Japanese version of the ABPincluded only 15 items out of the original 23 items due tocultural differences.RESULTS: The score distributions on the ABP and AQ20were skewed towards the mild end of the scale (medianscore [interquartile range]: 2 [1–4] and 13 [6–22], respec-tively). Moderate correlations were noted between theAnxiety on the HADS and the ABP and AQ20 scores(Spearman’s rank correlation coefficient [Rs] 5 0.44 and0.62 respectively). The ABP had a strong correlation withthe LWAQ (Rs 5 0.81), and its strongest correlationwith the General Health (Rs 5 20.64) among the 8 sub-scales on the SF-36. The AQ20 had a moderate correlationwith the LWAQ (Rs 5 0.61). The FEV1, peak expiratoryflow rates, degree of airway hyperresponsiveness and treat-ment steps of asthma showed mild correlations with theABP and AQ20 (Rs 5 0.20 2 0.26).CONCLUSIONS: The ABP and AQ20 were short andsimple to complete, and both measures could easily be usedin clinical practice. The ABP can evaluate patients morespecifically with respect to the distress and bother inducedby asthma than the AQ20.

STUDY RESULTS MEASURING SYMPTOMATIC RELIEF

SR1

THE ECONOMIC IMPACT OF TOLTERODINEIN PORTUGALDias J1, Carvalho H2

1Pharmacia & Upjohn Laboratórios Lda, Lisbon, Portugal;

364 Abstracts

2Portuguese Association of Neurourology and Urogynecology, Lisbon, Portugal

OBJECTIVE: To estimate prevalence of Urinary Inconti-nence (UI) in the Portuguese population. To determine thecosts of treatment (including incontinent pad usage) withTolterodine for patients suffering from chronic UI.METHODS: Data from the National Health Survey per-formed in 1995–96, involving a representative sample of49718 people across mainland Portugal, was analyzed todetermine age, sex-specific and age-standardized preva-lence rates of UI. The average consumption of pads by in-continent patients was estimated from the information pro-vided by the Centro Regional de Saúde from MadeiraIsland, a region where pads are reimbursed.RESULTS: The global prevalence of urinary incontinencein Portugal was of 4,3%, being 2,6% in males and 5,8% infemales. Age and sex adjusted prevalence rates were of1,9% in males while 4,4% in females. Age and sex specificprevalence rates increase with age from 15 years onwardsin both sexes, being higher in females across all age groups.Above 65 years of age 10% of men and 12% of womensuffer from urinary incontinence. Of these, 68,8% arelikely to suffer from chronic urinary incontinence. This isthe estimated burden of the disease in the general Portu-guese population. Urinary incontinent chronic patients useon average 4 pads per 24 hours, normally using 3 pads dur-ing the day and 1 pad at night. Comparing the average padcosts per day with using Tolterodine, the difference is re-markable (648$00 PTE versus 334$60 PTE). The eco-nomic impact of such use would reduce the health expendi-ture in UI patients significantly.CONCLUSION: The prevalence of UI in Portugal iswithin the limits of what is mentioned in other community-based studies. The average daily costs of treatment withTolterodine are lower when compared with pad usage.

SR2

A PHARMACO-ECONOMIC CASE STUDY IN ANAESTHESIA, INCLUDING A RECENT RE-ANALYSIS USINGBOOTSTRAP TECHNIQUESMcNamara J, Gillis SBiostatistics Group, PAREXEL International, Sheffield, UK

OBJECTIVES: To apply bootstrap techniques to deriveconfidence intervals (CIs) for cost differences between twoanaesthetic regimes, and to compare with standard para-metric methods for deriving CIs.METHODS: In a cardiac surgery clinical trial, patients wererandomized to one of two anaesthetic regimes, either propo-fol or midazolam, for the induction and maintenance of an-esthesia, and sedation post-surgery in ICU. A cost compari-son of these two regimes was published in 1996, with alldrug usage and ICU nursing time costs. Parametric methodswere used to construct 95% CIs for the difference betweenpropofol and midazolam. Untransformed parametric analy-sis of cost data often fails because of distributional skew-

ness. Therefore, log-transformed analysis was undertaken toovercome the skewness problem. More recently, log-trans-formation has become a discredited strategy because health-care decision-makers focus on total budgets—for instance,the total annual budget available to provide surgery at aparticular centre. The only estimator directly linked to thistotal cost is the arithmetic mean—therefore any analysisbased upon geometric means, arising from log-transforma-tion, is inappropriate. Other authors have therefore recom-mended bootstrap techniques to produce CIs for cost differ-ences on an untransformed scale. This approach is notinvalidated by distributional skewness.RESULTS: With 37 propofol and 33 midazolam patients,a comparison of drug plus nursing costs showed an advan-tage per subject for propofol of £43.23 (95% CI: 2£3.21to £89.67). An untransformed two-sample t-test proved in-appropriate, because the necessary distributional assump-tions were not satisfied. Bootstrap CIs for the cost differ-ence were then constructed using four different bootstraptechniques. All bootstrap CIs showed a striking similarityto the untransformed estimates above, demonstrating therobustness of the t-test and conventional parametric meth-ods, despite distributional skewness.CONCLUSIONS: Both untransformed and bootstrap esti-mates should be presented—hopefully, the two methods willbe in agreement and provide mutually supportive evidence.

SR3

DEVELOPMENT AND VALIDATION OF THE QUALITY OF LIFE WITH MIGRAINE (QOLWM): A BRIEF MIGRAINE SURVEYCramer JA1, Silberstein S2, Winner P3, Becker L4, Chmiel J4, Ashraf T4, Noble K4

1Yale University School of Medicine, New Haven, CT, USA; 2Jefferson Headache Center, Philadelphia, PA, USA; 3Palm Beach Headache Center, West Palm Beach, FL, USA; 4Abbott Laboratories, Abbott Park, IL, USA

OBJECTIVE: To develop and validate a brief survey of mi-graine-related quality of life issues. Unlike other instru-ments that define the impact of acute treatment, the Qual-ity of Life with Migraine (QOLWM) questionnaire wasdesigned to assess the chronic impact of migraine.METHODS: Major issues related to living with chronic,frequent migraine headaches were condensed into seventopics. Each item was structured as a question about fre-quency and a question about bothersomeness. Item scoreswere the products of frequency and bothersomeness. Thetotal score was the sum of the seven item scores. Peoplewith migraine who voluntarily responded to a headachesurvey also completed the QOLWM and the HeadacheDisability Index (HDI).RESULTS: The QOLWM was completed in full by 994adults (81% women) with mean total score 77.98 6 40.49(range 7–175). There were no floor or ceiling effects. Inter-nal consistency reliability (Cronbach’s alpha) of 0.93 wasexcellent, with item-scale correlations of 0.59–0.83. Theeigenvalue of the single factor was 4.78, with item loadings

Abstracts 365

of 0.68–0.88. Scores correlated significantly with frequencyof migraines in the past month and use of prophylacticmedications daily (both p , 0.0001). Correlation of theQOLWM and HDI total scores for the 728 people whocompleted both questionnaires was 0.73. Item correlationswith the HDI total score were 0.47–0.67, demonstratingexternal criterion validity. Additional studies are ongoingto assess reproducibility and responsiveness.CONCLUSIONS: These data demonstrate the psychomet-ric properties of the QOLWM. The brief questionnaire maybe useful as a screening tool for clinicians to evaluate the im-pact of migraine on individuals. The two-dimensional ap-proach to patient-reported quality of life allows individualsto weight the impact of both frequency and bothersomenessof chronic migraines on multiple aspects of daily life.

SR4

LONGITUDINAL ASSESSMENT OF ASTHMA AND WORK OUTCOMESErickson SR, Kirking DMUniversity of Michigan, Ann Arbor, MI, USA

OBJECTIVE: Describe relationships between changes inwork outcomes (WO) and changes in asthma severityover a one-year period.METHODS: This longitudinal study surveyed adults en-rolled in a managed care organization using identicalmailed surveys. 106 respondents completed both the 1997and 1998 questionnaires and reported working outside thehome. Self-reported WO included missed work days dueto asthma and work performance (WP) from the Func-tional Status Questionnaire, a 6-item scale using a 4-pointLikert scale, transformed to a 100 point scale. A 4-weektime reference was used. Perceived severity was determinedby asking patients their perceived severity on a 5-pointscale from very mild to very severe; evaluated severity wasbased on reported symptoms matched to national guide-lines. Respondents were grouped by change in asthma se-verity: improvement; no change; or decline of at least oneseverity category. Analysis included descriptive statistics,paired Student’s t-test, and McNemer’s test.RESULTS: Initially, respondent’s mean age was 44.4(610.9) years; 72 (68.6%) were female; 76.7% had an an-nual family income of >$40,000; 96.2% were Caucasian;and 76.4% had some college education or more; 42.9%perceived their asthma severity as very mild or mild,38.1% moderate, and 19.1% as severe or very severe.Based on symptoms, 32.4% were classified as intermittentmild, 28.6% mild persistent, 36.2% moderate, and 2.9%severe asthma. WP scores changed in the expected direc-tion based on changes in asthma severity: no change inperceived severity (n 5 58) had no significant change inWP score (91.3 6 13.0 versus 92.9 6 12.4, p 5 0.52); im-provement in perceived severity (less severe, n 5 17) hadimproved WP scores (91.8 6 7.9 versus 94.4 6 9.4, p 50.30); and worsening of perceived severity (n 5 26) had adecline in WP score (94.4 6 8 versus 91.3 6 10.9, p 50.07). WP score trends were similar based on evaluated se-

verity. There were no significant differences in changes inthe number of respondents who reported missing work.CONCLUSIONS: Changes in WP are related to asthmaseverity. The measure of missed days work is less consis-tently related.

SR5

MEDICAL EVENTS AND RESOURCE UTILISATION IN CANCER PAIN PATIENTS TREATED WITH STRONG OPIOIDS: AN ANALYSIS OF THE UK-GPRD DATABASENuyts GD1, Haazen LM1, Jones MP2

1Janssen Pharmaceutica, Beerse, Belgium; 2Jones and Just Pty Ltd, Sydney, Australia

OBJECTIVES: To study the incidence of medical eventsand resource utilisation in patients treated for severe cancerpain with strong opioids with the general practitioner (GP).METHODS: We analyzed data on 2,323 cancer patientswho switched from a weak to a strong opioid. Patientsstarted either on TTS Fentanyl (TTS: N 5 270), Immedi-ate Release Morphine (IR: N 5 1,909) or Sustained Re-lease Morphine (SR: N 5 144). We compared the medi-cal events during their pain treatment, the number ofpatients receiving concomitant medication and the num-ber of GP visits. A descriptive analysis was performed aswell as relative risks (RR) calculated, adjusting for rele-vant co-variates.RESULTS: The groups were comparable for age (avg. 68yrs). There were more female patients in the TTS group(61.5%) compared to 48.1% and 49.3% for IR and SR,respectively. Median duration of cancer prior to the startof strong opioids was 7.1 months for TTS, 6 for IR and5.3 for SR. Mean duration of treatment: 68 days for TTS,97 for IR and 92 for SR. No differences in concomitantillnesses were observed except for cancer types with sig-nificantly more GI and fewer lung cancers in the TTSgroup. Compared to TTS, patients on IR had more con-stipation (RR 1.49: 95% CI 1.14–1.94), more nauseaand vomiting (RR 1.43: 95% CI 1.09–1.88) and morecardiac events (RR 1.95: 95% CI 1.15–3.29), while SRpatients differed from TTS only with respect to cardiacevents (RR 2.79: 95% CI 1.49–5.22). IR patients hadalso a higher rate of hospitalizations (RR 1.95: 95% CI1.14–3.31) and GP visits (RR 1.21: 95% CI 0.98–1.49).Fewer TTS patients consumed additional pain medica-tion, laxatives as well as antibiotics and CNS medication.CONCLUSION: This analysis of GP derived observa-tional data indicate that TTS Fentanyl results in a lowerconsumption of medication and other health care re-sources compared to morphine treated cancer patients.

ECONOMIC & OUTCOMES STUDY RESULTS OF GASTROINTESTINAL DISORDERS

GI1

COST-EFFECTIVENESS OF THE COMBINATION OF MISOPROSTOL WITH DICLOFENAC IN THE TAYSIDE POPULATION

366 Abstracts

Morant S, Davey P, MacDonald TMedicines Monitoring Unit (MEMO), Ninewells Hospital, Dundee, Scotland

OBJECTIVES: To compare observational data with alarge clinical study (MUCOSA) which showed that miso-prostol reduced NSAID complication rates by 40% in pa-tients with arthritis. To measure the cost of prescribing,endoscopy and hospitalizations for patients receiving non-steroidal anti-inflammatory drugs (NSAIDs) and estimatethe cost-effectiveness of misoprostol in routine practice.METHODS: A cohort study using all patients in the Tay-side region that received NSAIDs and anti-ulcer drugs be-tween 1989 and 1995. Thirty-day treatment was esti-mated from the cost of the NSAID plus endoscopies plushospitalization for GI events. Costs of hospitalizationsand endoscopic procedures were obtained from ScottishInformation and Statistics Division for 1997.RESULTS: Among 54,807 eligible patients the risk ad-justed rates of hospitalization for gastrointestinal diag-noses were 50% lower on Arthrotec (a fixed combinationof misoprostol and diclofenac) than on diclofenac alone.Statistically significant risk factors were: a prior historyof gastrointestinal events (p , 0.001), a prior history ofcardiovascular events (p , 0.001), increasing age (p ,0.001), social deprivation score (p 5 0.072), concurrentexposure to anti-ulcer drugs (p , 0.001) or steroids (p 50.001) and type and dose of NSAID (p , 0.001 and p 50.047 respectively). Only nabumatone had a lower eventrate than Arthrotec, but at a higher expected cost. Ar-throtec had lower 30 day treatment and complicationcosts than diclofenac alone in high risk patients (e.g., £52vs £86 for a patient aged 80–89 with a prior history ofGI events) but not in low risk patients (e.g., £16.50 vs£15 for a patient aged 50–59 with no prior GI events).CONCLUSIONS: There was close agreement betweenthis observational study and the MUCOSA study on theextent to which the prophylactic use of misoprostol re-duced NSAID associated gastrointestinal complications.The combination of misoprostol with diclofenac shouldreduce thirty day treatment and complication costs inhigh risk patients, in comparison with diclofenac alone.

GI2

COST-EFFECTIVENESS ANALYSIS OF HIGH DOSE IV OMEPRAZOLE INFUSION AS ADJUVANT THERAPY TO ENDOSCOPIC HAEMOSTASIS FOR BLEEDING PEPTIC ULCERSLee KKC1, You JH1, Lau JYW2, Sung JJY3, Yung MY2, Suk-San Ho S1, Lee CW1, Chung SSCC2

1Department of Pharmacy, The Chinese University of Hong Kong, Hong Kong, China; 2Department of Surgery, The Chinese University of Hong Kong, Hong Kong, China; 3Department of Medicine & Therapeutics, The Chinese University of Hong Kong, Hong Kong, China

OBJECTIVES: To investigate the cost-effectiveness ofhigh dose IV omeprazole as an adjuvant therapy to endo-

scopic hemostasis for prevention of early ulcer rebleedingfrom a hospital perspective. A randomized, placebo-con-trolled clinical trial was conducted previously in a localhospital to study the effects of high dose IV omeprazole in-fusion on patient outcomes after endoscopic ulcer hemo-stasis. Although the interim data suggested that IV ome-prazole significantly decreased the rate of early rebleedingthat occurred 72 hours after therapy, the cost implicationof this therapy has not been examined.METHODS: The data of 157 patients who completed theabove study was analyzed. The percentages of patients whoexperienced early rebleeding were obtained from medicalrecords. The health care resources consumed by each pa-tient during the first 72 hours post endoscopic hemostasiswere also retrieved from their records and studied.RESULTS: Four of 80 (5%) patients in the omeprazolegroup and 17 of 77 (22%) patients from the placebogroup had rebleeding within 72 hrs after endoscopy andrequired further treatment. The treatment cost within 72hours post endoscopy of the IV omeprazole group waslower than that of the placebo group (HK$1312 per pa-tient vs. $3223 per patient, 1 US 5 7.8 HK). The cost-effectiveness ratios for the omeprazole group and placebogroup were $9,946 and $12,821, respectively, per early re-bleeding episode prevented.CONCLUSIONS: High dose omeprazole is more cost-effective in preventing early ulcer rebleeding than placeboafter endoscopic hemostasis.

GI3

EVALUATION OF PHARMACISTS’ INTERVENTIONS ON PRESCRIBING ERRORS OF NONSTEROIDAL ANTI-INFLAMMATORY DRUGS: COSTS SAVINGS AND CLINICAL EFFECTSCousin C, Misrahi L, Chamba G, Bauguil G, Colin CFaculty of Pharmacy, Biological and Pharmaceutical Science Institute, Lyon, France

OBJECTIVES: To determine the economical and clinicalimpact of pharmacists’ interventions in ambulatory carewithin the context of a prescribing error of NonsteroidalAnti-Inflammatory Drugs (NSAIDs).METHOD: A national survey was carried out in 900town pharmacies during 12 weeks to record all prescrip-tions with an anomaly like contraindication, interactionwhich could be dangerous for the patient. We used a deci-sion analysis to compare two strategies, with or without asystematic pharmacist’s intervention before a prescriptionof NSAIDs. The outcome was upper gastrointestinal sideeffects of NSAID therapy (peptic ulcer and complications)and we used a prescribing errors rate varying between0,5% to 2% to estimate the costs savings and to measurethe occurrence of peptic ulcer avoided. Computer simula-tion was performed with Tree Age 3.0.RESULTS: 446 cases of NSAIDs prescription errors werenotified including combination of NSAIDs, NSAIDs over-dose and NSAIDs prescription with risk factors like peptic

Abstracts 367

ulcer history or age .65. Study gave an estimation of49.390 prescriptions involving NSAIDs prescription atrisk of upper gastrointestinal side effects in France for oneyear. The pharmacist’s intervention before a NSAID pre-scription with an error lead to save 1.037.190 FF (158.119Euros) to 3.951.200 FF (602.356 Euros) per year for theSocial Security and lead to avoid 58 to 232 peptic ulcersper year (30 to 120 complicated ulcers).CONCLUSION: This study confirmed that pharmacist in-tervention is cost saving and has a clinical impact as regardprescribing errors. Moreover this effect may be larger aslong as we only took into account the upper gastrointesti-nal NSAIDs side effects.

GI4

PREDICTORS OF SAMPLE INCLUSION FOLLOWING THE APPLICATION OF CONTINUOUS ELIGIBILITY REQUIREMENTS IN SUBJECTS RECEIVING COLONOSCOPIESBramley T1, Margraf T2, Okamoto L3

1University of South Carolina, Columbia, SC, USA; 2PharMetrics Inc., Boston, MA, USA; 3NDC Health Information Services, Phoenix, AZ, USA

OBJECTIVES: To investigate the predictors of sample in-clusion following the application of continuous eligibilityrequirements for subjects receiving colonoscopies.METHODS: All subjects with a colonoscopy (CPT 545355–45385) in PharMetrics’ Integrated Outcomes Data-base between January 1, 1997 and June 30, 1997; who didnot have a prior colonoscopy were eligible for the study.For each subject, the date of his or her first colonoscopyduring the study period was identified. Subjects were thenclassified as In-Sample if they had data available for analy-sis 6 months prior to and following the index colonoscopydate. Conversely, subjects without data available for analy-sis 6 months prior to and following the index date wereconsidered Out-of-Sample. Subjects tumor types were clas-sified as benign (ICD-9-CM 5 211.3, 211.4,569.0), malig-nant (ICD-9-CM 5 154,153) or no-polyps (absence of anyof the tumor codes). Chronic comorbidities and gas-trointestinal complications were identified via the presenceof ICD-9-CM codes. For each subject, the number ofchronic conditions identified during the study period wascounted. To determine the effect of subject characteristicson sample inclusion, sample membership was modeled us-ing logistic regression.RESULTS: 6958 subjects met the inclusion criteria, ofwhich 58.28% were classified as In-Sample. Increasing age,female gender and increasing number of comorbiditieswere significantly associated with a greater likelihood ofsample inclusion. Benign tumors increased the likelihood ofsample inclusion (OR 5 1.17; 95% CI: 1.04–1.32), butmalignant tumors were not significantly associated withsample inclusion (OR 5 1.14; 95% CI: 0.95–1.39).CONCLUSIONS: Continuous eligibility requirementstend to favor the inclusion of older subjects, subjects withmore comorbid conditions, and females. Subjects with be-

nign tumors, but not malignant tumors, are more likely toremain in sample than those with no-polyps following theapplication of continuous eligibility requirements.

ECONOMIC ANALYSIS OF SERVICE DELIVERY

SD1

USING CONJOINT ANALYSIS TO ASSESS WOMEN’S PREFERENCES FOR MATERNITY CARE SERVICES DURING THE INTRAPARTUM STAGERatcliffe J1, Longworth L1, Boulton M2

1Health Economics Research Group, Brunel University, Uxbridge, UK; 2Department of Epidemiology and Public Health, Imperial College School of Medicine, London, UK

OBJECTIVES: To estimate the relative value attached toseveral main characteristics associated with the process ofmaternity care during the intrapartum stage for womenwho have chosen to deliver at home compared withwomen who have chosen to deliver in hospital.METHODS: The attributes included in the exercise werecontinuity of care, the location for delivery, the level ofpain relief available, the locus of control for decision-mak-ing and the probability of being transferred to another lo-cation during labor. The attributes and their associatedlevels were chosen as a consequence of a literature reviewand several focus groups held with women who had re-cently delivered either at home or in hospital. A self-com-pletion questionnaire using the conjoint analysis techniquewas developed and administered to two samples. The firstsample comprised 250 women who had delivered in hospi-tal within the preceding 12 months and who were clini-cally defined as low risk prior to their delivery. The secondsample comprised 180 women who had delivered at homewithin the preceding 12 months.RESULTS: Response rates of 52% and 63% were achievedfor the hospital and home birth groups respectively. For thehospital birth group, statistically significant attributes af-fecting preferences were the level of continuity of care expe-rienced (p , 0.001) and the likelihood of transfer duringlabor (p 5 0.012), with higher continuity of care and alower likelihood of transfer being preferred. For the homebirth group, statistically significant attributes included loca-tion of care (p , 0.001), pain relief (p 5 0.033) and thelikelihood of transfer during labor (p 5 0.028) with ahome location, more natural methods of pain relief and alower likelihood of transfer being preferred.CONCLUSIONS: If, as suggested by recent evidence, it isthe case that there are no differences in the clinical out-comes between home and hospital birth for low riskwomen, then traditional measures used in health econom-ics (e.g., quality adjusted life years or QALYs) would de-tect no difference between the alternative modes of deliv-ery. However, the results of this study suggests that aspectsassociated with the process of maternity care are also im-portant and these preferences differ considerably for

368 Abstracts

women who choose to deliver at home relative to womenwho choose to deliver in hospital. This study has wider im-plications in demonstrating the application of a techniquethat can be used to ascertain the utility or value attachedto attributes or characteristics associated with the processof health care.

SD2

GENERICS ONLY OR GENERICS ANDEFFICIENT DRUGS?Bendahan G, Benaque ADirecció d’Atenció Primària Eixample, Institut Català de la Salut, Barcelona, Spain

BACKGROUND: During 98 several strategies have beenimplemented to increase the prescription of efficient (ED)and generic (GD) drugs. In Catalonia to improve the ge-nerics’ use, economic incentives have been offered to somePrimary Care Teams. Although the general practitioners(GP) in our study were not included in this incentive’sproject, they received the intervention presented in thisstudy.OBJECTIVE: To evaluate an intervention to implementED and GD in a Primary Care District in Barcelona. METHODS Type of study: Pre-post intervention with con-trol group. Setting and period: Drug prescription follow upof the primary care district (296.000 inhabitants) and Cata-lonia (control group) during 97, 98 and January-March 99.Information source: Catalan Institut of Health’s pharma-ceutical computerised system. Intervention: Design of asheet of paper with information about 1) prices of differentdrugs for a same active principle, including those drugs re-sponsible for higher pharmaceutical cost and with ED 2) allgeneric drugs on the market and how to prescribe them. InJanuary 98, it was distributed to all GP. Variables: Numberand price of the drugs of 18 active principle and savings ob-tained.RESULTS: The savings obtained prescribing ED of ome-prazole, enalapril and nimodipino during the period 99–97 have increased a 182% in the district versus 103% inCatalonia. The % of GD is also greater in our district:Ranitidina 24% vs. 14%, Captopril 13,2% vs. 8,6%, Aci-clovir 14,8% vs. 11.8%; Alprazolam 8.2% vs. 5% Fluox-etina 3,73% vs. 2,7%. During January-March 99, our dis-trict saved 69 millions (pesetas), 90% by prescribing EDand 10% by generics.CONCLUSION: The GP collaboration and an instrumentwith updated information and easy to use have allowed toachieve a higher percentage of ED and GD prescription inour district. To rationalise drug’s prescription, it is neces-sary to take into account the ED.

SD3

USING ECONOMIC INFORMATION IN HOSPITAL FORMULARY DECISION-MAKINGSpäth HM1,2, Ferdjaoui N1,2, Carrère MO1,2

1GRESAC, Lyon, France; 2Université Lyon, Lyon, France

OBJECTIVES: Over the last 15 years pharmacoeconom-ics research has grown rapidly, but little is known aboutthe actual use of economic information by health-care de-cision-makers. We conducted a literature review of pa-pers addressing the use of economic information in drugselection for hospital formularies to determine its role indecision-making, the barriers that prevent its use and theimportance of these barriers.METHODS: Five bibliographic databases were searched:(1) Medline (2) Embase (3) Pascal (4) NHS EconomicEvaluation Database, and (5) International Pharmaceuti-cal Abstracts, from 1991 to 1998. In addition a manualsearch of the journal PharmacoEconomics and of the ref-erence sections of all retrieved papers was performed. Toappraise and summarise the publications we used a 7-point checklist.RESULTS: We assessed 34 papers. Case studies based onpersonal observations were reported in 27 (79%) papers,surveys in 5 (15%) and literature reviews in 2 (6%). In-put that were reported to have a greater impact on deci-sions than economic information were: (1) the efficacyand safety of drugs (mentioned in 100% of the papers)(2) dosage and administration procedures (50%) (3)quality of life considerations (32%), and (4) clinical ex-perience of practitioners (29%). The most important ofthe seven identified barriers to the use of economic infor-mation in decision-making were: (1) lack of credibility ofeconomic evaluations (65%) (2) economic data did notapply to the setting of the decision-makers (62%) (3) de-cision-makers had no expertise in appraising economicinformation (59%) (4) decision-makers did not have eco-nomic information and it was not possible to collect it(47%), and (5) lack of collaboration between producersof economic information and decision-makers (41%).CONCLUSIONS: The current use of economic informa-tion is limited compared with its potential use. More re-search is needed to determine decision-maker’s needs foreconomic information and to identify incentives to takethis into account.

SD4

QUALITY OF LIFE ASSESSMENT AS A NEW OUTCOME MEASURE IN CLINICAL PRACTICEPratheepawanit N, Salek M, Finlay I, Luscombe DMedicines Research Unit, Welsh School of Pharmacy and University of Wales College of Medicine, Cardiff, UK

OBJECTIVES: To assess the practicality of routine mea-surement of patients’ quality of life during an outpatientpalliative care clinic.METHOD: The study was a prospective evaluation ofpalliative care services provided at an outpatient clinic atVelindre NHS Trust, Cardiff. During a one-year period ofthe study, patients’ self-report of quality of life was imple-mented into the routine care. Each patient was asked tocomplete the revised McGill Quality of Life Questionnaire(MQOL) in the waiting room pre-consultation. After eachconsultation, doctors recorded their comments on the pa-

Abstracts 369

tients’ responses to MQOL that had been helpful in theirdecision making and the interventions arising from thatprocess.RESULTS: The total of 348 consultations took place dur-ing the study period; 274 consultations had the MQOLforms completed. This was accounted for 127 patients.There were 46 male and 81 female with mean age of 59years old (range 26–85). 66 patients (52%) attended theclinic for the first time. Of these 66 new patients, 31(47%) had a record of the MQOL on their subsequent vis-its. Compared with the first visits, the results show thequality of life had improved in every domain (i.e., physicalsymptoms, physical wellbeing, psychological and existen-tial domains) on their subsequent visits. The overall qual-ity of life score in particular was significantly improvedduring the follow-up visits (P , 0.05). Considering the useof the MQOL, 101 (37%) consultations had a record ofdoctors’ comments on utilizing the quality of life informa-tion and the majority (95%) indicated the MQOL as beinguseful during consultations, particularly information re-lated to physical symptoms, psychological and existentialdomains.CONCLUSIONS: The findings support the practicality ofquality of life assessment in clinical practice and its valueas an outcome measure and monitoring tool of palliativecare services in this population.

ECONOMIC & OUTCOMES STUDY RESULTS OF ENDOCRINE DISORDERS

ED1

COST-EFFECTIVENESS OF INTENSIVE METFORMIN VERSUS CONVENTIONAL THERAPY FOR OBESE PATIENTS WITHTYPE 2 DIABETESPalmer AJ1, Sendi PP1, Gozzoli V1, Brandt A1, Spinas GA2

1IMIB, Institute for Medical Informatics and Biostatistics, Riehen, Switzerland; 2Division of Endocrinology and Diabetes, University Hospital, Zürich, Switzerland

OBJECTIVES: The clinical results of the metformin sub-study of the United Kingdom Prospective Diabetes Study(UKPDS) were combined with published cost data throughmodeling techniques, allowing a cost-effectiveness analysisof the management of obese type 2 diabetes patients witheither conventional therapy or intensive therapy with met-formin from the NHS perspective.METHODS: Occurrence of diabetes-related complicationswere simulated using a Markov model. Probabilities forcomplications were taken from the UKPDS, and costs wereretrieved from published sources. Total direct costs andmortality over an 11-year period were determined for pa-tients randomized to either conventional therapy or inten-sive therapy with metformin. Incremental costs per life yeargained for metformin versus conventional therapy were cal-culated. Extensive sensitivity analysis was performed.RESULTS: Mean (SD) costs per patient over the 11 year fol-low-up period (discounted at 5% per annum) were £4043

(10’987) and £3613 (12’122) for patients randomized toeither conventional or intensive therapy with metformin, re-spectively. Intensive therapy with metformin led to a 0.21life years gained (discounted at 5% p.a.), and thereforedominated conventional therapy, with lower costs and in-creased life expectancy. Outcomes were most sensitive tovariations in the acquisition cost of metformin. Changes inevent rates and costs related to myocardial infarction, renalfailure and stroke also had an important impact.CONCLUSIONS: Within the limitations of the modelingstudy, intensive therapy with metformin is cost saving andimproves life expectancy in obese type 2 diabetes in theUK setting. These findings should be confirmed by the eco-nomic study performed alongside the UKPDS.

ED2

COST-EFFECTIVENESS OFACROMEGALY TREATMENTSWilson LS1, Shin JL2, Ezzat S2

1Department of Pharmacy, University of California, San Francisco, CA, USA; 2Department of Medicine, University of Toronto, Toronto, Ontario, Canada

OBJECTIVES: Acromegaly is a chronic disease with in-creased morbidity and mortality. For this and other pitu-itary tumors, the treatment choice changes between surgi-cal or medical. The new generation of somatostatinanalogues resulted in a recent challenge in attitude towardsthe traditional dominant role of pituitary surgery, but costsof medical treatment are a concern. We conducted a cost-effectiveness analysis comparing surgical treatment alonewith combinations of surgical and medical treatments andcomparing new and old formulations of medical treatment.METHODS: We evaluated retrospectively 53 patientswith acromegaly who underwent pituitary trans-sphenoidsurgery alone or in addition to medical therapy. Outcomeswere analyzed as person months spent in different healthstate outcomes, defined according to Growth Hormoneand Insulin-like growth factor-I levels. We used a decisionanalytic modeling approach and included cost per addedyear in a cured health state and cost per life year savedcomparing surgery only with either the addition of medicaltreatment to all those non cured, or to all post surgery pa-tients.RESULTS: Patients spent as much as 65% of time in non-cured health states. The average annual cost of acromegalytreatment was $8,700. The cost-effectiveness ratio of treat-ing patients with microadenomas or macroadenomas andpersistent disease following surgery was $30,733 and$35,892 respectively per added year spent in a cured state.New somatostatin formulations had little effect on theoverall results. The cost per life year saved was approxi-mately $32,000. The results were cost-effective over awide range of costs and outcomes.CONCLUSION: Medical management of acromegaly iscost-effective providing further justification for more ag-gressive treatment of patients with persistently active post-operative disease.

370 Abstracts

ED3

DEALING WITH CO-MORBIDITY AND RARE COMPLICATIONS IN A DIABETES TYPE 2 COST OF ILLNESS STUDYSpannheimer A1, Goertz A2, Reitberger U1, Neiss A3, Liebl A4

1Kendle GmbH & Co. GMI KG, Munich, Germany; 2SmithKline Beecham Pharma, Munich, Germany; 3Institute for Medical Statistics and Epidemiology at the Technical University, Munich, Germany; 4Diabetes Center Bogenhausen, Munich, Germany

OBJECTIVES: There is a general lack of data on the prev-alence of diabetic complications, especially on co-mor-bidity. As part of a pan-European cost of illness study(CODE-2: Cost of Diabetes in Europe, Type 2) an epide-miological cross-sectional study has been conducted withpractitioners in Germany prior to the main study in orderto investigate the frequency of major complications (car-diovascular, cerebrovascular, nephrological, neuropathy,ophthalmological).METHODS: A representative random sample of 138 phy-sicians (GPs and specialists) documented the age and com-plications of 2701 randomly selected type II2 diabetic pa-tients on the basis of medical files. Different complicationschemes were grouped in five strata out of which 809 pa-tients were randomly selected for a more detailed evalua-tion. Rare complications were over-represented and for es-timation of event rates weighted by the prevalence dataderived from the pre-study.RESULTS: Absence of complications was found in 48.5%of the patients. Co-morbidity appeared in the majority ofpatients with complications (44%–89%, depending on themain complication). Overall incidence in the diabetes popu-lation was estimated to be 0.78% for myocardial infarction,1.28% for stroke and 0.80% for amputations. Prevalenceof blindness accounted for 1.34%. Due to the samplingstrategy even rare complications like dialysis were suffi-ciently reported (three times the number expected in an un-stratified random sample) to estimate an overall prevalenceof 0.62%.CONCLUSIONS: The study provides for the first timeprecise data on co-morbidity. The extrapolated event ratesin diabetic patients showed a strong correspondence withpublished data where available. This method proved to bevery effective for the extrapolation of epidemiological databy avoiding double counting. The method is also useful forimproving quality of estimations in rare subgroups. In ad-dition, it builds a reliable and precise basis for the estima-tion of cost on a population level.

ED4

QUALITY OF LIFE AND USE OF HEALTHCARE RESOURCES IN GROWTH HORMONE-DEFICIENT ADULTS AFTER GROWTH HORMONE REPLACEMENT THERAPYHernberg-Ståhl E1, Abs R2, Bengtsson B-Å3,Feldt-Rasmussen U4, Monson JP5, Wilton P1, Wüster C6

1Pharmacia & Upjohn, Stockholm, Sweden; 2University Hospital Antwerp, Egedem, Belgium; 3Sahlgrenska University Hospital, Göteborg, Sweden; 4Rigshospitalet, Copenhagen, Denmark; 5St. Bartholomew’s Hospital, London, UK; 6University Hospital, Heidelberg, Germany

OBJECTIVES: Healthcare and indirect costs have beenshown to be higher in hypopituitary adults with untreatedgrowth hormone deficiency (GHD) than in the general pop-ulation. The present study assessed the effect of growth hor-mone (GH) replacement therapy, in terms of quality of life(QoL) and use of healthcare resources, in GH-deficient pa-tients included in KIMS (Pharmacia & Upjohn Interna-tional Metabolic Database), which is a pharmacoepidemio-logical survey of adult patients with GHD.METHODS: The number of visits to the doctor, numberof days in hospital, amount of sick-leave, physical activ-ity during leisure time and assistance required with nor-mal daily activities were recorded at entry into KIMS andafter 12 months of GH replacement.RESULTS: Data were available from 2260 patients atbaseline and from 1019 patients at 12 months. For thetotal group, there were significant (p , 0.001) decreasesover 12 months in mean number of visits to the doctor(from 2.7 to 1.6) and mean number of days of sick-leave(from 6.5 to 3.5) The mean number of hospital days de-creased from 0.92 to 0.59 (p , 0.12). Patients also re-ported significantly (p , 0.001) increased leisure-timephysical activity. Naïve patients (not previously treatedwith GH) reported significantly (p , 0.001) greater im-provements in physical activity over 12 months than non-naïve patients. Naïve patients needed significantly moreassistance with normal daily activities than non-naïve pa-tients at baseline (22% vs 15%; p , 0.001), but not after12 months (16% vs 14%). Significantly (p , 0.001)more women required assistance in daily activities thanmen, both at baseline (28% vs 10%) and after 12 months(23% vs 7%).CONCLUSIONS: GH replacement significantly reducesthe use of healthcare resources and improves QoL inadults with GHD.

Abstracts 371

Talking PostersDiscussion Groups

QUALITY OF LIFE METHODOLOGY ISSUES

TPQ1

THE EUROPEAN REGULATORY ISSUESON QUALITY OF LIFE ASSESSMENT(ERIQA) PROJECTAcquadro CMAPI Research Institute, Lyon, France, for the ERIQA Group

BACKGROUND: The last 20 years have seen the develop-ment of hundreds of health-related quality of life (HRQL)instruments which are increasingly being used in multi-national clinical trials. As a result, regulatory authoritiesare more and more faced with HRQL as a new outcomemeasure for the evaluation of new therapies. In 1997,Mapi Research Institute organised an exploratory meet-ing which brought together HRQL researchers and Euro-pean regulatory representatives. The main objective was tohighlight the regulatory authorities’ concerns about HRQLevaluation in clinical trials. Two conclusions were made:1. There is a need to rationalise the field of HRQL re-search and to make HRQL a credible criterion for theevaluation of new therapies by health authorities; 2. Thisissue can only be resolved through a better collaborativeeffort between key players: health authorities, HRQL re-searchers and pharmaceutical companies. Following theseconclusions, the ERIQA Project was launched.OBJECTIVE: To provide European regulatory authoritieswith guidance on how to assess the quality of HRQL stud-ies in clinical trials, and how to evaluate the validity ofHRQL claims.METHODS: The project is structured in 2 phases and 4steps. Phase I will lead to the production of 3 types of doc-uments, and presentations to regulatory authorities: 1. Areview of existing guidelines, focusing on EMEA docu-ments; 2. A reference document, i.e., a guidance for the reg-ulatory authorities to assess the quality of HRQL evalua-tion and the validity of HRQL claims; 3. Pilot guidelines inspecific indications: cancer and COPD asthma. Phase IIwill lead to consensus conferences with key players.RESULTS: Preliminary results of Phase I will be pre-sented. Collaboration with other similar US initiatives arealso planned.

TPQ2

A REVIEW OF AVAILABLE HEALTH-RELATED QUALITY OF LIFE (HRQL) REFERENCE VALUES IN THE MEDICAL LITERATUREKeininger DL, Conway K, Acquadro C, Anfray CMAPI Research Institute, Lyon, France

BACKGROUND: Within the last 20 years, a large num-ber of HRQL instruments have been developed and are in-

creasingly being used in multinational research and clinicalpractice. However, methodological issues relating to inter-pretation of HRQL results still need to be clarified andthere is a need for improved interpretation of HRQL re-sults. This could be achieved through the development ofreference values. Reference values provide a means ofcomparing the results of an instrument’s administrationacross populations and across disease severities. The use-fulness of reference values to interpret the burden ofchronic disease has been illustrated in numerous studies.OBJECTIVES: The purpose of this study was to reviewthe available medical literature to assess the current avail-ability of HRQL reference values.METHODS: A thorough search of the Medline and IndexMedicus database from 1966 to the present was con-ducted that included the following search terms: “norma-tive data”, norms, “reference levels”, “reference values”,“quality of life”, and “health-related quality of life”.RESULTS: A search of the Medline and Index Medicus da-tabases yielded eight articles that referred to the develop-ment of reference values and 61 articles describing compar-ative studies where comparison were made of SF-36 scoresto those of established reference values. These articles in-cluded the development and evaluation of reference valuesof the Medical Outcomes Study Short Form Health SurveySF-36 and the European Organization for Research andTreatment of Cancer Quality of Life Questionnaire-C30.CONCLUSIONS: Currently, few reference values exist,particularly, for disease specific questionnaires. Clearly,there is a critical need for the development of reference val-ues for different diseases and populations according toclinical and socio-demographic criteria. These referencevalues may be used to make comparisons across popula-tions and disease severities which will provide improvedinterpretation of the results of HRQL studies.

TPQ3

EUROPEAN GUIDANCE DOCUMENT FORTHE IMPROVEMENT OF THE INTEGRATION OF HEALTH-RELATED QUALITY OF LIFE(HRQL) ASSESSMENT IN THE DRUG REGULATORY PROCESSChassany O1, Fullerton S2,3, Sagnier P4, Aaronson N5, Marquis P6, for the European Regulatory Issues on Quality of Life Assessment (ERIQA) Group1Lariboisiere Hospital, Paris, France; 2Astra Hässle, Mölndal, Sweden; 3UCLA Department of Medicine, Los Angeles, CA, USA; 4Bayer AG, Leverkusen, Germany; 5The Netherlands Cancer Institute, Amsterdam, The Netherlands; 6MAPI Values, Lyon, France

OBJECTIVE: Many clinical trials include an HRQL as-sessment, but very few drugs have obtained labeling orpromotional approval. This is due in part to the poor qual-ity of HRQL assessment and reporting. The objective ofthe ERIQA group is to provide European regulatory au-thorities with guidance on how to assess the quality of

372 Abstracts

HRQL studies in clinical trials, and how to evaluate thevalidity of HRQL claims, for appropriate decision-mak-ing.METHOD: A guidance document and a checklist havebeen designed following a literature search and using theexperience of the ERIQA group’s members, i.e., HRQL re-searchers, pharmaceutical industry representatives, aca-demic people, and reviewers for regulatory authorities.RESULTS: The guidance document reviews the major is-sues of HRQL assessment in clinical trials, and especiallypractical considerations such as: selection of an HRQLquestionnaire (i.e., minimal properties required, validationof translated versions); implementation of a HRQL assess-ment (i.e., training of study personnel, mode of adminis-tration, eligibility criteria, data collection, prevention ofmissing data, respondent burden, multicenter trial); statis-tical analysis (i.e., justification of the sample size, handlingof missing data, handling of multiple statistical tests), andinterpretation. For each issue, recommendations are made,even when there is no definite answer (e.g., interpretationof results). All the issues to be prespecified in the researchprotocol are mentioned. The checklist summarizes all theissues. It is intended to help both regulatory authority re-viewers in performing their clinical trial reviews and spon-sors and investigators in conducting a clinical trial withHRQL data and writing the study report.CONCLUSION: The final objective is to reach a large Eu-ropean agreement upon this guidance document, to im-prove the quality of HRQL studies and to convince Euro-pean regulatory authorities of the usefulness and scientificvalue of HRQL assessment.

TPQ4

A EUROPEAN EQ-5D VALUE SET. MYTHOR REALITY?De Charro F, Weijnen T, on behalf of all the participants in the EQ-Net ProjectInstitute for Medical Technology Assessment (iMTA), Erasmus University, Rotterdam, The Netherlands

OBJECTIVES: To harmonize VAS values resulting from 11studies in 6 countries (England, Finland, Germany, TheNetherlands, Spain and Sweden) using the EQ-5D valuationquestionnaire, developed by the EuroQol Group. The aim isto make available a European EQ-5D value set and to gaininsight into the influences resulting from the prevailing dif-ferences in methodology and sample characteristics.METHODS: The database constructed for this researchproject contains data on more than 110.000 health statesfrom 8370 respondents. Appropriate transformation wasapplied to deal with ceiling and floor effects. Regressionmodels and a single value decomposition were applied toexplore the differences between the valuations in the 11samples and the influences of background variables onthe VAS values of a subset of 18 health states. Multilevelregression analysis on data of 49 health states was con-ducted to take into account characteristics at the individ-

ual respondent level while testing the impact of healthstate characteristics on the value set.RESULTS: In general, the impact of the background vari-ables such as age and experience of illness on the VAS val-uations appeared to be modest. Differences in study meth-odology have a higher impact on the VAS values butdifferences in the health states themselves were found to bethe major determinant of VAS values.CONCLUSIONS: When corrected for the influence of thebackground variables, the VAS values from the 6 countriesshowed a remarkable agreement. Differences can be attrib-uted to differences in survey methodology and samplecharacteristics rather than ‘cultural’ aspects. By using asimple instrument, ie., the EQ-5D valuation questionnaire,it seems that an internationally consistent EQ-5D value setcan be established.

TPQ5

MAPI RESEARCH INSTITUTE HEALTH-RELATED QUALITY OF LIFE OUTCOMES DATABASE (IQOD) PROJECTKeininger DL1, Conway K1, Acquadro C1, Marquis P2, Jambon B1

1MAPI Research Institute, Lyon, France; 2MAPI Values,Lyon, France

BACKGROUND: Health-related Quality of Life (HRQL)of patients is increasingly being used as an endpoint inmultinational studies. In this era of international collabo-ration, there is a critical need for improved interpretationof HRQL studies and the confirmation of conceptualequivalence of linguistically adapted HRQL instrumentsused in multinational studies. In order to facilitate thesetwo objectives, Mapi Research Institute has launched thedevelopment of IQOD.OBJECTIVES: The primary objective of the IQOD proj-ect is the development of an international database thatwill contain HRQL, socio-demographic, and clinical datafrom multinational HRQL studies. Once this database isdeveloped the following objectives can be realised: (1)Development of reference values to facilitate interpreta-tion of scale scores and making it possible to estimate therelative burden of various medical and psychological con-ditions across populations and across disease severities;(2) Evaluation of the conceptual equivalence of linguisti-cally adapted HRQL instruments used in multinationalstudies; (3) Development of item banking.METHODS: During the initial phase of this project, datawill be obtained from sponsors of international studies inwhich linguistically valid HRQL questionnaires, providedby Mapi Research Institute, were administered. There are13 instruments included in the initial phase of this project.The first step of the initial phase is the development of ref-erence values for the Women’s Health Questionnaire(WHQ) and the Minnesota Living with Heart FailureQuestionnaire (MLHF).RESULTS: Results of the initial phase will be presentedupon completion of analysis.

Abstracts 373

PHARMACY ISSUES

TPP1

IMPLEMENTING A MODEL OF RISK SHARINGOF PHARMACEUTICAL BENEFITS IN PRIMARY CAREGilabert A, Casals A, Sedano ECatalan Health Service, Autonomous Government of Catalonia, Barcelona, Spain

OBJECTIVE: To design and assess a model of risk shar-ing of pharmaceutical benefits in primary care based onallocating prescription budgets, assigning prescriptionquality standards and linking the results to an economi-cal benefit.METHODOLOGY: In 1998 a prescribing budget was as-signed to primary health care teams (PHCTs) of 22 basichealth areas (BHAs) (488,900 inhabitants). The methodol-ogy for allocating drug budgets was based on the calcula-tion of a maximum expenditure permitted (MEP) for eachPHCT by assigning a percent increase above the previousyear’s expenditure, proportionally inverse to its cost/inhab-itant. This cost/inhabitant was modulated by the percentageof population with free pharmaceutical coverage (popula-tion 65 years old or over and other specific populationgroups ) and the typology of the BHA (urban or rural). Pre-scribing quality standards (PQS) assigned for each PHCTconsisted of the intrinsic pharmacological value (IPhV) ofthe prescription and the defined daily doses per 1,000 inhab-itants per year (DHD) for anti-bacterial drugs and NSAIDs.Values for qualitative indicators were assigned dependingon the distance between the PHCT’s current value and thegroup average. Non-achievement of the objectives assignedfor both MEPs and PQSs was linked to a loss of 2% of thetotal budget allocation contracted with the PHCT.RESULTS: The assesment of the prescribing budget ob-jective showed that 36.4% of PCHTs had an expenditurebelow its MEPs and 63.6% over. Of PHCTs exceedingMEP, 42.9% presented a deviation of less than 5%,28.6% between 5–10%, 21.4% between 10–15% and7.1% greater than 15%. Qualitative evaluation showedthat 86.4% of PHCTs achieved the objective of IPhV,81.8% reached the DHD assigned for anti-bacterials and68% the DHD for NSAIDs.CONCLUSION: The model has permitted the CatalanHealth Service to implement the concept of risk sharingin the management of pharmaceutical benefits in primarycare, and the introduction of greater accountability forprescription costs in general practice.

TPP2

EVALUATION OF A THERAPEUTIC INTERCHANGE PROGRAM FOR CALCIUM CHANNEL BLOCKERS IN HYPERTENSIONNichol MB1, Harada ASM1, Jones J1, Gilderman A2, Salas J21University of Southern California, Los Angeles, CA, USA; 2Prescription Solutions, Costa Mesa, CA, USA

OBJECTIVES: This project evaluated a therapeutic inter-change program of a pharmacy benefit management firmthat initiated an amlodipine to felodipine therapeutic in-terchange program. The purpose of this study was to in-vestigate the effect of switching hypertension therapy in agroup of managed care patients on health-related qualityof life, patient satisfaction, and prescription utilizationand cost.METHOD: Three groups of managed care patients wereprospectively selected from five large primary care medi-cal groups in Texas and Oklahoma during the last sixmonths of 1997. A study group of patients undergoinginterchange (switching from amlodipine to felodipine),and two comparison groups of patients not undergoing amedication change were used (continuous amlodipineand continuous felodipine). Baseline (before a plannedinterchange) patient questionnaires were mailed to a totalof 802 patients (Groups 1–3) and eight-week follow-up(after the interchange) patient questionnaires were mailedto 295 patients who responded to the first survey. Thepatient questionnaires included the SF-36, symptom mea-sures, and satisfaction with care. Prescription claims datawere used for the utilization and cost analysis.RESULTS: Patients maintaining therapy on felodipinecontinuously were older than amlodipine or switched pa-tients and were more likely to be female. Switch patientshad lower total prescription utilization after the switchcompared to the continuous-use groups. Switch patientsalso had fewer hypertension-related prescriptions afterthe switch. There were no statistically significant differ-ences in quality of life. It appeared that switch patientswere more likely to change hypertension medications af-ter their initial switch.CONCLUSIONS: The clear difference in patient charac-teristics between the treatment groups implies an impor-tant selection bias. These differences may mean that pa-tients subject to the therapeutic interchange were selectedfor clinical or other reasons not immediately apparent.Subsequent analyses should control for these effects.

TPP3

PHARMACEUTICAL CARE AND HEALTHCARE UTILIZATIONFischer LR, DeFor T, Cooper S, Scott L, Boonstra D, Eelkema M, Hase K, Goodman MHealthPartners, Minneapolis, MN, USA

OBJECTIVES: To assess if pharmaceutical care is associ-ated with a decrease in overall health care utilization andcost. Pharmaceutical care entails an expanded role forpharmacists to provide patient counseling and monitoringabout medications. Few adequately designed studies haveinvestigated the impact on health care utilization, and, todate, findings are mixed.METHODS: A quasi-experimental design, using utiliza-tion/cost data from an HMO database. Intervention siteswere 3 staff and 3 contract pharmacies. HMO enrolleeswith chronic health conditions (lung or heart disease) were

374 Abstracts

identified through the HMO’s electronic pharmacy data-base. Two years of utilization data (Year-Before and Year-After intervention) were extracted from the HMO databasefor participants, refusers and controls, excluding peoplewho died, disenrolled, or discontinued their pharmacybenefit. Multiple linear regression analyses were per-formed with the following dependent variables: differencebetween Year-Before and Year-After in number of clinicvisits, prescriptions dispensed, unique medications, hospi-tal admits, and hospital days, as well as inpatient, outpa-tient, and total charges. Separate regression analyses wererun for 2 models: participants vs. controls and interven-tion [participants 1 refusers] vs. controls. Other variablesin the regression models were gender, age, Charlson Co-morbidity Index in Year-Before, and group (lung vs. heartdisease).RESULTS: Participants were more likely than controls tohave an increase in both clinic visits (on average, 1.8 morevisits) and prescriptions filled (2.4 more fills). Those in theintervention sample (participants 1 refusers) were alsosomewhat more likely than controls to increase their num-bers of unique medications (0.6 more). None of the otherregression models were significant for the main predictorvariables.CONCLUSIONS: Our study does not find that pharma-ceutical care decreases health care utilization or cost in thisHMO setting. To the contrary, pharmaceutical care ap-pears to be associated with an increase in utilization.

TPP4

POTENTIAL COST SAVINGS BY PRESCRIPTION OF THE CHEAPEST DRUGS AMONG THOSE WITH EQUAL CONTENT IN PRIMARY CAREAlegre de Rey E, Martinez Rodriguez L, Rabadan Asensio APharmacy Unit, Cadiz-Bahia Primary Care Area, Cadiz, Spain

OBJECTIVES: To calculate the cost savings if the cheapesttrade marks were prescribed. Competence among drugcompanies has contributed to a rising potential of saving,so making possible a more efficient use of drugs.METHODS: We take a database with every prescriptionmade in 1998 by a staff of 239 health-system physicians inCádiz-Bahía Primary Care Area. This Area has 468,908 in-habitants. The study was done by taking a sample of thosedrugs with more potential of cost saving: 6 antiulcer drugs,10 antibiotics, 8 antihypertensive and 9 NSAIDs.RESULTS: The real cost of prescription of these drugs in1998 was 3.583.500E. 790.841E (28,8%) could be savedif cheaper trade marks were prescribed. That proportionwas similar among anti-hypertensives (30,5%) NSAIDs(29,5%), antiulcer drugs (29,4%) and antibiotics (24,5%),but potential savings were greater for antihypertensives(414.762E) than for antiulcer drugs (232.573E), antibiot-ics (196.297E) or NSAIDs (187.265E). Related to specificdrugs, 10 drugs get more than 80% of potential savings,and only 4 get more than 50%. They are enalapril(23,5%, 268919E), omeprazol (12%, 138.000E), captopril(10,2%, 116.724 E), ibuprofen (8,2%, 93.459E). Among

antibiotics, we could reach 79% of potential savings withciprofloxacin and amoxicillin/clavulanate. 93% of poten-tial cost savings in antihypertensives could be reached withenalapril and captopril. With omeprazole only we couldreach 59% of savings in antiulcer drugs. There are notsuch differences among NSAIDs.CONCLUSIONS: A great proportion (25–30%) of costscould be saved if the physician knows and prescribes thecheapest alternatives among those with equal content. Spe-cial attention is needed about potential cost savings of fewdrugs.

ECONOMIC MODELING STUDY ISSUES

TPE1

ASSESSMENT OF THE APPROPRIATENESS OF CLINICAL GUIDELINES FOR CONTINUATION TREATMENT IN MAJOR DEPRESSION FROM A HEALTH ECONOMIC PERSPECTIVENuijten MJCMEDTAP International, Amsterdam, The Netherlands

OBJECTIVE: The primary objective of this study was toassess the appropriateness of the existing Dutch guidelines(“NHG-standard”) from a health economic perspective.Those guidelines recommend continuation treatment for pe-riod of 6 months. We also assessed additional maintenancetreatment for 1 year after the continuation period.METHODS: The assessment was based on a Markovmodel based on decision analytic techniques. For this analy-sis we defined 3 mutually exclusive states: no depression(ND), depression with treatment in ambulatory setting(DA), depression/treatment failure requiring hospitalization(DH). The outcomes for the model have been defined as:QALYs and direct medical, while the study was performedfrom the health insurance perspective in The Netherlands in1999. The probabilities of clinical events and therapeuticchoices as well as the utilities were based on published liter-ature. The medical resource use related to each state was ab-stracted from published literature and expert opinion. Theassociated 1999 unit costs of the used medical resourceswere derived from official Dutch tariff lists.RESULTS: The results of the baseline analysis showed thata treatment according to the guidelines will result in an in-cremental cost-effectiveness ratio of 81,700 NLG perQALY. Treatment according to the guidelines increased thetotal costs from NLG 49 to NLG 683, while the number ofQALYs increased from 0.45 to 0.47. Sensitivity analysesconfirmed the robustness of these findings. Extension oftreatment to maintenance treatment (6 months and 12months) reduced the incremental cost-effectiveness ratio toonly 12,330 NLG per QALY.CONCLUSION: In conclusion, based on the assumptionsused in the model, the Dutch clinical guidelines for depres-sion are appropriate from a health economic perspective,when they are extended to beyond continuation period untilmaintenance treatment.

Abstracts 375

TPE2

ECONOMIC EVALUATION OF ENOXAPARINAS POST DISCHARGE PROPHYLAXIS FOR DEEP VEIN THROMBOSIS (DVT) IN ELECTIVE HIP SURGERYDavies LM1, Richardson GA1, Cohen AT2

1Centre for Health Economics, University of York, York, UK; 2Department of Medicine, Guys, Kings, Thomas’s School of Medicine and Dentistry, Denmark Hill, London, UK

OBJECTIVES: Previous research suggests heparin to pre-vent DVT following orthopaedic surgery is value formoney. Recent trials indicate that extending the durationof prophylaxis may prevent additional DVT’s, but in-crease the cost of prophylaxis. The objective of this studywas to compare the expected costs and benefits of usingenoxaparin (a low molecular weight heparin) for the in-dex hospital admission only versus enoxaparin for hospi-tal admission and 21 days post discharge.METHODS: Decision analysis was used to combine prob-ability, resource use and unit cost data, from the perspec-tive of the UK NHS and patients. Expected costs and ben-efits of standard and extended regimes were estimated fora hypothetical cohort of 1000 patients undergoing electivehip replacement. Three alternative measures of outcomewere used: lives gained, life years gained and quality ad-justed life years (QALY) gained. Costs of events were esti-mated as quantity and type of health care resources used,multiplied by unit costs. Data for the model was based onpoint estimates from a review of clinical and economic lit-erature. Statistical techniques were not appropriate to as-sess the level of uncertainty associated with the results.Sensitivity analysis across minimum and maximum valuesof variables was used to test the robustness of the results.RESULTS: The expected cost per QALY for extendedenoxaparin was £3,145 (range £1268/QALY to £15212/QALY). Extended enoxaparin was cost saving if 66% ofpatients or carers could administer prophylaxis post dis-charge.CONCLUSIONS: Whilst conservative data were used tobias the analysis against extended enoxaparin, limitationsin the data and modelling techniques mean that the resultsof the analysis are only indicative of the potential valuefor money. The results indicated extended enoxaparinwas associated with a cost/QALY of £1268–£15212,which compares well with the cost/QALY for other healthcare interventions funded by the UK NHS.

TPE3

ECONOMIC EVALUATION OF DIFFERENT STRATEGIES OF A VACCINATION AGAINST HEPATITIS A AND HEPATITIS BSmala A1, Berger K1, Szucs T2

1MERG—Medical Economics Research Group, Munich, Germany; 2Universitäts Spital Zurich, Zurich, Switzerland

OBJECTIVES: To compare two combined vaccinationstrategies against hepatitis A and hepatitis B (at age 1–15

years or at age 11–15 years) with non-vaccination in chil-dren and youth, to analyse net-costs per strategy, cost-effectiveness and epidemiologic development in projectionfor the next 30 years in Germany.METHODS: A period of 3 times 10 years is given, model-ing present data on birth rate development, incidence andcourse of disease and treatment costs per case, calculatingincidence, cases of infection resp. disease and treatmentcosts, calculated from the TPP’s perspective. Costs per caseand courses were obtained retrospectively from standard-ized interviews with pediatricians and from the literature.Cost-effectiveness is assessed as costs per avoided hepatitisA or hepatitis B infection for each strategy.RESULTS: Significant drop in HAV and HAB incidence invaccinated age groups: Through vaccination of all 1–15year old children, a reduction of 57 600 new HAV infec-tions and of 45 800 new HBV infections in 30 years canbe demonstrated. In the vaccination group 11–15 years,19 800 new infections with HAV and 21 900 new infec-tions with HBV can be avoided. Vaccination costs over 30years amount to 1.5 billion DM in the vaccination strategy11–15 years and 3.7 billion DM in the vaccination strategy1–15 years. Significant decrease in new infections savestreatment costs of 1.6 billion DM for the vaccination at age11–15 years and 2.8 billion DM for the vaccination at age1–15 years. The cost-effectiveness of vaccination rangesfrom costs of 90 100 DM up to savings of - 20 200 DM peravoided infection, depending on cycle and vaccination strat-egy. Considering unreported cases of hepatitis A and B as 5fold of known incidence, the range of savings per avoidedinfection is between - 3 300 DM and - 38 700 DM.CONCLUSION: The initially more expensive combinedvaccine represents a cost-effective alternative to the solehepatitis B vaccination and also shows a more beneficialeffect from an epidemiological point of view. In Germany,the strategy of vaccination between 11–15 years seems tobe the most cost-effective—on the other hand the vaccina-tion of all between 1–15 years is superior considering theepidemiological effects.

TPE4

DEVELOPMENT OF A QUALITY ASSESSMENT CHECKLIST FOR THE APPRAISAL OF ECONOMIC MODELS: A CASE-STUDYIN OSTEOPOROSISHege U1, Pang F2, Tolley K3

1Department of Health Sciences, University of York, York, North Yorkshire, UK; 2Centre for Health Economics, University of York, York, North Yorkshire, UK; 3Pfizer Central Research, Sandwich, Kent, UK

OBJECTIVES: Deriving data on clinical outcomes to pop-ulate economic models requires a series of key decisionsupon the part of the economist. Ideally, to minimise thepotential for bias, the process of including data should besystematic. The objective of this research is to construct aquality assessment checklist for the inclusion of clinicaland epidemiological data in economic models, and fur-

376 Abstracts

thermore to illustrate its application in the area of os-teoporosis, where economic evaluations predominantlytake the form of decision-analytical models.METHODS: A systematic review involving electronic da-tabases and hand-searches was conducted of full economicevaluations of interventions in osteoporosis. The includedstudies were assessed using a checklist based on the Co-chrane Collaboration handbook for systematic reviews.The six main items of the checklist included the (i) specifi-cation of the research question (ii) comprehensiveness ofthe literature review (iii) critical appraisal of studies (iv)relevance of the data extraction (v) appropriateness of theanalysis and (vi) handling of uncertainty.RESULTS: 18 studies met the criteria for inclusion (UK 56, US 5 7, Australia 5 2). 16 studies considered hormonereplacement therapy, 3 calcitonin, 2 calcium and 2 bisphos-phonates. It was found that 28% satisfied 4 or more of thesix main criteria. It was also found that only 33% studiesdid a comprehensive literature review; primary studieswere predominantly epidemiological, but only 4 appraisedthe quality of the observational data.CONCLUSION: These results demonstrate methodologicalweaknesses in the process of including data in osteoporosismodels. Quality checklists for economic models shouldtherefore emphasise the identification, retrieval, extractionand pooling methods for the clinical outcome data. Furtherthere should be increased transparency in the reporting ofeconomic models to assist in model quality assessment andvalidation. Future research should focus on the impact of bi-ased data selection on cost effectiveness ratios.

TPE5

COST OF MANAGING BENIGN PROSTATE HYPERPLASIA IN BELGIUMAnnemans L1,2, Vanoverbeke N2, Closon MC3

1Free University of Brussels, Brussels, Belgium; 2HEDM, Mechelen, Belgium; 3Université Catholique de Louvain, Brussels, Belgium

OBJECTIVES: To study the cost of illness for patientsnewly diagnosed with benign prostate hyperplasia (BPH)in different scenarios.METHODS: Published data provide the clinical results ofcurrent relevant management for new BPH patients: al-pha-blockers (AB), watchful waiting (WW) or surgery (S).Based on these data, we simulated the history of patientswith moderate BPH in a state transitional model, startingwith one of the three treatment options. The model covers10 periods of 1 year, and simulates real life treatmentswitches (e.g., from WW to AB, from one AB to another,from WW to S, etc. . .). Ten centres were asked to providedata on medical management at different states, based ontheir local audits. Resulting direct medical costs were cal-culated from the health care insurance perspective and ap-plied into the model. Costs were discounted at 3%. Themodel was validated by comparison of predicted with ob-served surgery rates from a National hospital database (68hospitals).

RESULTS: Patients starting on WW have the lowest cumu-lative costs (3,750 Euro, SD 5 3,767 ), followed by patientsstarting with AB: 4,559 Euro for tamsulosin (SD 5 1,702)and 4,744 Euro for terazosin (SD 5 1,651). Introducing ABas second choice after WW has caused an observed reduc-tion in S rates in Belgium. If AB would not exist, the in-creased and earlier use of surgery would lead to a predictedextra cost per patient of 60 Euro per year during the first 5years, and in the long run to a break-even situation. Themodel overestimated observed surgical rates by 4.5%.CONCLUSIONS: The introduction of AB seems to leadto a delay of surgery in BPH patients, with short term sav-ings and long term break-even. The real life short-termsavings may be lower than the predicted, since the modeloverestimates the surgical rates.

ECONOMIC OBSERVATIONAL STUDY METHODOLOGY ISSUES

TPM1

DIRECT AND INDIRECT COST OF MIGRAINEIN ITALYArpinelli F, Manetta F, Roncolato M, Fabbri L, Rizzini PMedical Directorate, Glaxo Wellcome S.p.A., Verona, Italy

Migraine has an estimated prevalence of 11.4% in Italyand is associated with significant morbidity.OBJECTIVE: To estimate resource consumption (RC) andcosts related to migraine in Italy.METHODS: This was an observational, multicenter sur-vey of a patient population attended by general practitio-ners (GPs). Patients who visited their GPs on five consecu-tive days of two different weeks were asked if they sufferedfrom headache. Patients with headache were asked to par-ticipate in the study if their condition met the InternationalHeadache Society criteria for migraine. Migraineurs re-ceived a diary in which to record attacks and treatments.During subsequent GP visits, patients were administered anRC questionnaire on medical services and lost workdaysdue to migraine. Assessment of RC was from the commu-nity perspective. Values were assigned according to officialsources, and average annual costs in Euro (E) per patientwere calculated.RESULTS: During the first trimester of 1996, 902 GPs col-lected data on 71,588 evaluable patients. Prevalence of mi-graine in the sample was 11.6%. Health care resources (di-rect costs) accounted for 52.4% of the total annual cost,while lost workdays (indirect costs) accounted for the re-maining 47.6%. The total annual cost per migraneur wasE599. Direct costs included: medical consultations E49(8% of total); tests E63 (11%); drugs E43 (7%); and hospi-talizations E157 (26%). A total of 2.8 workdays per pa-tient were lost annually due to migraine. Extrapolated tothe total (active) population in Italy, this corresponds to15,016,870 lost workdays, amounting to over E1.5 billionin indirect costs per year. As Italian migraneurs exceed5,440,000, the total annual cost of migraine in Italy is overE3.2 billion.

Abstracts 377

CONCLUSIONS: Migraine is a costly condition, both interms of direct and indirect costs. Better therapeutic controlshould dramatically reduce costs related to healthcare re-source consumption and lost workdays due to migraine.

TPM2

ONE-YEAR REHOSPITALISATION RATEIN SCHIZOPHRENIA AND SCHIZOAFFECTIVE DISORDERTavcar R, Dernovsek M, Zvan VUniversity Psychiatric Hospital, Ljubljana-Polje, Slovenia

OBJECTIVES: Rehospitalisation rate is one indicator ofoutcome. Schizophrenia has a high relapse and rehospi-talisation rate (up to 50% per year). Antipsychotic main-tenance treatment is essential for relapse prevention. Theaim of our naturalistic follow-up study was to find outthe predictors of one-year rehospitalisation in patientswith schizophrenia or schizoaffective disorder.METHODS: All patients discharged in 1994 from theUniversity Psychiatric Hospital in Ljubljana, Slovenia,were followed up for one year. Symptoms were assessedwith Present State Examination. Data on rehospitalisa-tion were taken from the national Registry of PsychiatricInpatients. Predictors of rehospitalisation were assessedwith Cox proportional hazards survival analysis.RESULTS: Included were 447 patients (202 males and 245females) with a mean age of 39.1 years and 5.9 previoushospitalisations. The majority of patients (n 5 322) weredischarged with depot antipsychotics, 82 were on classicaloral antipsychotics while 43 were prescribed atypical agentsclozapine or risperidone. In one year 122 patients (27.3%)were rehospitalised, which is comparable to the results ofother studies. The following risk factors were found: diag-nosis of schizoaffective disorder (Cox regression coefficient(B) 5 0.56, p 5 0.027), frequent previous hospitalisations(B 5 0.31, p 5 0.021), inappropriate behaviour (B 5 0.43,p 5 0.004), and oral classical antipsychotics versus depot oratypical agents (B 5 0.37, p 5 0.047).CONCLUSIONS: This study yields some insight into theinfluences on rehospitalisation rate. Although rehospital-isation seems to depend mostly on disease-related factors,the choice of antipsychotic drugs, which is a modifiablerisk factor, may be important too.

TPM3

THE EFFECT OF OBSERVATION DURATION ON CLAIMS-BASED RISK ADJUSTMENT SCORESAndrews K, Powell-Threets K, Zhao ZCenter for Outcomes Measurement and Performance Assessment, Merck-Medco Managed Care, L.L.C., Montvale, NJ, USA

OBJECTIVES: To determine the association betweenclaims-based risk adjustment scores and duration of obser-vation and to ascertain whether statistical control for anyassociation is possible.

METHODS: We analyzed 18 months of claims data (1/1/1996–6/30/1998) from a large American company. We re-stricted the study population to those individuals that werecontinuously eligible for benefit coverage for the entire 18months (n 5 42,054). Drug-based risk adjustment scores(Chronic Disease Score) and medical-based risk adjustmentscores (Elixhauser Index) were calculated at 3,6,9,12, and18 months.RESULTS: We found a pronounced effect of time, withthe population’s mean Chronic Disease Score increasing23% (p , 0.05) in 1 year and the population’s mean Elix-hauser Index increasing by 19% (p , 0.05). More impor-tantly, the relationship between time and the mean risk ad-justment scores was linear and demonstrated a strongcorrelation (r2 . 0.96).CONCLUSIONS: The high correlation between some riskadjustment scores and duration of observation precludesdirect comparisons between individuals with differentlengths of observation. However, statistical control, e.g.,standardized scores derived from linear regression, for thebias due to length of observation is possible. Thus, restrict-ing analyses to individuals with a common duration of ob-servation may be unnecessary.

TPM4

EVALUATION OF HIV/AIDS TREATMENT REGIMENS AND CONCOMITANT THERAPY USING A LARGE PHARMACY CLAIMS DATABASEHutchison S, Beno CInformation Operations, PCS Health Systems, Inc., Scottsdale, AZ, USA

OBJECTIVES: To examine initial and ending treatmentregimens for HIV/AIDS patients, and determine rates oftreatment for opportunistic infections.METHODS: We conducted a retrospective databasestudy using a large pharmacy claims database. We identi-fied patients newly treated for HIV/AIDS during the first6 months of 1998 (N 5 494), then examined their phar-maceutical prescription fill patterns for the calendar year1998. All patients had continuous insurance eligibilityfrom July 1997 through December 1998, and did not fillprescriptions for AIDS drugs between July 1997 and De-cember 1997.RESULTS: Most patients initiated therapy on a 2- or 3-drug regimen (41% and 42% respectively). Most patientscontinued on the same treatment regimen they initiated on(67% to 71%) with exception of patients initiating on a 2-nucleoside regimen (only 45% continued on this regimento the end of the study period). The most common aug-mentation involved adding a protease inhibitor (14%) fol-lowed by adding a nucleoside (11%). PCP, Candidiasis,MAC, and depression were the most frequently treatedconcomitant therapy conditions. The rate of treatment forcomorbid conditions varied considerably depending on thepatient’s initial treatment regimen.CONCLUSIONS: This study provides direct evidence ofregimens used to treat HIV/AIDS, estimates of switching

WITHDRAW

N

378 Abstracts

patterns, and rates of comorbid conditions that exist in pa-tients treated on each regimen. These estimates can be usedto calculate and project cost of therapy for patients suffer-ing from HIV/AIDS.

TPM5

BURDEN OF BRONCHIAL ASTHMA: FAMILY EXPENSES FOR CHILD’S DISEASEProsekova E, Geltser B, Pashcenko OChildhood Town Clinical Hospital, Vladivostok State Medical University, Vladivostok, Russia

OBJECTIVES: The object of the present assay was to ana-lyze the type and amount of expenses for the family with achild suffering from bronchial asthma living in Vladivostokbetween 1996–1998.METHODS: 500 families with asthmatic children weresurveyed retrospectively and anonymously.RESULTS: Direct expenditures for the child’s diseasewere 10,98 6 0,94% of annual income. Pharmacotherapywas the main expense item (75,96 6 1,2%). Drug expensesfor one child were US$302, 86 6 10, 4 in 1996, US$356,72 6 19, 8 in 1997. In 1998 the dollar amount of pharma-cotherapy didn’t change greatly (US$344, 4 6 20, 2) but aslump in exchange value of the ruble without an adequateincrease of ruble income led to expenses growing as a per-centage of family income (22, 1 6 2, 1%). Self-controlmeans expenditures appeared in 1997 and by the end of1998 19 6 2,25% of families had sustained them. Doctorvisits and sanatorium and health resort expenses per childwere 7–8% of direct expenditures for asthma. Less than5% of families have the means to buy an air cleaning appa-ratus and hypoallergic linen. The unit price is US$100–2100 but annual income of a family with an asthmaticchild in Vladivostok in 1997 was US$3493 6 50, 4, in1998 2 US$2996 6 94, 5. Owing to the child’s disease oneof the parents in a third of families either had to be totallyunemployed or take low qualified work that aggravated fi-nancial loss. Annually 12,2 6 2,1 disablement days fall onthe family because of asthma.CONCLUSIONS: In order to evaluate a family’s realmeans to follow medical recommendations it’s necessary totake into consideration family expenses for asthma. Eco-nomic state aggravation in the country in August 1998 ledto abrupt reduction or interruption of basic asthma therapyin 37, 86 6 4, 36% families for financial reasons. 82, 8 61, 67% of parents find the disease to be a considerable fi-nancial burden for the family.

RANDOMIZED CLINICAL STUDY METHODOLOGY ISSUES

TPC1

COST-EFFECTIVENESS ANALYSIS OF BUDESONIDE VERSUS DISODIUM

CROMOGLYCATE THERAPY IN CHILDREN WITH MODERATE BRONCHIAL ASTHMASzende A1, Hoffer G2, Kelenhegyi K1, Cserhati E3, Spicak V4

1AstraZeneca, Torokbalint, Hungary; 2DOTE School of Public Health, Debrecen, Hungary; 3Semmelweis Medical University, Budapest, Hungary; 4Faculti Nemocnice Bulovka, Prague, Czech Republic

OBJECTIVES: To investigate the cost-effectiveness of us-ing an increased dose of disodium cromoglycate versusswitching to budesonide treatment in those children withmoderate asthma whose symptoms can not be controlledwith the recommended dose of cromoglycate.METHODS: An open, randomized, parallel group designclinical trial was conducted at 20 study centres. After a4-week run-in period, 131 children with uncontrolledsymptoms were randomized either to the treatment ofincreased disodium cromoglycate 4 3 40 mg daily orbudesonide 2 3 200 mg daily for a 12-week treatmentperiod. An intention-to-treat analysis was performed. Ef-ficacy variables were the morning PEF value and thenumber of symptom free days. The economic analysiswas based on a societal perspective. Protocol drivencosts were excluded from the analysis. Direct costs weremeasured as costs of study medication, rescue medica-tion, non-protocol driven visits, and number of daysstayed at hospital. Indirect cost measurement was basedon time missing from school. A sensitivity analysis wascarried out on resource use data to test final cost-effec-tiveness.RESULTS: The morning PEF value increased more frombaseline in the budesonide group (30.8 vs. 10.6; p 50.01). The average number of symptom free days washigher in the budesonide group (65.7 vs. 59.5; p 5 0.09).The average cost of therapy was lower in the budesonidegroup (HUF 7844 vs. HUF 16962; p , 0.0001). The costper symptom free day was HUF 119 in the budesonidegroup and HUF 285 in the cromoglycate group.CONCLUSIONS: Budesonide treatment was the domi-nant strategy as it lead to both superior clinical outcomesand cost savings.

TPC2

COST-MINIMIZATION ANALYSIS OF TWO TRIPLE REGIMENS FOR THE TREATMENT OF HELICOBACTER PYLORI-RELATED PEPTIC ULCER DISEASEYou JH1, Lee KKC1, Ho SSS1, Sung JJY2, Yung MY3, Lee CW1, Yee GCY4

1Department of Pharmacy Practice, The Chinese University of Hong Kong, Hong Kong, China; 2Department of Pharmacy, Medicine & Therapeutics, The Chinese University of Hong Kong, Hong Kong, China; 3Department of Surgery, The Chinese University of Hong Kong, Hong Kong, China; 4University of Nebraska, Omaha, NE, USA

Abstracts 379

OBJECTIVES: To compare the direct medical costs asso-ciated with two different regimens for the treatment ofHelicobacter pylori-related peptic ulcer disease (PUD)from a hospital perspective. The efficacy of a new regimen,including ranitidine bismuth citrate, amoxycillin and cla-rithromycin (RAC), was recently compared in a study at alocal teaching hospital with that of a regimen consisting ofomeprazole, amoxycillin and clarithromycin (OAC). Thestudy results showed no significant difference in the effi-cacy of the two regimens, but the cost implication of thetwo regimens was not examined. The drug cost of OAC(HK$364, $US 5 7.8HK) was 20% more than that ofRAC (HK$304) for a 7-day therapy.METHOD: Data from a controlled, randomized clinicaltrial conducted in Hong Kong was reanalyzed. Therecords of 100 patients with Helicobacter pylori-relatedPUD, who were previously randomized to receive eitherRAC or OAC, were reviewed. The hospital resources con-sumed during the period of PUD treatment were retrievedand studied. The total cost associated with each regimenper ulcer-healed patient was calculated and analyzed.RESULTS: Twelve of the 100 patients were excluded fromthe analysis because of incomplete documentation or non-compliance with the protocol of the clinical trial. Forty-one inpatients and 47 outpatients were included in theanalysis. In the inpatient group, there was no significantdifference between the median direct cost associated withOAC and RAC ($13,042 and $11,622, respectively; P 50.168). In the outpatient group, the median direct cost as-sociated with RAC was significantly lower than that ofOAC ($4,096 and $3,839, respectively; P 5 0.003).CONCLUSION: The direct medical costs associated withOAC and RAC were similar for inpatient treatment ofHelicobacter pylori-related PUD but RAC was less costlyin the outpatient setting.

TPC3

COST OF AML TREATMENT IN BELGIUM: RESULTS OF A RANDOMIZED TRIAL WITH AND WITHOUT FILGRASTIM USENoens L1, Fillet G2, Zachée P3, Van Schoubroek K4, Standaert B4

1UZ Gent, Gent, Belgium; 2CHU Sart Tilman, Liège, Belgium; 3AZ Stuyvenberg, Antwerpen, Belgium; 4Amgen, Brussels, Belgium

INTRODUCTION: A randomised multi-centre phase IIItrial using Filgrastim (5 mg/kg/day until neutrophil recov-ery) in induction and consolidation therapy for ‘de novo’adult acute myeloid leukaemia (AML) patients showedsafety and efficacy of the drug with significant reduction inhospital duration and IV anti-infective drug therapy (Heilet al, Blood, 1997, 90, 4710–4718).OBJECTIVE: Considering the Belgium patients enrolledin the trial to estimate the financial impact of Filgrastimuse in the treatment of AML for that country.METHODS: Retrospective data collection of resource usewas obtained from the 36 Belgian patients (20 cases and16 controls) enrolled through 3 hospitals. The data were

retrieved from Case Report Forms and hospital bills. Thecost perspective considered is the reimbursement authorityof Belgium. A cost-minimisation model is developed in-cluding the following resource items: hospital duration, IVanti-infective drug days, lab test days, blood transfusionunits, vials of Filgrastim, other drug use excluding chemo-therapy, and use of other diagnostic tests (Rx, Scans). Unitcosts in 1998 BEF are retrieved from the reimbursementauthority (RIZIV/INAMI), the Red Cross Blood Bank, thedatabase of the Belgium Pharmaceutical Association (APB),and a private database on cost of health care in Belgiumhospitals (CECODI).RESULTS: The cost model shows an average cost decreaseof 73.31 BEF (5,7%) per patient for induction and consol-idation therapy with Filgrastim. Sensitivity analysis onhospital day costs that may widely vary, shows a break-even point reached at a cost per day much lower than theminimum reimbursement cost (break-even point 5 952BEF).CONCLUSIONS: Filgrastim use in the treatment of AMLpatients in Belgium is likely to induce cost savings. Thecost results are conservative estimates that do not includeindirect cost evaluations and quality of life improvementof the patient due to earlier hospital discharge.

TPC4

AN ECONOMIC EVALUATION OF AMLODIPINE FOR THE TREATMENT OF NONISCHEMIC DILATED CARDIOMYOPATHY: THE PROSPECTIVE RANDOMIZED AMLODIPINE SURVIVAL EVALUATION (PRAISE)Glick HA1, Polsky D1, Schulman KA2, Martin BC3, O’Connor CM4

1University of Pennsylvania, Philadelphia, PA, USA; 2Georgetown University Medical Center, Washington, DC, USA; 3The University of Georgia, Athens, GA, USA; 4Duke University Medical Center, Durham, NC, USA

OBJECTIVE: To evaluate the economic implications ofamlodipine therapy in patients with advanced left ventric-ular dysfunction due to nonischemic dilated cardiomyopa-thy by using data from the Prospective Randomized Amlo-dipine Survival Evaluation (PRAISE).METHODS: By using a decision analytic model, costs andeffects were estimated for the first 2 years of observation inPRAISE and were projected for 30 years after initiation oftherapy (referred to as the lifetime projection).RESULTS: While statistical tests of the survival curvesindicated that amlodipine significantly improved survival(P , 0.001), differences in life expectancy (amlodipine,10.19 years (95% CI, 20.03 to 0.41 years during the first2 years of the trial; 12.89 years, 95% CI, 20.37 to 6.14years projected for the patient’s lifetime) were not signifi-cant. The ratios of cost per year of life saved were,$8000; those of cost per quality-adjusted year of lifesaved were ,$14,300. The confidence intervals for thecost-effectiveness ratios indicated that for the first 2 yearsof the trial, amlodipine was unlikely to have ratios

380 Abstracts

.$58,000 per year of life saved and $124,000 per quality-adjusted year of life saved. For the lifetime projection, itwas unlikely to have ratios .$28,000 per year of life savedand $54,000 per quality-adjusted year of life saved.CONCLUSIONS: Among patients whose heart failurewas due to nonischemic dilated cardiomyopathy, amlo-dipine therapy was good value for the cost. An economicevaluation should be repeated when the second PRAISEstudy—which was designed to evaluate the mortality ef-fects in heart failure of nonischemic etiology—is com-pleted.

ECONOMIC EVALUATION METHODOLOGICAL ISSUES

TPL1

STATE OF THE ART IN PHARMACOECONOMIC EVALUATION: A REVIEW OF METHODSAND PRACTICEPang F1, Tolley K2

1Centre for Health Economics, University of York, York, UK; 2Outcomes Research, Central Research, Pfizer, Sandwich, UK

OBJECTIVES: An international review of cost-effective-ness studies of pharmaceuticals published in the last 5years in peer reviewed journals was undertaken to identifythe extent to which “state of the art” methodology and ana-lytical techniques had been employed.METHODS: Three main approaches were taken for thisreview: (1) A literature search for cost-effectiveness analy-ses of pharmaceuticals 1994–99 was undertaken usingMEDLINE and other databases; (2) A list of “state of theart” methods and analytical techniques (defined as newand innovative, rather than established) were drawn from“methods” papers published in leading health economicsjournals in the past 10 years, and peer opinion; (3) Appli-cation of a simple quality scoring system to assess thequality of the reviewed papers.RESULTS: 30 (currently) economic evaluations of phar-maceuticals have to date been identified, originating fromseveral countries. The “state of the art” methods list in-cluded developments in a several main areas: data collec-tion and modelling approaches (e.g., RCTs, meta analysisand scenario analysis), cost measurement, analysis andhandling uncertainty (e.g., confidence intervals for ICERs,Monte Carlo simulations, Bayesian approaches to sensitiv-ity analysis). The quality scoring system is still in develop-ment (although pilot results hopefully due soon). The mainfinding was that only a few of the studies reviewed used“state of the art” methods, relying mostly on establishedapproaches to CEA.CONCLUSIONS: The development of new state of the artmethods in CEA in recent years, in particular with new de-velopments in statistical applications, has increased poten-tial quality and rigour of CEA results. However, thesemethods are not yet routinely used in actual pharmaceco-nomic evaluations and may not be until more fully inte-grated into the growing number of pharmacoeconomic

guidelines (linked to drug reimbursement) being producedin different countries.

TPL2

MULTINATIONAL ECONOMIC EVALUATIONS: A REVIEW OF PUBLISHED STUDIES, METHODOLOGICAL ISSUES AND PRACTICEPang FCentre for Health Economics, University of York, York, UK

OBJECTIVES: There is growing interest in the economicevaluation of pharmaceuticals at the multinational level.The purpose of these evaluations is to inform healthcaredecision-makers about the cost-effectiveness of pharma-ceuticals, whose interests are largely specific to their owncountries. The objective of this research is to identify andcritically appraise multinational economic evaluations re-lating to pharmaceuticals in a number of disease areas, toproduce a comprehensive list of the methodological con-siderations and to demonstrate how previous work can in-form and ensure optimal design and analysis of futuremultinational economic evaluations.METHODS: A systematic review involving databases (in-cluding MEDLINE, OHE, NHS) and hand-searches ofjournals was conducted for multinational economic evalu-ations. Each economic evaluation was assessed using a 70-point checklist specifically developed for multinationaleconomic evaluations, which evaluated design issues (studyquestion, study bias, outcomes) and analysis issues (datapooling, data presentation, data robustness and data gen-eralisability) and also scored against a previously devel-oped 35 point generic checklist based on the BMJ guide-lines. Simultaneously a review was performed on theliterature on generalisability and a survey was conductedin a number of countries to ascertain attitudes to general-isability of data from multinational clinical trials.RESULTS: 16 economic evaluations met the criteria for in-clusion, which were based on a variety of frameworks andall took the form of cost-effectiveness analyses (5 5 multi-national clinical-economic trials, 7 5 adaptation of singlecountry clinical-economic trials and 4 5 multinational de-cision-analytic models) With the 70-point checklist, it wasfound that the studies addressed different subsections ofthe checklist adequately and the quality of the studiesshowed further variation using the 35-point checklist.There were very few papers on generalisability, but the re-sults of the survey demonstrated a degree of convergence.CONCLUSIONS: Worldwide, the issue of multinationaleconomic evaluations is generating huge interest and is oneof the biggest challenges facing health economics today.However this area has not been extensively researched andthere is an urgent need for additional methodologicalwork. This research is a first step towards developing a setof guidelines for use in future studies relating to the design,analysis and presentation of multinational economicevaluations with the purpose of maximizing the general-isability of these studies and hence their value to decision-makers.

Abstracts 381

TPL3

BEST EVIDENCE IN PHARMACOECONOMICSMallarkey G, Venter-Lewis LAdis International Ltd, Auckland, New Zealand

OBJECTIVE: To determine which studies provide best evi-dence of the cost effectiveness of drugs.METHODS: During 1998, we found 1398 pharmacoeco-nomic studies had been published in over 2300 medicaljournals. We summarised and evaluated all of these usingan objective scoring system. The evaluation looked at studydesign, its conduct, and the adequacy of the published re-port. Points were awarded for key elements in trials (e.g.,clear statement of aims of study and study perspective; va-lidity of treatment alternatives; relevance of outcomes mea-sured; clear statement of sources of cost and outcome data;appropriate discounting used; accurate reporting and dis-cussion of results, relevance of author’s conclusions; clini-cal relevance of results. Points for each element were addedto give an overall score out of 100 for each study.

RESULTS: There were 101 studies that achieved a score>66%. Cardiovascular medicine featured in 28 of the 101top studies, followed by cancer chemotherapy (12) respira-tory medicine (11), antiviral therapy (11), vaccines (10),antibacterial drug therapy (8) and neuropsychotherapeutics(7). There were 40 of 101 studies from institutions based inthe USA. Canada produced 12 of the top studies followedby, France (9), England (9), Italy (7), Germany (7), Sweden(4), The Netherlands (4) and Australia (2). Glaxo Wellcomesupported 11 of the top studies followed by Bristol-MyersSquibb, Eli Lilly, Pharmacia & Upjohn and Wyeth Ayerst(3 each). The following journals are publishing the best-designed pharmacoeconomic studies: Pharmacoeconomics(17 studies); AIDS (6); Archives of Internal Medicine (3);British Medical Journal (3); Circulation (3); Clinical DrugInvestigation (3); Journal of Clinical Oncology (3); Journalof the American Medical Association (3); Vaccine (3).CONCLUSION: This analysis provides valuable qualita-tive and quantitative evidence of the extent and nature ofworldwide pharmacoeconomic publishing.

382 Abstracts

Poster Presentations

ECONOMIC AND OUTCOMES ISSUESOF CANCER

PCA1

PHARMACOECONOMIC EVALUATION OF FILGRASTIM (r-metHuG-CSF) TREATMENTIN AUTOLOGOUS BONEMARROW TRANSPLANTATIONPrivitera G1, Della Volpe A2, Tagliaferri E2, Vegni FE4, Bosisio C1, Bottoni A3, Pierfederici P3, Auxilia F1, Lambertenghi Deliliers G2

1Istituto di Igiene e Medicina Preventiva—Ospedale Maggiore di Milano (IRCCS), Milan, Italy; 2Centro Trapianti di Midollo Osseo—Ospedale Maggiore di Milano (IRCCS), Milan, Italy; 3Roche SpA—Ricerca Clinica e Drug Safety, Milan, Italy; 4London School of Hygiene & Tropical Medicine, Environmental Epidemiology Unit, London, UK

Minimization cost analysis of hospitalization of 40 patientsaffected by hemopoietic malignancies that underwent autol-ogous bone marrow transplantation has been performed.Clinical records of 27 patients treated with Filgrastim havebeen compared with records of 13 patients not treated. Theretrospective comparative analysis included days of hospi-talization, antibiotic therapy, nursing time, transfusion epi-sodes, microbiological cultures required. Overall averagehospitalization cost since transplantation for treated patientswas £65,537,774 versus £69,479,878 (p , 0.004). Averagehospitalization time since transplantation for treated pa-tients was 29.3 days versus 32.7 for not treated. Thus, hos-pitalization after transplantation was 3.4 days (HR 5.3;95% C.I. 2.0 to 13.9) shorter for treated, accounting for a10.4% saving in hospitalization time. Treated patients hadaverage 5.6 days less in sterile room (HR 5.0; 95% C.I. 2.0to 12.4). Every treated patient hospitalization raised savingsfor at least £3.942.104, taking into account the average sav-ings in hospitalization days after transplantation. Moreover,Filgrastim use lead to overall and “sterile” nursing timesav-ings of 6.5% and 16.8%, results in accordance with shorterhospitalization. Our results are similar to those reported byother studies. Suitable use of Filgrastim, reducing hospital-ization length and disease course, appears to be a valid wayto improve health-care efficiency.

PCA2

IRINOTECAN IN FIRST LINE TREATMENT OF METASTATIC COLORECTAL CANCER: IMPROVED SURVIVAL AND COST EFFECTIVENESS COMPARED WITH INFUSIONAL 5-FLUOROURACILJackson DL, Burrell ARhône-Poulenc Rorer, West Malling, UK

Douillard et al. (1999), have reported a multicenter, ran-domized, controlled open label study comparing irinote-

can (Campto®) in combination with 5-fluorouracil (5-FU) and folinic acid (FA) therapy with fluorouracil/FAalone as first line treatment for metastatic colorectal can-cer. They concluded that irinotecan in combination with5-FU offered consistent significant advantages in terms ofefficacy and clinical benefit over single agent 5-FU with-out any detrimental effects upon quality of life. We aimto relate these data to relevant costs within the UK and toevaluate the economic implications of the difference insurvival between the two treatment arms from the view-point of a UK NHS commissioner. This work developsthe second line assessment of irinotecan by Iveson et al.(1999) in the EJC, which concluded that irinotecanachieved results of £14,000 per life year saved in the sec-ond line setting. This cost-effectiveness analysis comparesthe economic implications of replacing 5-FU therapy as asingle agent (de Gramont regimen or AIO regimen) plusfolinic acid rescue with irinotecan in combination with5-FU/FA (de Gramont regimen, or AIO regimen). Drug ac-quisition costs are derived from the British National For-mulary (March 1999); unit costs for clinical consultationsand services are derived from relevant 1997/1998 costdatabases. Costs associated with treatment delivery anddisease complications are also considered. Indirect costsalthough important are not included, in line with theviewpoint of commissioners within the NHS. Althoughcumulative drug acquisition costs per patient are higherwith irinotecan and 5-FU/FA than with infusional 5-FU/FA therapy alone, these costs are at least partially offsetby lower cumulative costs per patient associated withtreatment of complications in the irinotecan plus 5-FU/FA arm than in the 5-FU/FA alone arm.

PCA3

TREATMENT PATHWAYS, RESOURCE USE AND COSTS IN THE MANAGEMENT OF SMALL CELL LUNG CANCEROliver E1, Kiebert G1, Hutton J1, Hall R2, Killan J3, Higgins B3, Paschen B4

1MEDTAP International, London, UK; 2Northern Cancer Network, Newcastle-upon-Tyne, UK; 3Royal Victoria Infirmary, Newcastle-upon-Tyne, UK; 4Merck KgaA, Darmstadt, Germany

OBJECTIVE: Lung cancer is a major cause of death in theUK and small cell lung cancer (SCLC) represents about20% of primary lung tumors. The costs associated with themanagement of SCLC are significant, however few studieshave been conducted in the UK to determine their true ex-tent. The aim of this study was to obtain an estimate of thecurrent patterns of treatment and associated resource useand costs for SCLC in the UK.METHODS: Study sites were two hospitals in Newcastle-upon-Tyne. A focus group meeting with local cliniciansclarified the expected pathways of SCLC care. Forms forretrospective patient record data extraction were developedon a per treatment phase basis. Data was collected on aconsecutive series of 106 patients diagnosed with SCLC be-tween 1994 and 1997. Unit costs were determined from lo-cal hospital accounts and secondary sources.

Abstracts 383

RESULTS: Per treatment phase average costs and associ-ated percentages for the first 56 patients were: referraland diagnosis £1,964 (11.21%); active therapy £6,147(35.09%); follow-up £1,103 (6.30%); retreatment(s) £3,804(21.72%); subsequent follow-up(s) £560 (3.20%) and ter-minal care £3,937 (22.47%). The high cost of active ther-apy was primarily due to the long inpatient stays associatedwith chemotherapy. Similarly, the high cost of terminalcare was predominantly due to hospitalizations and/or hos-pice stays. The total average cost of treatment pre-recur-rence exceeded that of post-recurrence (52.6% of costs vs.47.4%).CONCLUSION: The results of this analysis show that theaverage costs of treating SCLC are considerable, althoughthe variability between patients in terms of the type andquantity of resource use was very high. Such analyses pro-vide a useful insight into resources used in actual clinicalpractice. This study has identified the major cost driversand when combined with outcome data will form the basisof future economic evaluations of SCLC treatments.

PCA4

TRANSLATION OF THE QOL-RTI H&N QUESTIONNAIRE FOR ASSESSMENT OF HEALTH RELATED QUALITY OF LIFE IN PATIENTS UNDERGOING RADIATION THERAPY FOR HEAD AND NECK CANCERJohnson DJ1, Janda M2, Bressmann T3, Trotti A1, Günthner A4

1Division of Radiation Oncology, University of South Florida, Tampa, FL, USA; 2Department of Radiation Therapy, Vienna General Hospital, Vienna, Austria; 3Department of Oral and Maxillofacial Surgery, University of Technology, Munich, Germany; 4Department of Psychiatry, University of Tübingen, Tübingen, Germany

OBJECTIVES: To translate the QOL-RTI H&N for usewith German patients. The Quality of Life-RadiationTherapy Instrument Head & Neck (QOL-RTI) is a multi-dimensional psychometric instrument designed specificallyto assess health related quality of life (QoL) in patients un-dergoing radiation therapy. The head and neck module(H&N) is a companion module to the QOL-RTI for headand neck patients. This instrument differs from other ex-isting tools such as the EORTC H&N or the FACT H&Nby including more specific aspects of saliva, taste, swal-lowing, and local pain in a concise format.METHODS: Healthcare professionals from Austria andGermany translated and linguistically validated the instru-ment. The current literature reports that translation of ex-isting health status measures is best accomplished with thedouble-back translation method. This method was used toassure the accurate translation of the tool. It was deter-mined that there is no exact match for some words orphrases, so appropriate translations were derived.RESULTS: Specific difficulties encountered in the transla-tion were determining cultural nuances or slang in thesource document. For example, question number 7 in En-glish says I worry about my bills and in the German trans-

lation it now says I worry about my financial situation. Itwas determined that these and several other questions mayhave different literal meanings in English and German, sowords were changed to ensure the same intent (impliedmeanings) as the original questionnaire.CONCLUSIONS: It is necessary to take great care whentranslating existing health care measures from one lan-guage into another so that the translated version will havecontent and semantic equivalence across the cultures rep-resented in the project. Psychometric testing is currentlybeing conducted. In addition to German, the QOL-RTIH&N has been translated into Japanese, Italian, Greekand Spanish and is in various stages of validation and test-ing in these languages.

PCA5

EVALUATION OF POTENTIAL DRUG-INDUCED CHANGES IN HEALTH CARE UTILIZATION AND ATTENDANT COSTS: MODELS, DATABASE, AND EXAMPLESFarquhar IHoechst Marion Roussel, Bridgewater, NJ, USA

OBJECTIVE: The suggested approach (models and the da-tabase) enables one to evaluate changes in disease-specifichealth care utilization and associated costs which could re-sult from a health care intervention, including the introduc-tion of new medicines in various therapeutic areas.METHODS: Disease-specific estimates of health care utili-zation and associated medical expenses distributed by typesof care and sources of payment have been constructed byapplying a linear model with exogenous population weightsand normalizing coefficients to relational data files derivedfrom the National Medical Expenditure Survey data. Aproportional attributable risk model, which incorporatesexogenous parameters accounting for causality, inflation,discounting, human costs associated with lost earnings, andpopulation structural changes, has been developed for esti-mating direct and indirect costs of diseases. These con-structed cost estimates (direct, direct by types of care andindirect) have been used as input variables for benefit-costand cost-effectiveness models.RESULTS: Health care utilization profiles have been con-structed for lung cancer, non-small cell lung cancer(NSCLC), and NSCLC in stage III-IV. For a regimen in-cluding a new drug in addition to Taxol and Carboplatin,which will add at least 97 days to a median survival, to becost saving, it has to reduce the individual disease cost by28.3%. For the same regimen to enable a system cost sav-ing it has to ensure 34.6% changes in a system health carecost. This new product will be cost-effective at the thresh-old of $80,000 per life year gained if it is priced within$8,000–$13,700. Within this price range, the estimated in-dividual averted cost might reach $7,000–$17,000.CONCLUSIONS: The suggested approach allows one tolink clinical efficacy to plausible changes in health out-comes, health care utilization and attendant costs at any ofthe levels and types of health care delivery.

384 Abstracts

PCA6

COST OF POSTMENOPAUSAL BREAST CANCER IN BELGIUM: AN INCIDENCE BASED APPROACHAnnemans L1,2, Moeremans K2, Van Belle S3, Cocquyt V3, Gilis P4

1Free University of Brussels, Brussels, Belgium; 2HEDM, Mechelen, Belgium; 3University of Ghent, Ghent, Belgium; 4Eli Lilly Benelux, Brussels, Belgium

OBJECTIVES: To study the incidence-based cost of illnessfor postmenopausal women newly diagnosed with breastcancer, and to extrapolate this to a national level.METHODS: Published data from long term studies de-scribe the disease history of women diagnosed with breastcancer. We simulated this history in a state transitionalmodel, including 5 states (early cancer node negative; earlycancer node positive; local relapse; metastasis, all with fol-low-up; and death), covering 10 periods of 1 year. To cal-culate the costs of each state, an analysis of medicalrecords of 120 patients in different states was organized insix centers, representative for size and regional distribu-tion. In each centre, the records of 20 patients with either anew diagnosis, or a diagnosis of recurrence in early 1997were randomly selected and an independent physician in-vestigated the medical resource use up to end 1998. Re-sulting direct medical costs were calculated from thehealth care insurance perspective and applied into themodel, discounting at 3%. Based on national incidencedata, a population-based cost was derived. The model wasvalidated by comparison of predicted with observed me-tastasis rates and mortality.RESULTS: The cumulative 10 year cost per patient 531,900 Euro (range 5 10,300–152,400). The average costconstitutes of 47% hospital, 23% surgery and 20% drugcosts. The majority of costs occur in the year after diagno-sis and in the year before death. The total 10 year cost foran absolute incident population 5 237 Million Euro. Therelative difference between predicted and observed mortal-ity was 2.6%, and between predicted and observed me-tastases 3.5%.CONCLUSIONS: This incidence-based approach identi-fied the cost of postmenopausal breast cancer over timeand may serve as a baseline against which to assess new in-terventions in either its prevention or in its early treatment.

ECONOMIC AND OUTCOMES ISSUES OF DISEASES OF THE CENTRAL NERVOUS SYSTEM

PCN1

DIRECT AND INDIRECT TREATMENT COMPARISONS OF SECOND-GENERATION ANTIPSYCHOTIC DRUGSSauriol L1, Suissa S2, Laporta M2, Edwardes M2, Deslandes M2

1Hoechst Marion Roussel Canada Research Inc., Montreal, Quebec, Canada; 2Division of Clinical Epidemiology, Royal Victoria Hospital, Montreal, Quebec, Canada

OBJECTIVES: The objective of this meta-analysis is to as-sess whether the results from an indirect treatment com-parison are similar to those from randomized clinical trialswith antipsychotic drugs. A direct and indirect treatmentcomparison was conducted on olanzapine and risperidone,using haloperidol as the control group.METHODS: Clinical trials comparing olanzapine and ris-peridone to haloperidol were identified through literaturesearches, and were reviewed by four people. Extractedwere changes from baseline to endpoint for the BPRS to-tal, the PANSS negative, the percentage use of anticholin-ergic drugs, and the percentage of drop-outs. To conductthe meta-analysis, the fixed effect model was chosen overthe random effect method. Missing data were estimatedfrom regression analyses. Safety measures were difficult tocompare since the instruments used were different for theolanzapine and risperidone trials.RESULTS: The results reported a similar efficacy betweenthe two drugs. With the indirect method, differences foundwere a BPRS Total Change of 0.97 (21.1; 3.0) and aPANSS Negative Change of 0.95 (20.23; 2.1). With thedirect method, the results were 1.80 (21.4; 5.0) and 1.10(20.44; 2.6) for both outcomes, respectively. Results forthe other percentage of patients using anticholinergicdrugs, and the percentage of drop-outs estimated with theindirect method, tend to favor olanzapine. 19.3% morepatients in the olanzapine group, compared to the risperi-done group, used no anticholinergic drugs. As for thedrop-out rates, there were no statistically significant differ-ences between the two treatment groups.CONCLUSIONS: The comparison of the direct and indi-rect method using olanzapine and risperidone showed dif-ferences in results, although the conclusion for efficacywas the same. Differences may be due to the study dura-tion. Hence, this method must be used carefully when con-sidering different study designs.

PCN2

DOSES OF OLANZPINE, RISPERIDONE, AND HALOPERIDOL IN CLINICAL PRACTICE: RESULTS OF A PROSPECTIVE PHARMACOEOPIDEMIOGICAL STUDY (EFESO)Sacristán JA1, Gómez JC1, Gregor KJ2, Montejo AL3, Vieta E4 and the EFESO Study Group1Eli Lilly and Company, Madrid, Spain; 2Eli Lilly and Company, Windlesham, UK; 3Centro de Salud “La Alamedilla”, Hospital Clínico Univeristario, Salamanca, Spain; 4Hospital Clínic i Provincial, Barcelona, Spain

OBJECTIVE: Evaluate the doses of olanzapine, risperi-done, and haloperidol in outpatients with schizophrenia.METHODS: A 6-month, prospective, comparative, study.Data were collected by psychiatrists at mental health cen-ters and other outpatient facilities. The antipsychotic drug,doses, adverse events, CGI scale scores, and the GAF scalescores were recorded. Clinical response was defined as adecrease of at least 2 points on the CGI, with a final CGIscore <4.

Abstracts 385

RESULTS: 2,657 patients were included in this analysis.Initial and overall mean daily doses were 12.2 and 13 mgfor olanzapine, 5.2 and 5.4 mg for risperidone, and 13.9and 13.6 mg for haloperidol. Median daily doses were 10mg for olanzapine, 6 mg for risperidone, and 10 mg forhaloperidol. Significantly (p < 0.001) lower proportion ofolanzapine-treated patients (36.9%) experienced EPS thanrisperidone (49.6%) and haloperidol (76.0%) patients.Significantly (p < 0.001) lower proportion of olanzapine-treated patients (47.8%) experienced any adverse eventthan risperidone (57.2%) and haloperidol (79.8%) pa-tients. Significantly (p , 0.05) greater proportion of olan-zapine-treated patients (37.3%) were responders than ris-peridone patients (31.5%). Patients who had an initialCGI <5 had significantly (p , 0.001) higher overall meandaily doses than patients with an initial CGI ,5. Signifi-cantly (p , 0.001) lower proportion of olanzapine-treatedpatients (10.2%) were receiving concomitant anticholin-ergic medication at the end of the study than risperidone(19.9%) and haloperidol (44%) patients. Significantly (p ,0.001) lower proportion of olanzapine-treated patients(18.4%) were receiving other antipsychotics at the end ofthe study than risperidone patients (25.6%).CONCLUSION: Mean daily doses from a naturalistic set-ting were consistent with recommendations based on clini-cal trials. Olanzapine-treated patients were less likely toexperience EPS, any adverse event, or use of anticholin-ergic medications than risperidone and haloperidol pa-tients were. Olanzapine-treated patients were less likely toreceive other antipsychotic medications and more likely toexperience a response than risperidone patients.

PCN3

COST ANALYSIS OF SCHIZOPHRENIA TREATMENT IN GERMANY: A COMPARISONOF OLANZAPINE, RISPERIDONE, AND HALOPERIDOL USING A CLINICAL DECISION MODELSpannheimer A1, Clouth J2, Gregor KJ31Kendle International, Munich, Germany; 2Eli Lilly and Company, Bad Homburg, Germany; 3Eli Lilly and Company, Windlesham Surrey, UK

OBJECTIVES: A pharmacoeconomic analysis of schizo-phrenia treatment with olanzapine, risperidone and halo-peridol in Germany.METHODS: A decision model, which included clinicalparameters, that was previously developed by an interna-tional expert panel was used. Medical resource utiliza-tion and costs in Germany were determined by a Germanexpert panel and integrated into the model. Subse-quently, a cost analysis, a cost-effectiveness analysis anda cost-utility analysis were performed.RESULTS: The total costs per patient for 5 years werehighest for risperidone (DM 123 004). Total costs for ha-loperidol were DM 121 868 and total costs for olanzap-ine amounted to DM 119 572. Cost-effectiveness of ris-peridone and olanzapine compared to haloperidol was

better by the factor 2.3 and 2.5, respectively. The costsper quality-adjusted-life-year were lowest for olanzapinefollowed by risperidone and haloperidol.CONCLUSION: If total costs are considered, the highmedication costs of new antipsychotics are compensatedby savings due to better efficacy in treatment of negativesymptoms and less relapses.

PCN4

A COMPARISON OF ADULT MIGRAINE TREATMENT REGIMENS IN A MANAGEDCARE POPULATIONEaddy M1, Mcloney A2, Margraf T3, Okamoto L2

1University of South Carolina, Columbia, SC, USA; 2NDC Health Information Services, Phoenix, AZ, USA; 3PharMetrics, Inc., Boston, MA, USA

OBJECTIVES: To compare costs and characteristics ofmigraine patients receiving triptan therapy versus thosenot receiving triptan therapy.METHODS: All patients possessing a migraine diagnosis(ICD-9-CM 5 346) in the PharMetrics Integrated Out-comes Database over the age of 18 between January 1,1997 and March 31, 1998 were eligible for inclusion. Sub-jects were required to be in the database for at least 3months prior to and 3 months following the index date.Any subject receiving migraine treatment prior to theirfirst diagnosis was excluded from the study. Subjects werecategorized as being in either the triptan or non-triptantreatment groups based on the presence of a triptan pre-scription during the 90 days following their index date.Subject demographics, comorbidities and utilization ofother migraine treatments were assessed, and comparedacross treatment groups. Charges were captured over a 90-day period following the index date.RESULTS: A total of 8,018 patients met the inclusion cri-teria. Approximately 85% were female and the mean agewas 41.62 (Std. Dev 5 12.64). Over 90% did not receive atriptan prescription within 90 days of the index date. Hy-pertension was less commonly observed in triptan users;all other comorbidities occurred equally in both groups.Triptan users were more likely to consume other anti-migraine drugs than non-triptan users. Costs were verysimilar in triptan users (mean 5 $1751 Std. Dev. 5 4142)and non-triptan users (mean 5 $1613 Std Dev 5193).CONCLUSIONS: Triptans did not appear to be com-monly prescribed following an initial migraine diagnosis.Despite increased anti-migraine drug usage in triptan us-ers, costs between triptan and non-triptan users were verysimilar.

PCN5

THE USE OF INTERNATIONAL, DISEASE SPECIFIC, CROSS-SECTIONAL DATABASES FOR PHARMACOECONOMIC RESEARCHCrawford B1, Evans C1, Benford M2, Karavali M2

1MAPI Values, Boston, MA, USA; 2Adelphi, Bollington, UK

386 Abstracts

OBJECTIVE: To assess the feasibility of using interna-tional disease-specific, cross-sectional databases for phar-macoeconomic research.METHODS: We evaluated several large internationalcross-sectional, disease-specific databases for their appro-priateness in early and late phase research. A migraine da-tabase is presented as an example. The migraine databasecontained 699 physicians and 8,537 subjects in 6 coun-tries.RESULTS: Data elements could be used in modeling, costof care, patient segmentation, and cost-effectiveness stud-ies. For example, average resource use for France, Ger-many and the US for PCP visits (over the previous 12months) was 3.3, 4.0, and 4.3 days, respectively. Inpatienthospitalization days for the same countries were 0.1, 0.1and 1.9, respectively. For modeling purposes, transitionprobabilities could be based on the length of the migraineattack (in France): 25 hours without treatment and 7hours with treatment. Lost work time, important to mi-graineurs, was examined by severity group. During theprevious 6 months lost time from work for moderate mi-graines for France, Germany and the US was 11.4, 4.5,and 4.8, respectively. For severe patients, lost work timewas 6.6, 6.2, and 7.8 days, respectively.CONCLUSIONS: International databases offer some in-formation useful for PE research. Their strength relies onthe large number of patients and the validity of the data ishigher than using experts to estimate resource utilization.In addition, the international aspect of the database facili-tates cross-country comparisons and the adaptation ofmodels to local settings.

PCN6

MODELLING THE LONG-TERM COST EFFECTIVENESS OF RILUZOLE FOR THE TREATMENT OF AMYOTROPHIC LATERAL SCLEROSISTavakoli M, Davies HTO, Malek MUniversity of Saint Andrews, Department of Management, Fife, Scotland, UK

This study reports the results of a long-term cost effective-ness analysis of riluzole in the treatment of amyotrophiclateral sclerosis (ALS) in the United Kingdom. Long termanalysis was carried out using Markov modelling extrapo-lating the transition probabilities derived from clinical trialdata. The implications of life extension offered by riluzolewere assessed in terms of the time spent in each healthstate which ALS patients experience. This data was takenfrom a cohort of 954 patients drawn from a large random-ized, double blind, placebo controlled, multi-center trialbetween 1992 and 1994.RESULTS: Findings suggest that riluzole is effective inprolonging life and that a larger proportion of patientstreated with riluzole will stay longer in states where func-tional status is least impaired. The additional life expect-ancy of the intervention over the three-year period sug-gests that treatment of three patients for three years would

result in obtaining one additional life-year. Riluzole offsetsapproximately half the additional costs of treatmentthrough extending the time that patients spend in themilder phases. Therefore, cost per unadjusted additionallife-year gained was estimated at £8,587 over the expectedlifetime of ALS patients. The findings also show that incre-mental cost figures are not sensitive to the cost of healthstates.

PCN7

ECONOMIC ASPECTS OF AN INTERNATIONAL STUDY OF MAJOR DEPRESSIVE DISORDER IN PRIMARY CARE PATIENTS (THE LIDO STUDY)Chisholm D, Knapp M, Simon G, Treglia MCentre for the Economics of Mental Health, Institute of Psychiatry, King’s College, London, UK

OBJECTIVES: As part of a multi-national, observationalinvestigation of the quality of life and economic correlatesof recognized major depression in primary care (the LIDOstudy), to develop a research method for the collection ofa) individual health care service utilisation and associatedcosts data, and b) site-level socio-demographic and serviceprofiles.METHODS: Alternative approaches to the collection andaggregation of data relating to the utilisation, cost, struc-ture and provision of local health services were reviewed,together with relevant existing instruments. An iterativeprocess of cross-cultural validation was pursued in orderto ensure standardized definitions and semantic equiva-lence of services across sites.RESULTS: Three sets of instrumentation were developed:1) a “local socio-demographic and service profile” sched-ule giving data on key socio-demographic indicators andprimary/secondary care service characteristics at the local,regional and national level of each participating site; 2) aninterviewer-led resource utilisation questionnaire contain-ing internationally comparable items of potential servicereceipt (subsequently translated into local site languagesand incorporated into the assessment of each recruitedstudy subject); and 3) a set of unit cost templates for allitemized services, based on a common protocol and stan-dardized methodology. Both “purchasing power parities”and a “market basket” approach were used (and com-pared) as the basis for converting national service costsinto a single currency for the purposes of pooled analyses.A conceptual model depicting the inter-relationship be-tween costs, depressive symptoms and quality of life wasconstructed in order to generate and structure study hy-potheses and modes of analysis.CONCLUSION: Cross-cultural comparison of health ser-vice utilisation and costs is complicated by the heterogene-ity of service systems and the need to reflect the contextswithin which people live and receive their care. In order tobe locally meaningful yet internationally comparable, acarefully constructed approach towards data collection isrequired.

Abstracts 387

PCN8

ANTIPSYCHOTIC DRUG USE AND HOSPITALISATION IN PATIENTSWITH SCHIZOPHRENIAde Vries CS, Boyle DIR, MacDonald TMDundee University, Medicines Monitoring Unit, Dundee, Scotland

BACKGROUND: The treatment of schizophrenia withpsychotropic drugs is complicated by side effects and lackof efficacy. The newer, atypical psychotropic drugs mayhave fewer side effects and may reduce the relapses thatlead to hospitalization. We performed a population-basedstudy of psychotropic use, co-prescription, and hospital-izations in schizophrenic patients in the Dundee catchmentarea of Scotland.METHODS: All individuals within the population basewho were hospitalized for schizophrenia between January1980 and October 1995 were identified and followed upfrom January 1993 until October 1995. The followingpsychotropic drug groups were studied: phenothiazines(P), thioxanthenes (T), butyrophenones (B), diphenyl butylpiperidines (D), and atypical psychotropics (A). For thesedrug groups, community-based dispensed prescribing wasstudied and the following events were recorded: discontin-uation of psychotropic use, switches to other psychotro-pics, co-prescribing of other psychotropics, or co-prescrib-ing for the treatment of side effects. Prescribing adherencewas calculated as the number of patient days of medica-tion dispensed per 100 patient days. In each drug group,the incidence of hospitalizations was calculated. Differ-ences between drug groups were calculated as odds ratios.RESULTS: 353 schizophrenic patients with sufficient fol-low-up time were identified. Patients who used phenothi-azines were at lowest risk of hospitalization (OR 0.1; CI95

0.5–0.14), and switch behavior (OR 0.4; CI95 0.3–0.6).With phenothiazines as the reference group, odds ratiosfor hospitalization varied from 9.2 (B; CI95 4.6–18.4) to24.0 (D; CI95 8.6–65.0) for the other drug groups; odds ra-tios for switching varied from 1.7 (B; CI95 0.9–3.0) to 3.7(A; CI95 1.8–8.0), and odds ratios for co-prescribing variedfrom 0.9 (A; CI95 0.6–1.5) to 1.2 (B; CI95 0.7–2.0). Adher-ence to all drug groups was poor (P: 53%; B: 55%; D:66%; T: 44%; A: 67%).DISCUSSION: Hospitalizations, switches, and co-pre-scribing all were less frequent with phenothiazines than inany of the other psychotropic drug groups. Phenothiazinesare the drugs of first choice for schizophrenia in the UK,and these results probably represent channeling of second-line drugs to those that have failed or had adverse drug re-actions with phenothiazines. Compliance was poor withall drug groups and this is a cause for concern as a propor-tion of treatment failure may be as a result of this.

PCN9

COST-EFFECTIVENESS ANALYSIS AND SOCIAL ASPECTS OF THE PALMITATE PIPOTIASINE OUTPATIENT TREATMENT IN SCHIZOPHRENIAGurovich I1, Kobina S2, Litvischenko Y1, Lyubov E1, Shmukler A1

1Moscow Research Institute of Psychiatry of the Russian Federation, Moscow, Russia; 2Rhone-Poulence Rorer, Moscow, Russia

OBJECTIVES: This study focused on evaluation of cost-effectiveness in the treatment with palmitate pipotiazine(Piportil L 4) vs. oral forms of traditional neuroleptics inpublic psychiatric outpatient clinic in Moscow.METHODS: To 25 outpatients with ICD-10 defined diag-nosis of schizophrenia was prescribed palmitate pipotiazinefor a period of six months in the present “mirror”—imageanalysis. Our intention was that the study should interfereas little as possible with routine clinical practice. Clinicalimprovement, drug tolerability and social functioning ofthe patients were evaluated with the PANSS, CGI scale,EPRS scale and original checklist on social functioning andquality of life (Gurovich & Shmukler, 1998). Economicevaluation was performed including relevant data aboutthe direct and indirect costs for all of the patients.RESULTS: The data obtained indicate some statisticallysignificant (p , 0.05) clinical improvement with thePANSS, CGI scale as well as reduced drug side effects withESRS (p , 0.05). The improvement of social functioningand quality of life was indicated. The treatment of palmi-tate pipotiazine ensures a reduction of relapses, improvingquality of remission states and a facilitation of the socialadaptation of the patients. So direct and indirect costs werereduced. The study demonstrates cost-effectiveness of thetreatment with palmitate pipotiazine (especially for the pa-tients at high risk for future relapse and rehospitalization.)cf oral forms of conventional antipsychotics.CONCLUSIONS: It was concluded that the constant andprolonged outpatient maintenance treatment with palmi-tate pipotiazine should be cost saving alternative for themanagement of outpatients with schizophrenia in “realworld” practice.

PCN10

A CROSS-SECTIONAL ANALYSIS OF HEALTH CARE COST OF SSRI USERSGoodman M, DeFor T, Cooper S, Maciosek MHealthPartners Research Foundation, Bloomington, MN, USA

OBJECTIVES: Many pharmaco-economic studies havesought to identify the cost-effectiveness of different SSRIantidepressants. In order to replicate randomized con-trolled trials, many of these studies have reduced the SSRI-user population to small, often unrepresentative numbers.This study seeks to inform health plan decision-makers bystudying a one-year cross-sectional population of SSRI us-ers. This population-based perspective is useful in under-standing natural differences in SSRI use.METHODS: We identified 27,134 SSRI users in 1997 in alarge, mixed model HMO. For each user we identified totalSSRI days, SSRI cost, total pharmacy cost, total health plancost using charges, paid amounts, and a standardized costalgorithm, and comorbidities using the Charlson comorbid-ity index. Statistical methods appropriate to the highly

388 Abstracts

skewed distribution of health care expense were used to cal-culate adjusted mean expense for users of each product.RESULTS: SSRI users included children and adults. Unad-justed total paid charges were lowest for users of fluoxe-tine. However, both the paroxetine and sertraline popula-tions were significantly sicker as measured by the Charlsoncomorbidity index (difference 5 0.11, p , 0.05). Adjustedfor age, gender, and comorbidities, the mean paid chargeswere 15% higher for paroxetine (p , .001) compared tofluoxetine and not significantly different for sertralinecompared to fluoxetine. When total pharmacy charges areadded to the non-pharmacy charges, the paroxetine is20% and sertraline is 27% lower than fluoxetine. Users ofparoxetine are about 50% more likely to switch amongthe SSRIs. Switchers have 34% higher total charges.CONCLUSIONS: The use of SSRIs represents an impor-tant and growing portion of medical expense. This paperillustrates that total medical care cost should be used inmaking population-level treatment choices.

PCN11

UK SCHIZOPHRENIA CARE AND ASSESSMENT PROGRAM (UK-SCAP)—A PROSPECTIVE, OBSERVATIONAL STUDY OF THE TREATMENT AND OUTCOMES OF DRUG THERAPY FOR SCHZOPHRENIA IN A NATURALISTIC SETTINGMcKendrick J1, Stephenson DA1, Kody M2, Gregor KJ3, Kinsman GD4

1Eli Lilly and Company Limited, Basingstoke, UK; 2Eli Lilly and Company, Indianapolis, IN, USA; 3Eli Lilly and Company Limited, Windlesham, UK; 4The Lewin Group, Bracknell, UK

OBJECTIVES: The Schizophrenia Care and AssessmentProgram (SCAP) is a prospective observational study estab-lished in the US and Australia to comprehensively measurecharacteristics of care for schizophrenia patients in actualclinical practice while comparing a wide range of clinical,humanistic, and economic outcomes among older andnewer medications. A feasibility study was conducted to in-vestigate the potential for adapting SCAP to Western Euro-pean countries and to determine appropriate study design.METHODS: The feasibility study was carried out in sev-eral countries in Europe and included (1) interviews withproviders and planners of healthcare in national associa-tions of psychiatrists, medical associations and nationaland regional government (2) literature searches on a rangeof aspects and outcomes of schizophrenia and (3) inter-views to assess the key operational aspects of conducting aSCAP study in representative clinical sites. RESULTS: The feasibility study indicated that a UK-SCAPwould capture the treatment and budgetary issues raised byUK prescribers and administrators. The wide range of out-comes assessments in SCAP (including sociodemographics,clinical status and treatments, functional status, generalhealth status, quality of life, resource utilisation and carerburden) will distinguish the study and complement datafrom clinical trials. UK-SCAP will enroll approximately600 inpatients and outpatients (limited exclusion criteria)

from late-1999 and will follow this cohort for 3 years. Theprimary outcomes measurement tools will be the SCAP in-strument, developed from items derived from establishedmeasures and validated in the US SCAP study.CONCLUSION: It is important to complement informa-tion from randomized clinical trials with information from“real life” settings. As new drug therapies have shownpromise for improved treatment of schizophrenia, moreresearch is needed to understand the clinical, humanistic,and economic opportunities of these therapies.Submitted to ECNP Conference, London, September 1999.

PCN12

HEALTH ECONOMICS FOR N-OF-1 TRIALSQizilbash N1, Karnon J21SmithKline Beecham Pharmaceuticals/Oxford University, Essex, UK; 2Brunel University, Uxbridge, UK

Series of N-of-1 trials have long been used to help decide theusefulness of interventions in psychology, often where con-ventional trial designs are impractical or unethical. Recently,N-of-1 trials have been adopted to make decisions on theusefulness of a proven therapy in individual patients. Theiruse is increasing because many people do not respond to aproven treatment, a greater desire to individualize therapy toenhance the risk/benefit ratio, and cost-containment pres-sures. N-of-1 trials reduce between-subject variability. Theyare useful in chronic diseases, where the effects of the activetreatment are unlikely to carry-over to the next treatmentperiod. N-of-1 trials provide a unique vehicle for the collec-tion of economic data. Unlike most crossover trials, N-of-1trials typically involve repeated episodes of the alternativetreatments being administered over consecutive cycles. Thismeans that each patient in a series of N-of-1 trials providesmultiple observations for each treatment considered. As therequired sample size for the economic components of clini-cal trials have been shown to be large, such trials offer an al-ternative to large samples that may be unfeasible. In addi-tion, N-of-1 trials provide the means to estimate individualcost-effectiveness ratios, which may be incorporated into thepatient, or clinician, decision making process. The problemof follow-up must be addressed, particularly as evaluationsmeasuring QALYs necessarily require a life time horizon.However, if an intermediate outcome can be identified, en-abling extrapolation, then this problem can be addressed.Specific examples of the use of this design will be illustratedfrom the field of dementia, where the ‘number needed totreat’ to obtain benefit in one patient is high from symptom-atic improvement by approved cholinergic agents. The roleof economic data from N-of-1 trials to help drug develop-ment will also be discussed.

ECONOMIC AND OUTCOMES ISSUES OF CARDIOVASCULAR DISEASE

PCV1

DIRECT COSTS OF MAINTAINING NORMAL SINUS RHYTHM IN PATIENTS WITH ATRIAL

Abstracts 389

FIBRILLATION IN THE NETHERLANDS: A MODELING APPROACHBhattacharyya SK1, Delea TE2, Vera M2, Oster G2, Kuy AVD3

1Procter & Gamble Pharmaceuticals, Cincinnati, OH, USA; 2Policy Analysis Inc., Boston, MA, USA; 3Loon op Zand, The Netherlands

OBJECTIVES: To estimate the direct costs of maintainingnormal sinus rhythm (NSR) in patients with atrial fibrilla-tion (AF) with four antiarrhythmic (AA) drugs (sotalol,amiodarone, flecainide, and propafenone) in the Nether-lands (NL).METHODS: A Markov model, with a one month cyclelength and a one year treatment duration, was used to esti-mate expected costs of maintaining NSR. An expert panelof cardiologists was interviewed to collect data on the per-ceived safety and efficacy of the drugs, resource use (inpa-tient and outpatient visits, tests and procedures) for recur-rence of AF, treatment initiation, monitoring, and sideeffects. Cost data were obtained principally from the Cen-tral Organization for Tariffs in Health Care and the Na-tional Association of Pharmacists. Costs were estimatedseparately by drugs and for patients with and without con-comitant structural heart disease (SHD), then weighted bythe joint distribution of presence or absence of SHD, spe-cific use of a drug in these two populations, and/or marketshare of AA drugs.RESULTS: The weighted average annual cost of maintain-ing NSR in an AF patient was US$2,181.31. Inpatienttreatment costs, dominated mainly by costs of symptomrecurrence and therapy initiation, ranged from 78%(flecainide) to 92% (sotalol) of total costs. Costs of moni-toring and side effects were the highest for amiodaroneand sotalol, respectively. Drug costs, ranging from $60 forsotalol to $252 for flecainide, were the minor contributorsto the total costs. In patients with SHD, annual costs ofmaintaining NSR were $2,685 and $2,505 for sotalol andamiodarone, the two most common drugs in the NL, re-spectively.CONCLUSIONS: The estimated annual cost of maintain-ing NSR in patients with AF in the NL is as high asUS$2,685 per patient. Relatively lower AA drug costs areoffset by significant medical treatment costs, particularlyfor inpatient therapy initiation and recurrence of diseasesymptoms.

PCV2

THE COST-EFFECTIVENESS OF CHOLESTEROL-LOWERING TREATMENT IN THE PRIMARY PREVENTION OF CORONARY HEART DISEASE: THE CASE OF A COMPREHENSIVE DIETARY INTERVENTION PROGRAMLong KSection of Health Services Evaluation, Mayo Clinic, Rochester, MN, USA

OBJECTIVES: No studies to date have addressed the cost-effectiveness of lifelong dietary treatment when adminis-

tered in “real world” community practice. This study ex-amines whether a comprehensive intervention program todetect and treat, through dietary therapy, hypercholester-olemic patients is a cost-effective approach in the primaryprevention of morbidity and mortality from coronaryheart disease (CHD).METHODS: I developed a multidimensional model ofCHD incidence using results from the Physician-AssistedCholesterol Education Program (PACE), a prospectiverandomized trial in rural clinical practice, in combinationwith multivariate logistic risk functions from the Framing-ham Heart Study. The cost and health effects of this di-etary intervention are projected over a lifetime for patientswith known CHD risk factors. Results were also simulatedfor the pharmacologic treatment of lovastatin taken dailyin addition to dietary intervention. Changes in life expect-ancy, quality-adjusted life expectancy, and incrementalcost-effectiveness ratios ($/QALY), were the primary out-come measures of interest.RESULTS: PACE participants, on average, reduced theirtotal serum cholesterol levels 1.3% (CI 0.43–2.14). Rela-tive to no treatment, men on dietary treatment gained anestimated 4 to 26 days of quality-adjusted life expectancyat an estimated cost-effectiveness of $14,463 to $77,574($/QALY), depending on the age of treatment initiation.Women gained an estimated 3 to 19 days with cost-effec-tiveness ratios from $49,998 to $102,574 ($/QALY). Rel-ative to dietary treatment alone, the cost-effectiveness ofthe combined diet-plus-lovastatin treatment ranged from$43,620 to $49,776 ($/QALY) in men, and from $52,996to $74,718 ($/QALY) in women.CONCLUSIONS: Dietary intervention is a relatively cost-effective method of CHD prevention and should remainthe preferred treatment option for younger men andwomen—especially for those initially at highest risk ofCHD. Combined diet-plus-lovastatin therapy, however, isthe optimal treatment choice for older individuals and forthose who experience a reduced quality of life on stringentdietary regimes.

PCV3

COSTS OF MAINTAINING NORMAL SINUS RHYTHM IN PATIENTS WITH ATRIAL FIBRILLATION OR FLUTTER IN THE UNITED STATESBhattacharyya SK1, Delea TE2, Vera M2, Oster G2

1Procter & Gamble Pharmaceuticals, Mason, OH, USA; 2Policy Analysis Inc., Brookline, MA, USA

OBJECTIVES: The objective of the study was to estimatethe direct costs of maintaining normal sinus rhythm (NSR)in patients with atrial fibrillation or flutter (AF/F) with fiveantiarrhythmic (AA) agents (amiodarone, flecainide, pro-pafenone, quinidine, sotalol).METHODS: A Markov model was used to estimate theexpected costs of maintaining NSR in patients with AF/Fusing AA therapies. Estimates of the resource utilization(inpatient therapy initiation and monitoring, recurrence,

390 Abstracts

and side effects), and efficacy and safety of AA agents wereobtained from an expert panel of fifteen cardiologists. Unitcosts were estimated from published studies and Medicareand state reimbursement rates. Expected costs for eachtherapy were estimated for patients with and without struc-tural heart disease (SHD), then weighted by correspondingestimates of prescription frequencies to obtain expectedcosts for a typical patient.RESULTS: The annual costs of maintaining NSR in pa-tients with and without SHD ranged from $3,763 (quini-dine) to $4,749 (amiodarone) and from $2,912 (quinidine)to $3,773 (sotalol) per patient, respectively. For a typicalpatient, the total annual cost of maintaining NSR was$3,809. Costs of therapy initiation and side effects were es-timated to be $1,341 and $1,198, respectively; together,these constituted 67% of total treatment cost. Costs of in-patient care represented over 64% of total cost.CONCLUSIONS: The cost of maintaining NSR withavailable AA therapies is $3,809 for a typical AF/F patient.Costs of inpatient therapy initiation and side effects consti-tute the majority of the total cost. Total treatment costs arehighest for amiodarone and sotalol in patients with andwithout SHD, respectively.

PCV4

COST ANALYSIS OF AMLODIPINE VERSUS FELODIPINE ER IN THE TREATMENT OF MILD-TO-MODERATE HYPERTENSIONBerliner E, Radensky PMcDermott, Will & Emery, Miami, FL, USA

OBJECTIVE: To determine drug costs to establish andmaintain treatment for hypertension with amlodipine andfelodipine.METHODS: Data from published reports of 4 controlledtrials directly comparing amlodipine and felodipine wereanalyzed. Three trials were identified from a comprehen-sive Medline search; one trial was cited as a reference. Forone study, the unpublished study report supplemented in-formation from the published report. In all 4 trials, sub-jects with mild-to-moderate hypertension were started on5 mg of study drug and titrated to higher dosages after 2to 8 weeks if hypertension was not controlled. Maximumdosages of amlodipine and felodipine were 10 mg in 3studies. One study allowed titration to felodipine 20 mg.In another study, lisinopril was added if hypertension wasnot controlled by monotherapy. Data were pooled acrossstudies to determine the fraction of patients titrated togreater than 5 mg or dual therapy. For the cost analysis,average drug costs to establish and maintain treatmentwere calculated from dosages during and at the end ofeach trial, respectively. Where published reports did notindicate timing of withdrawals, upper and lower limitsyielded high and low estimates of cost to establish treat-ment. Average (US) wholesale prices were used for drugcosts. Statistical significance was assessed with a chi-square test.

RESULTS: Fewer amlodipine (45%) than felodipine(55%) patients were titrated to greater than 5 mg or dualtherapy (p 5 0.013). Average drug costs/patient-dayacross the four studies ranged:

Varying assumptions about timing of withdrawalschanged results by 3-percent or less. Higher costs withfelodipine were found in the study that allowed titration tofelodipine 20 mg.CONCLUSION: The higher price of amlodipine may beoffset by higher dosages or greater need for dual therapywith felodipine. Actual drug costs depend on dosages andcombinations used in practice.

PCV5

MANAGEMENT OF PERIPHERAL ARTERIAL DISEASE (PAD) PATIENTS: MODELLING THE RISK OF ISCHEMIC EVENTS AND THEIR ECONOMIC CONSEQUENCESLevy E1, Gabriel S2, Dinet J2, Carita P2

1Université Paris IX Dauphine, Paris, France; 2Health Economics & Outcome Research Department, Sanofi-Synthelabo, Gentilly, France

OBJECTIVES: The prevalence of PAD is known to beunderestimated as a large percentage of patients are notdiagnosed. As PAD is a risk marker of atherothromboticdisease, potential ischemic events (ischemic stroke, myo-cardial infarction) have to be taken into account whilestudying the economic burden of PAD.METHODS: A hypothetical cohort of 1000 patients withestablished PAD was used for studying, on a 2-year fol-low-up period basis, the average management of PAD pa-tients with and without complications (amputations, is-chemic stroke, myocardial infarction). Probabilities ofoccurrence of ischemic events came from previous largeclinical trials and were used in a Markov model. Thecosting was performed using French data costs.RESULTS: The average cost of management of a PADpatient over a two-year period was assessed taking intoaccount the probability of occurrence of complications(amputation, ischemic stroke, myocardial infarction)without any preventive treatment. This cost is almost 3times higher than the average cost of management of aPAD patient over a two-year period without any compli-cations (4501 Euros compared to 1707 Euros).CONCLUSION: The evaluation of the economic burdenof PAD has to take into account the risk of occurrence ofischemic events. In these patients, a preventive treatmentwith antiplatelet agents can significantly reduce the cost

Amlodipine Felodipine

Establish treatment $1.49–$1.59 $1.17–$1.71Maintain treatment $1.63–$1.74 $1.28–$1.98

Abstracts 391

of management by reducing the number of ischemic com-plications.

PCV6

COST OF SECOND LINE PHARMACEUTICAL MANAGEMENT OF ATRIAL FIBRILLATION IN FRANCEAnnemans L1,2, Moeremans K2, Aliot E3, de Jong P4

1Free University of Brussels, Brussels, Belgium; 2HEDM, Mechelen, Belgium; 3Hopital Central, Nancy, France; 4Biosense, Waterloo, Belgium

OBJECTIVES: Despite multiple pharmacological treatmentmany patients with atrial fibrillation (Afib) are not com-pletely “arrhythmia-free,” and remain at risk of chroniccomplications such as stroke and cardiac morbidity. Cur-rently, if first line treatment fails, most patients receive sec-ond line pharmacological treatment. The emergence of newtechnologies, aiming at “curing” the arrhythmia, e.g., com-puter guided ablation techniques, has increased interest inthe economic aspects of current second line pharmacologi-cal treatment of Afib. Therefore, we analyzed the average 5-year costs of second-line pharmacological management ofparoxysmal and persistent Afib in France.METHODS: The two most relevant current treatmentstrategies were included: (1) rhythm control, i.e., com-bined therapy with class Ic and III antiarrhythmics and (2)rate control, i.e., digoxin therapy together with a beta-blocker or a calcium antagonist. Both strategies are oftencombined with either anticoagulation therapy or aspirin.The analysis included probabilities of stroke, sudden deathand other cardiac or non-cardiac death, and direct medicalcosts of drugs, follow-up and prevention or treatment ofcomplications from a health care insurance perspective(discounted at 3% per annum). Clinical data were col-lected from prospective randomized clinical trials, medicalresource use was obtained from expert opinion (tworound pure Delphi panel) from 10 centers, and unit costswere obtained from official sources.RESULTS: The average total 5-year cost of pharmaco-logical management of Afib was FF108,500 per patient(SD 5 66,300; MIN 5 26,500; MAX 5 335,000). Fol-low-up visits and hospitalizations represented the largestproportion of costs (39%), followed by the cost of treat-ing congestive heart failure (CHF) (33%). Patients withprevalent CHF at start cost on average FF149,000. Pa-tients with persistent Afib cost on average FF40,000more than patients with paroxysmal Afib.CONCLUSIONS: Current second line pharmaceuticalmanagement of Afib is expensive and provides opportu-nities for new surgical or medical interventions.

PCV7

AN ECONOMIC ANALYSIS OF TELMISARTAN (MICARDIS®) TREATMENT FOR PEOPLE WITH MILD TO MODERATE HYPERTENSION IN THE UKBackhouse M1, Barnett AH2

1Research Triangle Institute, Manchester, UK; 2The University of Birmingham, Department of Medicine, Birmingham, UK

Telmisartan (Micardis®) is a new A-II inhibitor licensedin the UK for the treatment of people with mild to mod-erate hypertension. Randomised controlled trials (RCTs)suggest that the clinical outcomes of treatment with telm-isartan may compare favorably with other products in itsclass as well as with products in other classes. Thereforethe purpose of this analysis is to evaluate the extent towhich the clinical benefits are translated into economicbenefits from the perspective of the UK NHS.Data pertaining to key clinical outcomes (efficacy and ad-verse events) are derived from RCTs comparing telmisar-tan with several alternative medications. The ranges arevalidated by a review of the clinical literature. The costsof hypertension treatment and the management of ad-verse events are derived from treatment algorithms andpublished sources. The clinical and cost data are com-bined in a decision-analytic model to perform four sets ofanalyses:

1. Within-trial\3-month follow-up comparisons of hy-pertension medication costs and outcomes

2. Beyond-trial\15-month follow-up comparisons of hy-pertension medication costs and outcomes

3. Within-trial\3-month follow-up comparisons of hy-pertension medication costs and adverse event treat-ment costs and outcomes

4. Beyond-trial\15-month follow-up comparisons of hy-pertension medication costs and adverse event treat-ment costs and outcomes

The results are reported in both cost-effectiveness andcost-consequences format and are evaluated for their sen-sitivity to changes in key parameter values. The policyimplications and directions for future research are dis-cussed.

PCV8

HOSPITALIZATION PROFILE OF PATIENTS WITH ATRIAL FIBRILLATION OR FLUTTER: AN ADMINISTRATIVE DATABASE ANALYSISBhattacharyya SK, Laitinen DL, Brezovic CPProcter & Gamble Pharmaceuticals, Mason, OH, USA

OBJECTIVES: To evaluate hospital utilization in patientswith atrial fibrillation or flutter (AF/F) using a managedcare claims database.METHODS: A longitudinal integrated medical and phar-macy claims database was used to identify patients whoreceived an AF/F diagnosis in the emergency room and/orhospital in calendar year 1995. The earliest date for ei-ther primary or secondary diagnosis of AF/F was desig-nated as the index date for the purpose of analysis. Hos-pital utilization both before and after the index date wasobtained for four AF/F patient cohorts: patients with nohistory of congestive heart failure (CHF), myocardial inf-arction (MI) or stroke as recorded within the last 12

392 Abstracts

months of the index date including the index date (cohortCnn), patients with no CHF but MI or stroke (cohort Cny),patients with CHF but no MI or stroke (cohort Cyn), andpatients with both CHF and MI or stroke (cohort Cyy).RESULTS: Number of patients identified for cohorts Cnn,Cny, Cyn and Cyy were 319, 91, 195 and 59; among them52%, 41%, 52%, and 47% were female, respectively.The mean ages of patients in cohorts Cnn, Cny, Cyn and Cyy

were 68.5, 73.5, 75.2, and 76, respectively. The follow-ing table shows the hospital utilization (percent of AF/Fpatients hospitalized for any reason):

CONCLUSIONS: Cost intensive resource utilization, suchas hospitalization, increases up to 139.3% after the diag-nosis of AF/F as recorded in a managed care claims data-base. Although patients with concomitant illnesses, suchas CHF, MI, and stroke, have higher hospital utilizationcompared to patients without such diseases, AF/F con-tributes to the highest percentage increase for the healthi-est patient population.

PCV9

COST-COMPARISON ANALYSIS OF HOME-MONITORING AND AMBULANT MONITORING OF ORAL ANTICOAGULATIONIN SWITZERLANDNeeser K, Weiss C, Brandt A, Palmer AJIMIB, Institute for Medical Informatics and Biostatistics, Basel, Switzerland

BACKGROUND: A disease model was developed to com-pare clinical and economic outcomes of home monitoringversus ambulant monitoring in anticoagulation patientswith heart valve implants in two different locations. Theimproved outcomes seen with home monitoring result fromtighter control of the anticoagulation (AC) level. Thereby ahigher proportion of patients is kept within the therapeuticINR target range leading to a decreased incidence of ad-verse events.OBJECTIVE: The model compares the event-rates of ma-jor thromboembolic events (TE), major bleeding (BL) com-plications and mortality rates in cohorts with mechanicalheart valve replacement requiring oral anticoagulation withwarfarin.METHODS: The AC monitoring strategies comparedwere A) 52 times/year home monitoring of AC with theCoaguChek home-monitor, and B) monthly monitoringin an outpatient clinic (with laboratory tested AC levels).Direct costs of basic treatment, patient education and mon-

itoring, as well as treatment costs of complications werecollected from the Swiss third party payer perspective. Com-plication rates (TE, BL, mortality) were derived from pub-lished literature and official agency reports. The health eco-nomic analysis simulated an observational time of 10 years.Cost outcomes were discounted at a real annual rate of 5%.RESULTS: Self-monitoring by patients reduces the totalcosts (monitoring and complication related costs) per pa-tient from Swiss Francs (CHF) 8’988 to CHF 7’532 per pa-tient over the observation time of 10 years. Sensitivity anal-ysis were performed and identified the main cost drivers as:frequency of home monitoring, cost per home monitoringtest, cost and durability of home monitoring device.CONCLUSION: Demonstrated within the limitations ofthe modelling approach, the application of home monitor-ing in heart valve implant patients is dominant to ambulantmonitoring in outpatient clinics, with increased life qualityand decreased overall costs.

PCV10

PREDICTORS OF ADHERENCE TO LIPID-LOWERING THERAPYde Vries CS, Donnan PT, Waugh N, MacDonald TMUniversity of Dundee, Medicines Monitoring Unit, Dundee, Scotland

BACKGROUND: High serum lipid levels are an impor-tant risk factor for cardiovascular disease such as myocar-dial infarction and stroke. Recently, an American studydemonstrated that 5 years after beginning treatment, ap-proximately half of the cohort studied had stopped statintherapy.1 The aim of our study was to estimate adherenceto statins in Tayside and to determine if co-morbidity, so-cial deprivation, and co-prescribing were predictors of ad-herence.METHODS: Two studies were performed: Study 1, to de-termine adherence to statins over a 5-year period of follow-up and predictors of that adherence. Study 2, to find pre-dictors of compliance with statins in a 13-month follow-upperiod after publication of the 4S-study.2 The study popula-tion consisted of all individuals from the fixed populationbase who received more than one prescription for a statin.For each individual, adherence to therapy was defined asthe days for which statins had been dispensed divided bythe number of days they should have been dispensed. Peo-ple who had statins dispensed for 85% or more of theirdays in the community were classified as adherent. Logisticregression analyses were performed to determine predictorsof short-term and long-term adherence.RESULTS: Study 1. After 5 years, 65% of the original co-hort (n 5 392) were still collecting prescriptions for statins.89 (23%) were adherent to statin therapy. Predictors ofadherence were cardiovascular disease (OR: 1.19), femalegender (OR 1.67), and co-prescribing of nitrates (OR1.62). Study 2. Of the 1548 people who received a statin,652 (42%) were adherent to therapy. In this study, predic-tors were age (OR 1.02), use of nitrates (OR 0.76), and di-abetes (OR 0.65).

Cnn Cny Cyn Cyy Total

Before index date 13.5 35.2 33.9 47.5 25.5After index date 32.3 42.9 57.4 54.2 43.1Percent increase 139.3 21.9 69.3 14.1 69.0

Abstracts 393

DISCUSSION: The majority of people that were treatedwith statins stayed on these drugs although they used lessthan was prescribed to them. Adherence to statins in-creased over the years, however patients became less adher-ent the longer they used them.

PCV11

STUDY OF HEALTH CARE RESOURCE USE FOR ELDERLY PATIENTS WITH MILD AND MODERATE ARTERIAL HYPERTENSIONGeltzer B, Frisman MVladivostok State Medical University, Vladivostok, Russia

OBJECTIVES: To study health care expenditures for pa-tients with arterial hypertension.METHODS: We observed elderly patients (n 5 186; mid-dle age 68,4 6 2,2 years) from January 1997 to May 1999.We investigated four medicines from the basic groups ofhypotensive drugs: nifedipine-GITS; enalapril; diltiazem;indapamide. We defined duration of hospital treatment,number of ambulant visits and emergency care.RESULTS: Before treatment the middle number of ambu-lant visits was 6,01 per year. Against a background ofmonotherapy the number of visits was decreased: in nife-dipine-GITS group (NG)—to 2,75 (CI 1,2–2,4); in enala-pril group (EG)—to 3,27 (CI 1,5–3,1); in diltiazem group(DG)—to 3,76 (CI 1,4–2,3); in indapamide group (IG)—to5,28 (CI 1,4–3,2). Expenditures of health care were dimin-ished: in NG—by 54,53%; in EG—by 46,14%; in DG—by37,82%; in IG—by 12,69%. Before the treatment the mid-dle duration of hospital treatment was 22,84 days. Againsta background of monotherapy the duration of hospitaltreatment was decreased: in NG—to 14,25 days (CI 1,74–3,48); in EG—to 16,63 days (CI 1,35–3,21); in DG—to15,76 days (CI 1,42–3,38); in IG—to 16,57 days (CI 1,41–3,23). Expenditures of health care were diminished: inNG—by 37,62%; in EG—by 27,2%; in DG—by 31,01%;in IG—by 27,47%. Combined economy in NG was48,2%; in EG—27,76%; in DG—36,61%; in IG—32,3%.RESULTS: This investigation provides comparative evalu-ation of economy of health care resource with differentmedicines.

PCV12

ECONOMIC EVALUATION OF ACE INHIBITORS IN SPAINAsensio J, Hermida H, Abbas IKendall Institute, Barcelona, Spain

OBJECTIVES: To evaluate which ACE inhibitor, of the10 that are marketed in Spain, offers the most favorablecost-effectiveness relationship in the treatment of light-moderate arterial HBP.MATERIAL AND METHODS: Only direct costs derivedfrom treatment were taken into consideration. We havecalculated the cost of treatment per day (CTD) in the fol-lowing way: By grouping the different pharmaceutical

products marketed in Spain by active ingredients. The priceof the daily defined dose (DDD) was established for eachproduct by dividing its price by the number of tablets thatthe DDD requires in accordance with the recommendeddose in the information leaflet.RESULTS: In more or less objective terms it is acceptedthat therapeutic efficiency within the group of ACE inhibi-tors is very similar and so efficiency depends exclusively onthe cost of the product. This has lead to the carrying out ofan analysis of the minimization of costs, and this showsthat the alternative with the lowest cost per controlled pa-tient is Enalapril, so substituting this option for the rest ofACE inhibitor prescriptions could result in very significantsavings. Potential savings could reach 10,801 Million Ptas.per annum if the lower priced Enalapril were used.CONCLUSIONS: Enalapril is the most cost-effective ACEinhibitor for the treatment of light and moderate HBP inSpain. Substituting Enalapril for prescriptions for the restof ACE inhibitors could lead to a significant saving poten-tial in pharmaceutical spending. Changing to lower pricedEnalapril could increase the figure for potential savings byup to almost 11,000 Million Ptas. per annum which makesit the most efficient option.

ECONOMIC AND OUTCOMES ISSUES OF ENDOCRINE DISORDERS

PEN1

QUALITY-OF-LIFE-EVALUATION OF THE GERMAN S.A.T.-TRIAL - COMPARISON OF SF-36, EUROQOL, TIME-TRADE-OFF METHOD AND WILLINGNESS-TO-PAYWendland G, Kurscheid T, Lauterbach KWUniversity of Cologne, Köln, Germany

OBJECTIVE: To assess the effectiveness of evidence-basedtreatment of obesity in a primary care setting, we con-ducted the S.A.T.-trial (Sibutramine in Adiposity Trial)which is a multicenter, double-blind, placebo-controlled,randomized parallel group comparison of 54-week contin-uous therapy with Sibutramine 15 mg or Placebo in addi-tion to a standardized non-pharmacological therapy con-sisting of a calorie-restricted diet, modification of eatingbehavior and lifestyle, and an activity programme. Patientswith a BMI between 30 and 40 are included. The primaryendpoint in this study is weight loss in kg between the lastmeasurement of body weight (visit 11) and the measure-ment at the first visit. Secondary endpoints are amongstother clinical parameters health economic and quality oflife data. Since the random code is still not broken, the re-sults given in this abstract refer to all randomized obesepatients in the study (both Sibutramine and placebogroup). The presentation in November will show the re-sults for the two patient groups separately.QUALITY OF LIFE MEASURES: Quality of life is mea-sured by 4 different instruments: the Short Form 36 (SF-36):the EuroQol, the Time-trade-off-Method (TTO) and theWillingness-to-pay-Method (WTP). The SF-36 is a general,

394 Abstracts

standardized and validated Quality-of-life-instrument, con-sisting of 36 questions that can be aggregated into 8 healthdimensions. The EuroQol is also a general, standardizedand validated quality-of-life-questionnaire, having the ad-vantage of aggregating the five health dimensions into onequality-of-life-index. Therefore, the EuroQol is frequentlyused in health economic evaluations to calculate Quality ad-justed life years saved (QALYs). In the TTO interview pa-tients are asked questions to calculate a quality-of-life-indexfor each patient stratified by age. In the willingness-to-pay-questionnaire, obese patients are asked how much theywould be prepared to pay per month for an immediate andsustained weight loss of 10 kg from their current bodyweight. The answers are stratified by net income classes.RESULTS: Quality of life measured by the SF 36 has im-proved in all 8 health dimensions at visit 2 compared tovisit 1. Especially energy/fatigue is impaired in obese pa-tients. The quality-of-life-index measured by the EuroQolhas slightly improved from visit 1 to visit 2 over all age-groups. The quality-of-life-index of the TTO-interviewand the EuroQol-questionnaire are very similar except forthe age group ,35 years. Willingness-to-pay correlatespositively with net income.DISCUSSION: The results are preliminary, but in line withthe published literature. They reveal the massive burden ofobese patients.

PEN2

METHODOLOGICAL ASPECTS OF MULTI-COUNTRY COST OF ILLNESS STUDIES—2: INTERNET BASED DATA MANAGEMENTKirsch J1, Wilfling M2, Elfering I2, Hood S1

1Health Economics Dept., SmithKline Beecham Pharmaceuticals, Harlow, Essex, UK; 2MedicalData Service, Münster, Germany

OBJECTIVE: Managing the data collected in multi-countrycost of illness studies is a major task. The objective of thistask is to ensure the accuracy and confidentiality of the datacollected and to provide a flexible platform for analysis.METHOD: The CODE-2* (Costs of Diabetes in Europetype 2) study was conducted simultaneously in 8 Europeancountries to evaluate the total burden of type 2 diabetes.Resource utilisation, clinical and demographic data werecollected using specifically designed data collection instru-ments. Over 3 million discrete items of data relating to over7000 patients were collected and entered, transmitted andstored within a centralized database. This was achieved us-ing HTML and XML programming and software technol-ogy as the basis for data entry and data file creation andIDT** encryption techniques to ensure data privacy fordata sent via the Internet to a central server database. Thesenovel systems offer a number of advantages such as: userfriendly data entry screens, on line data validation, ‘limit-less’ data entry for certain resource items (e.g., practitionervisits), on line program updates, exception data reports, filestatus reports, confidential data transfer, central data vali-dation, derived variable creation, on line analytics and

progress reports, and data export functionality to analyti-cal packages such as SAS, SPSS and Excel.RESULTS: As a result of using this data management ap-proach, a consistent and high quality data set for analysiswas assured for the CODE-2 study.CONCLUSIONS: Internet technology, such as that used inthe CODE-2 study, should have a greater role to play in thecollection and management of data in future multi-countrycost of illness studies. *Registered Trademark of SmithKlineBeecham. **Registered Trademark of MDS.

PEN3

DEVELOPMENT OF AN ALGORITHM FOR ESTIMATING ASTHMA SEVERITY FROM AN ADMINISTRATIVE COST DATABASELeidy NK1, Paramore LC1, Watrous M2, Doyle J3, Zeiger RS4

1MEDTAP International, Bethesda, MD, USA; 2Genentech, Inc., South San Francisco, CA, USA; 3Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA; 4Kaiser Permanente, San Diego, CA, USA

OBJECTIVES: The purpose of this study was to developand test an algorithm for estimating disease severity foreconomic studies involving administrative cost databaseanalyses. The intent was to go beyond the traditionalthree-group classification (mild, moderate, severe) and de-velop a more refined four-group system consistent withNHLBI Guidelines for estimating asthma severity: mild in-termittent (MI), mild persistent (MP), moderate persistent(MoP), severe persistent (SP).METHODS: A retrospective cohort design was used, em-ploying administrative claims from a fee-for-service pro-vider of health care benefits to approximately 3.5 millionfederal employees in the U.S. Patients with asthma weredefined by: >1 medical encounters for asthma (exclusiveof COPD and allergic rhinitis) during 1994–5 and continu-ous enrollment in the health plan for 24 months; 22,833patients with asthma were identified. Four algorithmswere developed using NHLBI guidelines for pharmaco-logic intervention with adjustments for practice patternsduring 1994-5. Evaluation was based on the observed se-verity distribution in the population and results from bi-variate and multivariate analyses in random samples, ex-amining the relationship between severity level and thefollowing clinically-grounded variables: deaths, hospital-izations, ER visits, and use of an allergist/pulmonologist.RESULTS: The final algorithm was a 2-step procedurebased on b-2 agonist and oral steroid use. Severity distri-bution in the population was as follows: 69.5% MI,16.9% MP, 11.1% MoP, and 2.5% SP. The final, popula-tion-based bivariate analyses showed a logical stair-steppattern with significant relationships between severity anddeath rates, number of hospitalizations, and specialist use(p , .001). The ordinal logistic regression model was alsosignificant (p , .0001; chi-square 5 597.45, df 5 4), witha Goodman-Kruskal Gamma statistic of 0.25.CONCLUSION: Results suggest the algorithm is usefulfor classifying patients into four levels of severity using an

Abstracts 395

administrative cost database. Clinical validation of the al-gorithm is warranted.

PEN4

MEASURING THE REAL ANNUAL DIRECT COSTS OF DIABETES TYPE 2 PATIENTSIN GERMANYGoertz A1, Spannheimer A2, Reitberger U2, Liebl A3, Neiss A4

1SmithKline Beecham Pharma, Munich, Germany; 2Kendle GmbH & Co., GMI KG, Munich, Germany; 3Diabetes Center Bogenhausen, Munich, Germany; 4Institute for Medical Statistics and Epidemiology at the Technical University, Munich, Germany

OBJECTIVES: The total economic impact of type 2 diabe-tes in Germany is widely unknown and underestimatedsince its complications are often not taken into account.The pan-European cost of illness study (CODE-2: Costs ofDiabetes in Europe, Type 2) aimed at measuring the totalcost of patients with type 2 diabetes including both costsof management and complications.METHODS: For Germany, a bottom-up, prevalence-based design was used to collect data on resource utilisa-tion of 809 patients with diabetes type 2 for the year 1998.A representative random sample of 135 office-based physi-cians (GPs and specialists) provided data using patients’medical records in face-to-face interviews. For the extrapo-lation on a population level, costs were weighted accord-ing to prevalence data on complications derived from anepidemiological pre-study. Data were analysed from a so-cietal perspective.RESULTS: Total direct medical costs per patient per yearwere on average DM 7,012, ranging from DM 156 to DM137,512. The main part of the costs (61%) was inducedby hospitalisations and long-term care. Observed costs pertreatment type were DM 5,890 for diet, DM 5,621 forOAD and DM 9,799 for insulin-treated patients. Theanalysis per complication status showed a range from DM3,716 for patients without, and up to DM 13,283 for pa-tients with both micro- and macrovascular complications.Assuming a diabetes type 2 prevalence of 4.24%, the totalannual direct costs for the respective population were esti-mated to be DM 24,388 million.CONCLUSIONS: The data substantiate the enormous so-cio-economic burden and that costs are much higher thanestimated in previous studies. Treatment costs of diabetesshow a broad variance. Costs increase with age and arehighly related to occurrence of complications and onset ofinsulin treatment. Furthermore, the study highlights thatcosts are directly related to the severity of the patient’s ill-ness.

PEN5

METHODOLOGICAL ASPECTS OF MULTI-COUNTRY COST OF ILLNESS STUDIES—1: DATA COLLECTION INSTRUMENT DESIGNKirsch J, Hood SHealth Economics Dept., SmithKline Beecham Pharmaceuticals, Harlow, Essex, UK

OBJECTIVES: Underlying the success of multi-countrycost of illness studies are the methods used to identify, enu-merate and value the resources that constitute the overallcosts of the disease being studied. The design objective ofthe data collection instruments is to ensure consistency be-tween countries and retrieval of all relevant data.METHODS: The CODE-2* (Costs of Diabetes in Europetype 2) study is a bottom-up, prevalence based cost of ill-ness study conducted to evaluate the total burden of type 2diabetes in 8 European countries. The objective of thestudy was to fully identify all costs relating to the manage-ment of the disease and its associated complications in ad-dition to patient demographic and clinical parameters.Data for over 7000 patients were collected retrospectivelyover a 6-month period from a variety of sources includingpatient interviews, medical records and registries. This re-quired the design of common data collection tools thatcould be used in all eight countries by over 600 practitio-ners included in the study. Identification of core items ofresource utilisation and clinical assessments was done inconsultation with European and national advisory boardsconsisting of health economists, clinicians and healthcareresearchers. A process of forward and back translation en-sured the cultural and linguistic integrity of the eight lan-guage versions used. Layout and instrument design weredeveloped by iteration between the central study coordi-nating and country teams and assessed by mini-pilots. Fi-nal field tests ensured acceptability and suitability forstudy participants.RESULTS: The resulting country specific practitioner andpatient data collection instruments were consistent in allaspects for core resource utilisation items necessary for in-ter-country comparison of the study results.CONCLUSION: The instruments and development pro-cess used for the CODE-2 study should serve as a modelfor other multi-country cost of illness studies. *RegisteredTrademark of SmithKline Beecham

PEN6

COSTS OF TREATING NAIVE PATIENTS WITH TYPE 2 DIABETES—A COMPARISON BETWEEN THE UNITED KINGDOM AND FRANCEPashos CL, Tafesse E, Botteman MAbt Associates Clinical Trials, Cambridge, MA, USA

OBJECTIVES: To estimate for the United Kingdom (UK)and France the first year diabetes-related costs of naive pa-tients with type 2 diabetes initiated upon therapy with thesame oral hypoglycemic agent (Diamicron). In both theUK and France, responsibility for the care of patients withtype 2 diabetes rests with general practitioner physicians.METHODS: Patient records of consecutive naive patientswere collected from general practices (two in the UK, one inFrance). Diabetes-related resource utilization and costs forthe first year post-initiation of oral therapy were analyzed.This included medication, professional visits, hospital out-patient visits, emergency visits, inpatient hospitalizations,laboratory tests and procedures, and self-administered tests.

396 Abstracts

RESULTS: Healthcare utilization and concomitant costs(denominated in the Euro) for naive type 2 diabetic patientsinitiated on Diamicron varied greatly between these prac-tices in the UK and France. The biggest difference in utiliza-tion was greater use in France of hospital inpatient and hos-pital outpatient services. Total annual direct medical costsin France were approximately 1,620 Euro (10,600FF), com-pared with 454 Euro (320 BP) in the UK. Costs of Diami-cron, as actually prescribed and used by the patients, alsovaried considerably: 112 Euro in France versus 65 Euro inthe UK.CONCLUSION: Patient care and associated costs for pa-tients newly diagnosed with a common clinical conditionsuch as diabetes can vary considerably even within popu-lous western European nations. Analyses of cost of illnessand cost effectiveness must be careful to use country-spe-cific data in order to maximize the utility of these analysesby relevant decision-makers within those countries.

PEN7

DIRECT COSTS OF DIABETES IN ITALYGarattini L1, Tediosi F1, Parazzini F2, Chiaffarino F2 on behalf of Gruppo di Studio RECORD1Cesav, Centro di Economia Sanitaria, Istituto Mario Negri, Ranica, Italy; 2Unità di Epidemiologia Clinica, Istituto Mario Negri, Milano, Italy

OBJECTIVE: To offer direct cost estimates of patients withdiabetes mellitus based on retrospectively collected data.METHODS: A multicenter, retrospective observationalstudy on 30 diabetic centres located in 12 different re-gions of Italy. Each diabetes centre randomly selected be-tween 60 and 100 patients with diabetes mellitus. A totalof 2,135 patients were enrolled. For each patient data onresource consumption was collected retrospectively—onthe two months preceding the enrolment—using a ques-tionnaire filled in by physicians of the participant centres.Per patient-year direct cost estimates were produced, tak-ing the Italian National Health Service (INHS) perspec-tive, by linking per patient service use data with INHStariffs of day-hospital admissions, outpatient visits, labo-ratory and diagnostic tests, and drug prices. Drug costwas calculated by multiplying the daily dose by the pub-lic price when taken at home, and by 50% of the publicprice when taken in hospital and ambulatory setting.RESULTS: Total direct medical cost per patient was on av-erage Italian L. 1,708,364 for Type I diabetes (N 5 572;mean age 51.7) and L. 1,623,728 for Type II diabetes (N 51,563; mean age 61.4). Drugs accounted for 32% of thetotal cost in Type I patients and 17% in Type II, while out-patient care, day-hospital admissions included, accountedfor 68% and 83%. Inpatient costs were not included.CONCLUSION: This study suffers some major limits. Onlypatients, and not centres, were randomly selected; resourceuse data was retrospectively collected over two months ofobservations, and unit costs were based on INHS tariffs.However, since in Italy the cost of diabetes has received lim-ited attention, we believe this study presents some interest-ing information on the burden of diabetes in this country.

PEN8

CODE-2* UK: THE CURRENT COSTS OF TYPE-2 DIABETES IN THE UKBottomley J1, Baxter H2, Lawlor D2, Williams W2, Burns E3, Young R3, Harvey J4, Sheaves R4, Sheldon P5

1SmithKline Beecham Pharmaceuticals, Welwyn Garden City, UK; 2Nuffield Institute, Leeds, UK; 3Hope Hospital, Salford, UK; 4Health and Social Services, St Helier, Jersey, Channel Islands, UK; 5Bradford Health Authority, West Yorkshire, UK

OBJECTIVES: As part of a pan-European cost of illnessstudy in eight countries (CODE-2*: Costs of Diabetes inEurope, Type 2) the CODE-2* UK objective was to mea-sure cost of care for patients with type 2 diabetes (T2D) interms of management and complication costs in the UK.METHODS: This was a bottom-up, prevalence-basedstudy design, which collected resource use and clinical out-come data across primary and secondary care. Datasources include comprehensive population-based diabetesregisters (the ‘primary databases’) and hospital and pri-mary care records (‘secondary databases’). 756 T2D pa-tients (defined as aged 30 or above at diagnosis) were iden-tified from 3 registers. Clinical data was reviewed againsttreatment guidelines to assess status of T2D management.Preliminary results based on data in the 6-month periodJuly 1st to December 31st 1998 were analyzed first.RESULTS: Assuming a UK prevalence of T2D of 2% thetotal direct medical costs for people with T2D in the UK in1998 were estimated to be £1.83 billion or 3.4% of health-care expenditure. Average costs for patients with T2D were£1550 per year. The costs attributed to oral anti-diabeticdrugs comprised 2.7% of overall direct costs. Costs for pa-tients with T2D with microvascular and macrovascularcomplications with indicators of Insulin Resistance Syn-drome were higher than for patients without complica-tions. The proportion of patients achieving good glycemiccontrol (,6.5% HBA1c) decreased as patients progressedfrom diet and exercise through to insulin alone or in com-bination, highlighting the progressively increasing need fortreatment as duration of disease increases.CONCLUSIONS: CODE-2* UK provides evidence of di-rect costs and management of T2D in the NHS. It is a richdata set to inform Health Boards, Health Authorities andPrimary Care Groups of how costs are composed and thelikely outcomes seen for T2D management. *Trademarkof SmithKline Beecham plc.

PEN9

PHARMACODYNAMIC-PHARMACOECONOMIC (PD-PE) MODEL OF ASTHMA ANTI-INFLAMMATORY THERAPIESHandley DA, Reasner DS, Vaickus L, Sullivan A, Ellsworth KSepracor, Inc. Marlborough, MA, USA

ABSTRACT OBJECTIVE: To develop a pharmacody-namic-pharmacoeconomic model to compare effectivenessand costs of anti-inflammatory therapies used in asthmatreatment.

Abstracts 397

PARTICIPANTS WHO WOULD BENEFIT: Membersof P&T, MCO and others assessing formulary inclusionand cost-benefits of asthma therapies. Asthma has in-creased 10-fold in the last 2 decades, costing $6 B/year inthe US. The most severe asthmatics account for the majorityof the total health care costs, endpoints involving emergencydepartment treatment (z$250) and hospital admission($3100). Anti-inflammatories (steroids, leukotriene modifi-ers) are recommended for most levels of asthma severity andshould be able to alter these endpoints. However, initial pre-scription filling (50–80%) and refilling (20–40%) rates areso poor as to preclude comparisons of drug effectivenessor health care plans. While disease management of asthmawould wish to alter these endpoint costs, it is not possibleto develop a PD-PE model to compare effectiveness ofasthma anti-inflammatories across or within plans.OBJECTIVES: To develop a PD-PE model to compare ef-fectiveness of anti-inflammatory therapies. To obviate theabove limitations, we used published clinical results in-volving moderate-severe asthmatics in terms of improvedlung function ((D0.1L FEV1) calibrated against the anti-inflammatory inhaled therapy costs (per-day 3 treatmentduration). Additional cost reductions related to reducedneed for metered dose inhalers were also tabulated.RESULTS: These analyses show $US costs to achieve a0.1L improvement in moderate-severe asthmatics usingsteroids such as budesonide ($232) or beclomethasone($110), leukotriene modifiers (zileuton $68 or montelukast$101), as well as the single isomer b-agonist levalbuterol($39). Chronic treatment with racemic albuterol did notimprove lung function but rather resulted in a decrease inlung function. Each final cost is inclusive of reduced al-buterol metered dose inhaler costs.CONCLUSIONS: This cross-stratified PD-PE model as-sessed anti-inflammatory therapy impact on improvementsin asthma lung function and reduced need for rescue MDI.As these results are derived from well-controlled clinicalstudies, they reflect optimized treatment compliance andtherefore maximal PD-PE effect. Accordingly, these rela-tive scores may provide insights relating to formulary in-clusion of drugs which face class competition.

ECONOMIC AND OUTCOMES ISSUES OF GASTROINTESTINAL DISORDERS

PGI1

EFFECTIVENESS OF OMEPRAZOLE, AMOXICILLIN, AND CLARITHROMYCIN IN ERADICATING HELICOBACTER PYLORI IN PATIENTS ON CHRONIC ACID SUPPRESSION FOR PEPTIC ULCER DISEASE ORULCER-LIKE SYMPTOMSAllison J, Hiatt R, Levin TR, Lieu T, Ackerson L, Hurley L, Libran DKaiser Permanente Division of Research, Oakland, CA, USA

OBJECTIVE: To evaluate the effectiveness of the omepra-zole, amoxicillin, and clarithromycin (OAC) regimen in

eradicating H. pylori (HP) in patients on long term acidsuppression medicine for documented ulcer or ulcer likesymptoms.METHODS: Kaiser Permanente data bases were accessedto identify all patients receiving acid suppression therapyfor at least three 30-day periods within the most recent 12-month time frame. Those identified were then matched todatabases that capture physician recorded diagnoses ateach patient visit. Patients not previously treated for HPand with documented (by endoscopy or UGI) diagnoses ofduodenal ulcer (DU), gastric ulcer (GU), gastritis, and un-specified peptic ulcer disease (PUD) or ulcer like symptomswere selected for recruitment. Study participants were ran-domly assigned to C14 urea breath testing (UBT) and HPeradication, if positive, versus standard maintenance acidsuppressive therapy. All patients positive on UBT (exceptthose allergic to penicillin N 5 12) were treated with ome-prazole 20 mg, amoxicillin 1gm, clarithromycin 500 mgbid for 10 days. Proof of eradication was documented byUBT administered 4 weeks post treatment. Each patienttesting negative after treatment was taken off acid reduc-ing medication and followed for up to one year.RESULTS: Ninety-five of 96 patients completed 10 daysof OAC, and 89 have been retested 4 weeks later. Of thoseretested, 85% (95% CI 5 76%, 92%) were breath testnegative and 15% (95% CI 5 8%, 24%) were breath testpositive. Fourteen participants reported recurrent symp-toms following HP eradication. Side effects of patients onthe regimen were reported by 6 out of 96 treated (6%),and included bloody diarrhea (1), rash (4), and abdominalcramps (2).CONCLUSION: OAC is an effective regimen for eradicat-ing HP in infected patients on chronic acid suppression forPUD or ulcer like symptoms. Compliance with the regi-men is high and serious adverse reactions are low.

PGI2

DIAGNOSIS AND TREATMENT OF CONSTIPATION IN A MANAGEDCARE POPULATIONBramley T1, McLaughlin T2, Margraf T3, Okamoto L2

1University of South Carolina, Columbia, SC, USA; 2NDC Health Information Services, Phoenix, AZ, USA; 3Pharmetrics, Inc., Boston, MA, USA

PURPOSE: To assess the cost of care associated with thetreatment of constipation subjects.METHODS: All subjects contained within PharMetrics In-tegrated Outcomes database possessing an ICD-9-CM 5564.0 between January 1, 1996 and March 31, 1998 wereeligible for study inclusion. Subjects were required to havedata available for analysis 6 months preceding and follow-ing the date of their constipation diagnosis. Charge datawere evaluated over the 6-month intervals surrounding theindex date. Additionally, monthly charges were plottedover the entire 12-month interval to allow examination oftrends associated with the diagnosis and treatment of con-

398 Abstracts

stipation. Comorbidities and drug utilization related toconstipation were also assessed.RESULTS: 9,301 patients met all inclusion criteria. Themean age was 39.74 years and 63.41% of the sample wasfemale. The most commonly observed comorbidities werehypertension (16.3%), diabetes (7.5%), diverticulitis (7.4%)and irritable colon (5.9%). Approximately 8% of the sam-ple received medications thought to have constipating ef-fects in the 6-month period prior to their constipationdiagnosis. Prescription laxative use was observed in ap-proximately 3% of subjects after the constipation diagno-sis. Average costs rose from $277 (SD 5 1717) 6 monthsprior to the constipation diagnosis to $1031 (SD 5 4402)during the month of the diagnosis. Similarly, average costsdecreased from $890 (SD 5 3852) in the month directlyfollowing the index date to $479 (SD 5 3470) in the sixthmonth following the index date.CONCLUSION: Costs associated with the diagnosis andtreatment of constipation patients appear to be concen-trated around the index visit. Early detection and resolu-tion of constipation could result in significant cost savingsfor managed care organizations.

PGI3

DIFFERENTIAL COST OF USING NSAIDS IN A MEDICAID POPULATIONDickson WM, Reeder CECollege of Pharmacy, University of South Carolina, Columbia, SC, USA

OBJECTIVES: To estimate the differential cost of usingnon-steroidal anti-inflammatory drugs (NSAIDs) in aMedicaid population. NSAID use is associated with ad-verse gastrointestinal (GI) events such as GI distress, symp-tomatic ulcer, and anemia, which vary in severity and mayrequire symptomatic treatment or hospitalization. Theeconomic burden of these events is represented by the dif-ferential cost of treating NSAID-related toxicities.METHODS: Treatment cost and adverse GI event rateswere compared for two cohorts (NSAID and non-NSAIDusers) selected retrospectively from the South CarolinaMedicaid population. To be included in the analysis, allsubjects must have been at least 18 years old and continu-ously eligible for 12 months following an index event in1995. For NSAID users, the index event was at least twoNSAID prescriptions or a 60 day supply (n 5 21,278). Fornon-NSAID users (comparator group), the index eventwas at least one prescription during 1995 (n 5 152,072).Average cost per subject by type of service (physician,other ambulatory, inpatient hospital, outpatient hospital,emergency department, prescription drugs) was comparedfor the cohorts. Regression analysis was used to evaluatethe effect of NSAID use on treatment cost, controlling forage, gender, prior exposure to NSAIDs, use of gastro-pro-tective agents, and adverse GI events.RESULTS: NSAID users had higher physician ($1885 vs.$1488), other ambulatory ($5785 vs. $5060), inpatient

hospital ($5966 vs. $3326) and prescription drug ($1012vs. $738) costs than the non-NSAID group. Regressionanalysis showed that NSAID use was a significant explana-tory variable (p , 0.0001), as were age, gender, race, andadverse GI events.CONCLUSIONS: NSAID use is associated with higher av-erage treatment costs and is a significant predictor of mod-eled treatment costs after controlling for other effects. Sub-stantial cost savings may be realized if NSAID-related GItoxicities and adverse events can be managed or avoided.

PGI4

COST OF ITALIAN DYSPEPTIC PATIENTS: A FEASIBILITY STUDY FROM THEDYSPEPSIA PROJECTTriossi O1, Tampieri I1, Casetti T1, Degli Esposti E2, Buda S3, Cuttin S3

1Department of Gastroenterology and Digestive Endoscopy, S.M. Croci Hospital, Ravenna, Italy; 2Ravenna Health Care, Revenna, Italy; 3Clicon Srl, Data Warehouse Company, Ravenna, Italy

The Dyspepsia Project is a General Practitioners’ (GPs)project, aiming to establish an epidemiological database inorder to describe and follow up dyspeptic patients in theRavenna area.OBJECTIVE: To investigate the feasibility of a cost of ill-ness study on the dyspeptic patient in terms of healthcareresource use, and to produce an average cost per patient.METHOD: Retrospective cost of illness analysis based onrecords of 106 dyspeptic patients followed by 10 GPs.This preliminary analysis has been performed in the per-spective of Health System. Patients analyzed were enrolledbetween September and December 1999 and followed upat least for 6 months. Hospital admissions; GP, specialistand emergency room visits; instrumental and laboratorytests; drugs were analyzed; each cost variable was valuedin Italian Liras 1999 (1800 ITL 5 1 US$) using publishedregional or national tariffs and marked prices for drugs.Patients have been divided into Group A (37 5 35%) andGroup B (69 5 65%), according to whether they under-went or not gastroscopy test.RESULTS: The average cost per patient was ITL1,121,654 in A Group and ITL 624,565 in B Group. Di-rect costs accounted for 76.5% of the total value in AGroup and for 70.9% in B Group, while the remaining23.5% in A Group and 29.1% in B Group was due to in-direct costs (i.e., productivity losses). In Group A the ma-jor cost driver was drugs (40.8%) (13.6% was the share ofantidyspeptic drugs) followed by tests (18.4%), visits(15.6%) and emergency room visits (1.7%); in B Groupthe major cost driver was drugs (38.6%) (10.1% was theshare of antidyspeptic drugs) followed by visits (19.5%),tests (9.2%) and emergency room visits (3.6%).CONCLUSIONS: Collection of cost data at General Prac-titioner’s level is very effective as it allows a precise and ap-propriate analysis.

Abstracts 399

PGI5

COST OF ILLNESS STUDIES MAY LARGELY UNDERESTIMATE THE COST OF DISEASES WITH UNMET MEDICAL NEED SUCH AS IRRITABLE BOWEL SYNDROME (IBS)Volmer T1, Gehlert I2, Behrens M1

1Health Economics, Glaxo Wellcome GmbH&Co, Hamburg, Germany; 2University of Applied Science, Hamburg, Germany

OBJECTIVES: The usefulness of cost of illness studies iscontroversial. Nevertheless, they are carried out frequentlyand at least raise awareness and draw attention to any eco-nomic impact a disease may have on a particular healthcare system and society. The aim of this analysis was toquantify the cost of a disease with a high unmet medicalneed (Irritable Bowel Syndrome, IBS, ICD9 564.1), usingofficial German health statistics. Although IBS is reportedto have a prevalence of up to 20%, current evidence sug-gests, that existing statistics based on the ICD classifica-tion misrepresent the true economic burden.METHODS: We used 1997 data from various availablesources of health information and identified direct and in-direct costs using a “top-down approach” employing se-lected ICD9 codes (564.1, 564.9), which were commonlyused to document patients with IBS.RESULTS: The total calculated burden to German societywas 53.5 million Euro (?)

CONCLUSIONS: Medical conditions with unmet needsuch as IBS are not well represented in official statistics.The number of identified patients, the documented re-source use and the total cost of illness calculated using in-formation based on ICD9 coding might largely misrepre-sent the burden of these diseases. Therefore, representativeprospective studies are needed to document such areas ofunmet medical need.

ECONOMIC AND OUTCOMES ISSUES OF WOMEN’S AND MEN’S HEALTH DISORDERS

PMW1

COSTS OF TREATMENT WITH DOXAZOSIN VERSUS TAMSULOSIN IN MEN WITH BENIGN PROSTATIC HYPERPLASIA

Thompson D, Edelsberg J, Pozniak A, Oster GPolicy Analysis, Inc. (PAI), Brookline, MA, USA

OBJECTIVES: To compare the costs of treatment withdoxazosin mesylate versus tamsulosin hydrochloride inmen with benign prostatic hyperplasia (BPH).METHODS: We used claims data (spanning August 1996through December 1998) from a large health insurer in theNew England region of the United States to examine dif-ferences in the utilization and costs of care over a 90-dayperiod among men with BPH who received doxazosin ver-sus tamsulosin. Study measures included all BPH-relatedprescription drug therapy, office-based physician services,tests and procedures, and hospital inpatient and outpatientservices.RESULTS: Subjects who received doxazosin (N 5 394)and tamsulosin (N 5 167) were generally similar in termsof age, comorbidity status, and pretreatment health-carecosts. Subjects in the doxazosin group had fewer physicianvisits during the 90-day period of follow-up (18 fewer per100 patients; 95% CI: 6, 30) than those in the tamsulosingroup (28 vs 46 per 100). The mean cost of BPH drugtherapy was $11 lower (95% CI: $5, $17) among patientswho received doxazosin ($71 vs $82 for tamsulosin), andmean costs of follow-up physician visits were $16 lower(95% CI: $4, $27) ($18 vs $34 respectively). Althoughthere were no significant differences in the utilization orcosts of diagnostic tests, surgical procedures, or hospitalservices, nominal differences in these measures favoreddoxazosin. The mean total cost of BPH treatment over 90days was $173 (95% CI: $92, $253) in the doxazosingroup and $247 (95% CI: $124, $370) in the tamsulosingroup, yielding a difference of $74 (95% CI: 2$73, $221).CONCLUSIONS: Men with BPH who received doxazosinversus tamsulosin had lower costs of BPH-related prescrip-tion drug therapy and follow-up physician services, andnominally lower overall costs of BPH treatment.

PMW2

THE VALUATION OF ERECTILE DYSFUNCTION: CONVERTING CLINICAL OUTCOMES INTO UTILITIESStolk EA, Busschbach JJVInstitute for Medical Technology Assessment, Erasmus University, Rotterdam, The Netherlands

OBJECTIVES: We converted disease specific outcomes oftreatment for erectile dysfunction (ED) using sildenafil (Vi-agra) into Quality Adjusted Life Years (QALYs). Due tosize and complexity of most disease specific question-naires, application of this (recommended) strategy is scarceand detailed reports about valuation exercises of diseasespecific health states are lacking. Such valuation studiesare needed to adapt existing methods to this new strategy.METHODS: 200 subjects from the general public and110 ED-patients valued 25 ED-states with Time Trade-Offto convert the outcomes of a clinical trial into utilities. Thehealth states reflect different gradations of ED, as categor-

Cost category ICD9 564.1 ICD9 564.9

Number of identified patients 810 000 102 000

1. Ambulatory services [mill. ∈ ] 22.5 3.12. Drugs [mill. ∈ ] 11.8 6.33. Hospital [mill. ∈ ] 7.0 3.94. Other direct [mill. ∈ ] 0.3 0.15. Indirect costs [mill. ∈ ] 11.9 1.56. Total costs [mill. ∈ ] 53.5 14.9

Mean cost per patient (direct only) [∈ ] 51.4 13.1

400 Abstracts

ised by questions 3 and 4 of the International Index ofErectile Function (IIEF). Those questions were suitable forour purposes because they describe the primary endpointsof ED-treatment, as defined by the National Institutes ofHealth. To test content validity and feasibility of valuingED-states, we explored the relation between the values andbackground variables and analysed the factor structure ofthe valuation space in patients and the general public.RESULTS: The factor structure of the valuation space wasthe same in patients and the general public and valueswere not influenced by background variables, supportingvalidity and feasibility of valuing disease specific out-comes. According to the general public the mean improve-ment on the IIEF equals an increase of utility from 0.76 to0.87 on a scale of 0 (dead) to 1.0 (perfect health).CONCLUSIONS: Treatment preferences can be measuredby attaching utility weights to disease specific outcomes.The implication is that in cost-effectiveness analysis resultscan be available quickly, even if only disease specific out-comes are available from previous research. On basis ofthe utility gain of 0.11, we expect the costs per QALY ofsildenafil to be about US $10,000.

PMW3

ECONOMICAL ANALYSIS OF THE RESULTS OF PROSTATE CANCER SCREENING PROGRAMKomarova LE1, Matveev BP1, Buharkin BV1, Kobina S2

1N.N. Blohin Science Oncology Center, Russian Academy of Medical Science, Moscow, Russia; 2Rhône-Poulenc Rorer, Moscow, Russia

Early prostate cancer (PC) diagnosis as a result of screen-ing may raise total survival index. Diagnostic level of PCin screening population is 2.5%; diagnostic level of PC forthose, who asked independently for urological consulta-tions is 1.5%. All these were the bases for starting the PCscreening program in the Oncology Science Center (OSC),Russian Academy of Medical Science in 1996. Since thebeginning of the program 852 persons aged 50–80, whopreviously were not diagnosed with PC, participated inscreening. Methods of diagnosis in present work weremanual rectal examination (MRE), transrectal ultrasound(TRUS), and immunology examination—determinationof prostatic specific antigene (DPSA). Prostate punctionalbiopsy with morphologic diagnostic was used to amplifythe results. We also evaluated the diagnostic parameters ofthe methods used and estimated the cost-effectiveness ofdiagnosis. In the first part of the research diagnostic costsof screening a group were compared with that of the inde-pendent patients of the OSC: 5000 people during 2-yearperiod. Of the latest 4700 persons were subjected to MRE,1560 have passed TRUS, and 315 have passed DPSA. To-tal diagnostic cost for screening group (2 rounds) wasequal to 1,120,655 RUB, while the same for the controlgroup of individual patients was equal to 3,331,400 RUB.The difference of 2,210,745 RUB between the diagnosticcost for the control group and the cost for screening groupwas primarily due to the cost of MRE for the control

group. Thus, the first part of the research has shown thatthe screening is more cost-effective for the group of theadult aged 50–80. A decision tree model was figured out inthe second part of the research. It included data from theliterature, OSC’s clinical diagnostic results. Cost expecta-tions were counted for each treatment strategy to estimateeconomic effectiveness. This was done for each branch bymultiplying the cost and probability. Cost expectations fortreatment strategies for different PC stages were then com-pared. The analysis of the decision tree mode has shownthat the average treatment cost in the screening group wasequal to 13,666 RUB per patient, taking into account theprobability of finding disease and various types of treatingincluding Taxotere. The same average for the controlgroup of independent patients was 17,082 RUB per pa-tient. The significant difference in average costs in twogroups (3,416 RUB) is due to the larger weight of the pa-tients with the later stages of the disease in control group,who needed more expensive treatment courses.

PMW4

A MICRO-COST ANALYSIS OF URINARY INCONTINENCE CARE IN LONG-TERMCARE FACILITIESHartzema AG1, Shih YCT2, Tolleson-Rinehart S3, Goldfarb SD4, Gorospe J51Center for Pharmaceutical Outcomes Research, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA; 2University of North Carolina School of Pharmacy, Division of Pharmaceutical Policy & Evaluative Sciences, Chapel Hill, NC, USA; 3University of North Carolina at Chapel Hill, Program on Health Outcomes, Chapel Hill, NC, USA; 4Pharmacia & Upjohn, Duluth, GA, USA; 5University of North Carolina at Chapel Hill Graduate School, School of Nursing, Chapel Hill, NC, USA

OBJECTIVE: Urinary incontinence (UI) is a conditionwith high prevalence and the second most important rea-son for admission to long-term care. The purpose of thisstudy is to use a micro-costing approach to estimate the in-cremental costs of caring for the UI patients residing inthese facilities. The three levels of long-term care consid-ered are skilled and intermediate nursing, and rest homes.METHODS: Data collection took place in three phases.Phase I data from a time and motion (TAM) study, is usedto develop a taxonomy of different care activities per-formed for the incontinent patients and a correspondingcontrol group. In Phase II, a patient-level observationalstudy assessed the frequency of the caring activities ob-served in Phase I. Caring activities of 130 patients (65 inthe case and control groups, respectively) from six longterm facilities are recorded at each shift for a week, pro-ducing a total of 2,730 observations. In Phase III, investi-gators distributed a facility-level survey to 49 facilities, in-cluding all phase I and Phase II facilities. The surveycollected information on hourly wage for different types oflabor, unit costs of supplies, and laundry services. Phase IIIsurvey data are combined with patient-level data from

Abstracts 401

phase II in fixed effect regression models to estimate the ef-fect of incontinence on caring costs for UI.RESULTS: Compared with long-term care residents with-out UI, those with occasional UI consumed $2.90 moreper shift on labor and supply costs, while those with fre-quent UI consumed $5.76 more. Incremental laundry costsper shift are $0.34. Our study suggests that, on average,the increased cost of caring for a UI patient in a long termcare facility is $5,650 per year, after controlling for factorssuch as patient demographics, insurance status, type oflong-term care, and shift.CONCLUSIONS: Behavioral, pharmacological, and medi-cal treatments that successfully control or alleviate UI condi-tions may yield substantial savings in long-term care set-tings.

ECONOMIC AND OUTCOMES ISSUES OF INFECTIOUS DISEASES

PID1

ESTIMATES OF THE INDIRECT COSTS OF HIV AND AIDS IN THE UNITED KINGDOMMullins CD1, Whitelaw G2, Beck E3, Cooke JL1

1University of Maryland, Baltimore, MD, USA; 2SMT—Center for Research, Shrewsbury, NJ, USA; 3Imperial College School of Medicine, London, UK

OBJECTIVES: To estimate the indirect costs of HIV andAIDS for individuals in the United Kingdom based on uti-lization of informal (non-medical) care and communityservices; to compare these figures with direct costs; and todemonstrate the multitude of budgets that are affected byindirect costs.METHODS: A prevalence-based model was used to deter-mine the aggregate utilization by stage of progression (as-ymptomatic HIV, symptomatic non-AIDS, AIDS). Sourcesof data included SOPHID97 for prevalence figures; pub-lished data by Petrou, et al. on service utilization and costs;published figures from Lubeck and Fries and data from thePositive Lives Employment Survey 1997 for disability; andthe UK Office of National Statistics Family ExpenditureSurvey 1997–98 for disability and income data.RESULTS: Annual indirect costs (converted to 1998 Britishpounds) are £3,881.14 for asymptomatic HIV patients;£7,865.17 for symptomatic non-AIDS patients; and£21,695.13 for AIDS patients. These figures can be com-pared to annual hospital direct cost figures published inPharmacoEconomics in 1998, which were £2,391 for as-ymptomatic HIV patients; £4,562 for symptomatic non-AIDS patients; and £15,895 for AIDS patients. Thus, ourestimates of indirect costs are higher than the hospital directcost estimates for all three groups of HIV/AIDS patients.CONCLUSIONS: The difference (in 1998 British pounds)in annual indirect costs per patient between asymptomaticand symptomatic non-AIDS patients is £3,984.03. The dif-ference in annual indirect costs per patient between symp-tomatic non-AIDS and AIDS patients is £13,829.96. Thesedifferences demonstrate the potential for significant savings

if disease progression can be prevented or delayed. Focus onindirect costs of HIV/AIDS demonstrates the complexity ofdecision making when data is limited to direct medical costsand raises an issue regarding who should have input into de-cision making when multiple budgets are affected.

PID2

ASSESSING THE DIRECT COST OFLYME DISEASEGoodman M, Nordin J, Maciosek M, Fillbrandt KHealthPartners Research Foundation, Bloomington, MN, USA

OBJECTIVES: Lyme disease is an increasingly commoncondition in the United States and temperate regions ofEurope and Asia. The reported cases have increased 25fold since 1982. Recently two vaccines have been devel-oped. Decisions about the cost effectiveness require consis-tent cost estimates. The objective of this study is to identifythe costs of diagnosis, treatment, treatment failure, andlate diagnosis in a large HMO population located in an en-demic area. Prior studies have not included the worriedwell who are tested but do not receive treatment.METHODS: This study is a matched retrospective cohortstudy. 8,017 individuals tested or diagnosed with LD wasmatched by age, gender, and zip code with a control popu-lation. Standardized HMO and pharmacy costs were gath-ered from the HMO claims system. Chronic disease statuswas controlled using the Chronic Disease Score. Costswere assessed for six weeks prior to first LD contact andsix weeks following first LD contact. Late-effect and treat-ment failures were assessed separately for longer post-con-tact windows. RESULTS: The mean 12 week per capita medical directexpense for LD cases is 4.1 times higher than for the con-trol group. The cost difference is highly significant (p ,0.001). Pharmacy expense is 1.77 times greater (p ,0.001).CONCLUSIONS: The cost effectiveness of LD vaccines isdependent on the identifying the cost of both treatmentand anxiety caused by the potential of tick-borne infec-tion. The costs differences illustrated in this study empha-size the importance of both true positives and the “wor-ried-well” in assessing the total cost of LD to healthinsurance sponsors.

PID3

HOSPITALIZATION RATES AND ASSOCIATED COSTS FOR INFLUENZA IN A MANAGEDCARE POPULATIONEaddy M1, Cox F2, Okamoto L3

1University of South Carolina, Columbia, SC, USA; 2Glaxo Wellcome, Research Triangle Park, NC, USA; 3NDC Health Information Services, Phoenix, AZ, USA

OBJECTIVES: The objective of this study was to measureinfluenza-related hospitalization rates and cost in a man-aged care population.

402 Abstracts

METHODS: All subjects possessing an influenza diagno-sis (ICD-9-CM 5 487) in PharMetrics’ Integrated Out-comes Database between January 1, 1997 and June 30,1998 were eligible for inclusion. Subjects with an influenzadiagnosis one-year prior to their index date and subjectswith less than 3 months of data available for analysis priorto or following first diagnosis were excluded. Comorbidi-ties, influenza vaccinations, antiviral and antibiotic pre-scriptions were also identified during the study period.RESULTS: A total of 18,000 subjects met all inclusioncriteria. The mean age of all subjects was 29.1 years (StdDev 5 20.7). Inpatient costs accounted for over 35% ofthe $12.8 million in total costs incurred during the studyperiod, even though only 3% (472) of the sample requiredhospitalization for influenza. The mean cost of an influ-enza related hospitalization was $9,667 (Std Dev 520,940) per subject hospitalized. Antibiotics were pre-scribed in approximately 28% of the subjects, however an-tivirals were prescribed to only (7%) of the subjects.CONCLUSIONS: Although hospitalization for influenzawas rare (3%), the costs associated with this outcomewere significant. Prevention of hospitalizations related toinfluenza could result in significant cost savings for man-aged care organizations.

PID4

COST-EFFECTIVENESS OF PARTNER PHARMACOTHERAPY IN SCREENING FOR CHLAMYDIA TRACHOMATISPostma MJ1, Welte R2, Van den Hoek JAR3, Countinbo RA3, Jager JC2

1Groningen University Institute for Drug Exploration (GUIDE), Groningen, The Netherlands; 2National Institute for Public Health and the Environment, Bilthoven, The Netherlands; 3Municipal Health Service, Amsterdam, The Netherlands

OBJECTIVE: To assess the cost-effectiveness of pharma-cotherapy for partners in a screening program for asymp-tomatic infection with Chlamydia trachomatis (CT) inwomen.METHODS: A decision-analysis model was constructedfor the health outcomes of a CT screening program, such asaverted cases of pelvic inflammatory disease and infertility(major outcomes). Prevalence estimates of CT-infectionand information on partner referral were derived from pi-lot studies in Amsterdam (the Netherlands). Reinfectiondue to failed partner referral or non-compliance with part-ner pharmacotherapy was included in the model. Cost-effectiveness from a societal perspective was estimated for ascreening program in General Practices in Amsterdam, di-rected at women aged 15–29. For diagnosis of asymptom-atic CT infection a Ligase Chain Reaction test on urine wasused, for pharmacotherapy of women and partners azithro-mycin was used. By linking health outcomes with health-care costs and productivity losses averted costs were esti-mated. Cost-effectiveness was expressed as net costs permajor outcome averted and was calculated in the referencecase and in sensitivity analysis.

RESULTS: Partner pharmacotherapy induces reductionsof approximately 50% in net costs per major outcomeaverted of the screening program. Within the broaderframework of the screening program partner pharmaco-therapy is a cost-saving activity.CONCLUSIONS: Inclusion of partner pharmacotherapyprovides significant improvements in overall cost-effective-ness of the CT-screening program. Partner pharmacother-apy lowers net costs per major outcome averted to realmswhere implementation of the screening program for womenaged 15-29 should be considered. Implementation of theprogram is currently discussed for the city of Amsterdam.Given the cost-saving potentials, male partner pharmaco-therapy should be pursued within broader framework of aCT-screening program for women.

PID5

RETROSPECTIVE STUDY COMPARINGCOST-EFFECTIVENESS OF NEWERAMPHOTERICIN B PREPARATIONSTsakalou A1, Mamzoridou K2, Doulianaki E1, Pirpasopoulos M1

1AHEPA University Hospital, Thessaloniki, Greece; 2 University of Thessaly, Thessaloniki, Greece

The economic impact of administration of the extremelyexpensive new amphotericin B (Amph B) preparations ishuge for Greek hospitals.OBJECTIVE: The purpose of this study is to present thecost of amphotericin B therapy and the cost-efficacy results,after administration, of the three amphotericin prepara-tions, available in the Greek market, in a third-grade univer-sity hospital.METHOD: 88 amphotericin prescriptions dispensed bythe Hospital Pharmacy were monitored over 2 months.RESULTS:

1. The three preparations available in the hospital Phar-macy are: A. Classic Amph B (Fungisone); B. Liposo-mal Amph B (Am Bisome); C. Lipid-complexed AmphB (ABLC). These are offered at the following prices (inUSD): A: 100 mg 5 1D, B: 100 mg 5 336D, C: 100mg 5 106D

2. Duration of Amph B therapy: 18 6 3 days3. Number of flacons per patient per day: Adults 5–12 fl;

Children: 5 6 14. Departments of major administration: Intensive Care

Units (Surgical-AIDS-Internal Medicine), Nephrology,Pediatrics, Unit of Peritoneal Clearance

5. Cost of therapy: a. Costs per day, per patient A: 3D, B:1680–3360D, C: 530–1060D. b. Costs of Total Ther-apy per patient: A 30D, B 54.000 6 4.000 D. C.:14.400 6 2.600 D

6. No clinical evidence has been provided yet, whether ad-ministration of liposomal or lipid-complexed Amph B iscost-effective and the optimum necessary dosage.

CONCLUSION: Extremely high costs for Amph B ther-apy become a threat for Social Insurance, because hospi-

Abstracts 403

tal fees will necessarily increase tremendously; proven ef-ficacy for these costs has not been provided in Greece yet.

PID6

TREATMENT OF ACUTE OTITIS MEDIA IN A MANAGED CARE POPULATIONMcLaughlin T1, Margraf T2, Okamoto L1

1NDC Health Information Services, Phoenix, AZ, USA; 2PharMetrics, Inc, Boston, MA, USA

OBJECTIVES: To assess treatment patterns and treat-ment-related costs of acute otitis media (AOM) in a Man-aged Care population.METHODS: All subjects diagnosed with AOM in Phar-Metrics’ Integrated Outcomes Database between January1, 1997 and December 31, 1998 were eligible for inclu-sion. Subjects were classified as: AOM with major surgery,AOM with minor surgery, or AOM with no surgery. Sub-ject demographics, AOM—related utilization and costs,and antibiotic prescriptions were measured during thestudy period.RESULTS: 322,626 subjects were included in the study,with 97% falling into the AOM with no surgery category.Approximately 50% of the subjects were 6 years of age orolder, while 52% of the subjects were female. The totalcost of treating AOM for these subjects was $76 million.Minor and major surgery for AOM was more prevalent inchildren aged 6 or older than any other age group, andwas associated with substantially higher costs. In 1997,the mean cost for subjects undergoing major surgery was$4,114 (std dev 5 3,290) compared to $142 (std dev 5335) for subjects not requiring surgery. The most com-monly prescribed antibiotic overall was amoxicillin, fol-lowed by sulfamethoxazole/trimethoprim; however, pre-scribing patterns differed across AOM categories and agegroups.CONCLUSIONS: Acute Otitis Media is a prevalent con-dition associated with significant medical costs.

USE OF PHARMACOECONOMIC AND OUTCOMES RESEARCH IN HEALTHCARE DECISION-MAKING AND PRODUCT DEVELOPMENT

PTH1

USE OF ADMINISTRATIVE DATA FOR POST-MARKET SURVEILLANCE OF THE OUTCOME OF TOTAL HIP AND KNEE REPLACEMENTSSilverman B1, MacDonald T2 Davey P2

1US Food and Drug Administration, Rockville, MD, USA; 2Medicines Monitoring Unit, Ninewells Hospital, Dundee, Scotland

OBJECTIVES: To use routine data to study outcomes ofall knee and hip joint replacements in an unselected popu-lation with universal coverage of healthcare.

METHODS: The study population was all resident in Tay-side, Scotland from 1989–1995. Joint replacements, andsubsequent hospitalizations were identified from a hospitaldischarge database. Death was recorded from a communityhealth database. The electronic records were validatedagainst primary case records for a random sample.RESULTS: We identified 1,649 people who had first pri-mary hip replacement and 989 first primary knee replace-ments. Case note validation was performed for 274 pa-tients, of whom 143 had no previous record of hip or kneesurgery and 131 had previous surgery. The proportion ofvalid cases was higher in the patients with no previous sur-gery (98% vs 57%). The most common mis-classificationwas revision arthroplasty as primary arthroplasty becauseof incomplete recording of previous joint replacement.There were 353 deaths in the hip implant cohort (21.4%)and 187 deaths in the knee implant cohort (18.9%); 6.4%of hip recipients and 5.5% of knee recipients died within thefirst year of implant. Four percent of hip replacements werefor management of hip fracture and they had significantlyreduced survival compared to the remaining patients (deathsat 60 days, 1.6% vs. 0.8%). Significantly more knee recipi-ents than hip recipients required revision of the joint re-placement (4.4% vs 2.4%, p 5 0.007). Although time tohip implant revision varied with the year of first implant(with 1990 recipients showing the shortest time to revision),this effect did not occur among knee implant recipients.CONCLUSIONS: This study demonstrates the suitabilityof the Scottish Morbidity record for postmarket surveil-lance of medical device implants. The electronic data comefrom a stable population with universal medical coverageand outcomes can be validated by review of primary caserecords.

PTH2

MALIGNANT RETROPERITONEAL TUMOURS OF CHILDREN: FIRST RESEARCH OF QUALITY OF LIFE OF 278 SURVIVED PATIENTSZaeva G, Moiseenko EScientific Research Institute of Oncology and Gematology, Moscow, Russia

OBJECTIVES: Research on quality of life (QoL) of pa-tients cured in childhood from malignant tumors. Wide us-age of the latest anti-tumor methods of treatment has ledto considerable growth of the number of recovered pa-tients during the last years. Only in Moscow between1978–97 2255 children with oncological diseases weresuccessfully cured, 278 (12,3%) who had retroperitonealmalignant tumors. Among these children 191 were curedfrom nefroblastoma, 66 – from neuroblastoma, 21 – fromadrenal tumors. Antitumor treatment deeply affects thechild’s organism, causing not only physical dysfunction,but also dramatic psychological, emotional and socialchanges in the patient’s life. QoL research of the peoplewho suffered from malignant tumors in childhood is a newarea in Russian children’s oncology.

404 Abstracts

METHODS: Special clinical methods were used for healthcondition evaluation. Our first research of some of theQoL factors was held in 1996–97.RESULTS: It was conducted on 561 patients with differ-ent oncological diseases, including retroperitoneal malig-nant diseases. The period of remission for 59% of the pa-tients was 5–10 years, for the rest 41%—5 years. Ages ofthe patients varied from 3–7 years to 15 years and over.Their physical condition was studied (including cardiovas-cular, urinary, immunological, reproductive, muscular andother systems), as well as their psychological conditionand social adaptation.CONCLUSIONS: Only 40% of the patients don’t havesignificant after-effect from their anti-tumor therapy. Dys-function of the immune system was found in 78% of theexamined patients, of the alimentary canal, 67%, of thecentral nervous system, 60%, of the muscle and bone andthe urinary tract systems, 56% and 50% respectively. To-gether with psychological and social problems this consid-erably influences the QoL of such patients. The researchwas continued by the evaluation of life conditions of 278patients cured from malignant retroperitoneal tumors, aswell as by the analysis of the obtained data in accordancewith the international life-level standards. For this purposewe are currently adopting the selected interrogatory sheetsEQ-50 (for adult patients) and SP-36 for children. The re-sults will be introduced in future publications.

PTH3

GUIDELINES FOR SUBMISSION OF REQUESTS FOR INCLUSION OF A PHARMACEUTICAL PRODUCT IN THE ISRAELINATIONAL FORMULARYShani S, Morginstin T, Shemer JMinistry of Health, Jerusalem, Israel

The Israeli National Health Insurance Law enacted in1995 determines a list of health services including a list ofreimbursed drugs (NLRD). Accordingly, each citizen isentitled to receive health services of established qualitywithin the framework of the available resources. In thelast decade, significant advances in R&D have resulted inthe emergence of numerous expensive drugs in the phar-maceutical market. Hence, the Pharmacoeconomics Unitat the Pharmaceutical Administration of the Ministry ofHealth established a mechanism for updating the NLRD,based on the evaluation of clinical, epidemiological andeconomic data. Prior to the presentation of guidelines forthe pharmaceutical industry on submission of economicdata as part of a request for the inclusion of a pharma-ceutical product in the NLRD, a survey was conducted toexamine existing pharmacoeconomic knowledge of theIsraeli pharmaceutical industry. The results of the surveyindicated that the pharmaceutical industry has very littlepharmacoeconomic knowledge. These results prompted alenient approach to the submission of requests for thenext two years during which the industry is expected todevelop a firm pharmacoeconomic basis. Guidelines on

submission of a request to include a pharmaceuticalproduct in the NLRD were presented to the pharmaceuti-cal industry. The dossier to be submitted by the industryconsists of three sections:

Clinical—treatment regimens, adverse effects, effi-cacy and safety in comparison to alternate therapies.

Epidemiological—morbidity figures, definition of tar-get population, sales figures.

Economic—cost of the suggested treatment comparedto that of alternative therapies and an economic evalu-ation.

260 different requests have been submitted for the 2000update of the NLRD, according to the guidelines, and arecurrently being processed. Following the 2000 update, areassessment will take place, and the guidelines will bechanged according to the experience incurred by both theindustry and the Ministry of Health.

PTH4

ECONOMIC EVALUATION BY UK PHARMACEUTICAL AND MEDICAL DEVICE FIRMS: GENERAL TRENDS ANDPOLICY IMPLICATIONSCookson RLSE Health, London School of Economics, London, UK

OBJECTIVES: To survey expert opinion on trends ineconomic evaluation by UK pharmaceutical and medicaldevice firms, from R&D to product launch and beyond.The background is the move in some EU countries, exem-plified by NICE in UK, towards emulating the Australian“fourth hurdle” of mandatory economic data require-ments before pharmaceutical purchasing.METHODS: Interviews with a range of experts from UKindustry and academia. Industry representatives includedboth science experts and policy experts from six pharma-ceutical firms and four medical device firms; academicsincluded both health economists and pharmaceutical pol-icy specialists.RESULTS: Dramatic increase over the last decade in pro-duction of pharmacoeconomic evidence at late develop-ment and early marketing stages in response to more ag-gressive management of pharmaceutical budgets. Medicaldevice firms lag behind, due to smaller size and less aggres-sive purchasers, but large firms are now starting to designtrials with economic evaluation in mind. Limited improve-ment in scientific quality of pharmacoeconomic evidencein relation to state-of-the-art methodology (“WashingtonGuidelines”); limited use of effectiveness trials, utility indi-ces, cost-utility analysis, sensitivity analysis. Limited use ofprospective cost-effectiveness modeling in R&D go/no godecisions, even though prospective modelling of returnon investment is routinely used from early developmentonwards. Some industry representatives argue that phar-macoeconomics cannot be used to prioritize R&D since“hard” data are not available and uncertainties are great.

Abstracts 405

Some academic economists disagree, however, arguingthat prospective economic modelling is better than infor-mal opinion.CONCLUSIONS: This study raises doubts about whetherpharmacoeconomics has yet had much impact on R&Dprioritization. If pharmacoeconomics is to succeed in guid-ing firms towards developing cost effective new products(i.e., ones delivering greater population health gains perunit of cost than existing products), then firms may needstronger incentives to use prospective cost-effectivenessmodeling in R&D decisions.

PTH5

A COST-MINIMIZATION ANALYSIS OF IRON DEFICIENCY ANEMIA TREATMENTVorobiev PA1, Gorokhova SG1, Avxentieva MV1, Kobina SA2

1Moscow Medical Academy named after I.M.Sechenov, Moscow, Russia; 2Rhône-Poulenc Rorer, Moscow, Russia

OBJECTIVES: To compare the costs of baseline standardtreatment course for iron deficiency anemia with differ-ent Fe11-content medicines.METHODS: The model for non-complicated iron defi-ciency anemia treatment was worked out by experts inthe field of hematology: a course of 200 mg Fe11 dailyfor 28 days. There were 5 Fe11-content medicines avail-able in Moscow pharmacies conventionally named H, S,T, F, G with the same suggested clinical effectiveness. Ac-tive Fe11 amount per tablet, necessary mean dose ofdrug, mean duration of course treatment, drug price weretaken into account. Data on medicine prices was calcu-lated as a mean value according to price-lists of Moscowpharmacies derived from the Internet on April, 20, 1999.RESULTS: The table below shows the results of this cost-minimization analysis.

CONCLUSION: The course of treatment for iron defi-ciency anemia with F-drug that is the cheapest really isthe most expensive. Cost-saving courses are those with Sand G-drugs. Although it is traditionally used in Russia,price per pack shouldn’t be an economical criteria forchoice of treatment.

PTH6

PATIENTS’ PERCEPTIONS OF THE OUTCOMES OF TREATMENT WITH TOPICAL NSAIDS, RUBIFACIANTS AND ORAL PAIN REMEDIES. A PROSPECTIVE STUDY OF MEDICINES AVAILABLE THROUGHCOMMUNITY PHARMACIESMcNaughton D, Libby G, MacDonald TMedicines Monitoring Unit, Ninewells Hospital, Dundee, Scotland

OBJECTIVES: To examine patients’ attitudes to and expe-riences of NSAIDs, Rubifaciants and oral pain remedies.METHODS: Subjects were recruited in one of 30 phar-macies in Scotland representing rural, semi-rural and ur-ban environments. Subjects who had purchased an Overthe Counter medicine from one of three Topical NSAIDs,three Rubifaciants or one of four oral pain remedies wereasked to participate in a questionnaire based study. Thequestionnaire examined pain assessment pre and posttreatment, side effects experienced, patients’ expectationsof treatment, treatment outcomes, opinion about the rel-ative merits of the three groups of medicines and demo-graphic data about the purchaser. The questionnaireswere anonymous but identified the study pharmacy. In-formation about the effectiveness of different rates of re-cruitment was also obtained. After an initial emphasis onpharmacist inducements, the study pharmacist was sta-tioned in the pharmacies and achieved a high rate of re-cruitment and returned questionnaires.RESULTS: Of 430 questionnaires distributed, 302(70.23%) were returned; 13 patients were excluded due tospoiled questionnaires. Of the remaining 289 patients, 94(32.5%) purchased Topical NSAIDs, 105 (36.6%) Rubifa-ciants and 90 (31.1%) purchased oral remedies. There wasno significant difference between the three groups in painscores pre and post treatment (p 5 0.22). There was a dif-ference in choice of treatment type by age (p 5 0.021) withpatients aged over 60 more likely to choose a TopicalNSAID. Those who suffered symptoms less than once aweek and those choosing a Topical NSAID were morelikely to be first time users of the study treatment (both p 50.001). Those in the Topical NSAID group were also morelikely to report that they would definitely buy the productagain than those in the other treatment groups (p 5 0.014).CONCLUSIONS: Initial findings demonstrate a very lowlevel of adverse events and a high rate of satisfaction par-ticularly with Topical NSAIDs as represented by a greaterwillingness to buy again.

PTH7

GLOBAL SCALE VERSUS GENERIC AND DISEASE-SPECIFIC INSTRUMENTS TO MEASURE THE HEALTH STATUS OF PATIENTS WITH ASTHMAHajiro T, Nishimura K, Tsukino M, Oga TDepartment of Respiratory Medicine, Graduate School of Medicine, Kyoto University, Kyoto, Japan

Indicators

Conventionally named drugs

H S T F G

Fe11 mg/tab 10 100 80 45 100Equivalent daily dose

(200 mg daily) tab 20 2 3 5 2Price per pack USD* 2.06 3.18 4.42 0.95 1.53Tablet number per pack 100 50 30 10 30Price per tablet USD 0.02 0.06 0.15 0.01 0.05Cost per course of

treatment (28 days) 11.5 3.56 12.38 13.3 2.86USD 4

*Prices in rubles were converted into USD according to the Central Bank rateon April 20, 1999

406 Abstracts

OBJECTIVES: Global quality of life (QoL) is an individ-ual’s overall evaluation of their quality of life, and can bedistinguished from the total QoL, which is an aggregatecomposed of items from various instruments. Hyland etal. have developed a new global quality of life scale (Hy-land scale). The purpose of this study was to investigatehow the Hyland scale evaluates the perception of healthstatus in patients with asthma, and to compare the Hy-land scale with generic and asthma-specific instruments.METHODS: A total of 167 patients with chronic asthma(age: 48 6 8 yr, 83 males) completed pulmonary functiontesting, a measurement of their airway hyperresponsive-ness, their dyspnea rating (Medical Research Councildyspnea scale), and assessments of their anxiety and de-pression (HADS; Hospital Anxiety and Depression Scale)and assessments of their health status. The global QoLwas assessed using the Hyland scale, and the health sta-tus was assessed using the SF-36 and the Living WithAsthma Questionnaire (LWAQ).RESULTS: The score on the Hyland scale was normallydistributed (mean:[SD]:74[14]). The test-retest reliabilityof the Hyland scale was found to be high (intraclass coef-ficient 5 0.80), and mild correlations were noted be-tween the Hyland scale and the Anxiety and Depressionon the HADS (Spearman’s rank correlation coefficient[Rs] 5 20.28 and 0.40 respectively). The Hyland scalehad a moderate correlation with the LWAQ (Rs 5 0.20),and had its strongest correlation with the General Health(Rs 5 0.48) among the 8 subscales on the SF-36. The Hy-land scale did not have any significant correlations withthe FEV1, peak expiratory flow rates, or the degree ofairway hyperresponsiveness. Multiple regression analysisshowed that the Hyland scale was affected by the Anxi-ety on the HADS, but its cumulative coefficient of deter-mination was very low (R2 5 0.02).CONCLUSIONS: The Hyland scale had mild correla-tions with generic and disease-specific measures, and didnot correlate well with clinical parameters. The Hylandscale may give insight into other aspects of the health sta-tus different from what generic and disease-specific in-struments evaluate.

PTH8

POTENTIAL COST SAVINGS ASSOCIATED WITH ASTHMA DRUG THERAPIES USING INTEGRATED DRUG AND MEDICAL CLAIMS DATAAndrews K, Boscarino J, Sokol M, Yao J, Zhao Z, Powell-Threets K, Maldonato D, Fulop G, Refowitz RCenter for Outcomes Measurement and Performance Assessment, Merck-Medco Managed Care, L.L.C., Montvale, NJ, USA

OBJECTIVE: To evaluate the potential cost savings ofasthma drug therapy in helping reduce incurred health-care expenses.METHODS: We integrated 18 months of drug and med-ical claims for 49,059 continuously eligible, unique indi-

viduals. Individuals were eligible for study inclusion if, inthe first 12 months, they received a medical diagnosis ofasthma (ICD-9 5 493) or filled an asthma-related drugprescription. A total of 1,830 asthmatics were included inthe study after exclusions for those under age 6, those overage 64, and those with chronic obstructive pulmonary dis-ease. We compared total health costs by type of medica-tion (controller vs. any/none and any vs. none) in the last 6months after stratifying for age and comorbidity. RESULTS: Annualized total health care costs were con-sistently lower for those on controller medications thanthose not on controller medications. The difference wasmost noticeable for asthmatics on controller medicationswho were 40–64 years old (n 5 261); these individualshad an average annual expense of $5,961 versus $10,138for those who were not on controller medications (n 5258) (p 5 .07). The relationship between the use of anyasthma medication and costs was inconsistent. While thedifferences in cost for those under the age of 18 and overthe age of 39 were insignificant, there was a significantdifference among individuals between the ages of 18–39who had lower comorbidity scores. Those individuals onany asthma medication (n 5 176) had an average annual-ized expense of $1,400 in the follow-up period, versus$4,143 among those who were not on these medications(n 5 52) (p 5 .0022).CONCLUSION: Our study seems to confirm the findingsof others, suggesting that the use of controller medica-tions, may reduce total health care costs. We plan now fur-ther research to evaluate the effects of patient drug utiliza-tion on these cost outcomes.

PTH9

RETROSPECTIVE OUTCOMES ANALYSIS OF THE EFFECTS OF ZAFIRLUKAST THERAPY IN THE MANAGEMENT OF PATIENTS WITH ASTHMA IN THE UNITED STATESSilverman S1, Wang Y2, Klingman D3, Joy K3

1AstraZeneca, Macclesfield, Cheshire, UK; 2AstraZeneca, Wilmington, DE, USA; 3PAREXEL International Medical Marketing Services Inc., Alexandria, VA, USA

OBJECTIVES: This study analyzed managed care claimsto find out how zafirlukast, a recently established oralasthma therapy, is being prescribed in clinical practiceand to evaluate the effects of zafirlukast therapy onhealthcare resource utilisation.METHODS: We examined the claims of 780 patients(.12 years) who participated in a Northeastern US man-aged care plan and had received at least 3 months’ therapywith zafirlukast, including at least one repeat prescription.Baseline medications were determined 3 months before thestart of zafirlukast. Use of healthcare services, includingnumber of outpatient visits, inpatient stays, emergency de-partment visits, and asthma-related drug prescriptions,were compared between 3 months before and 3 months af-ter commencing zafirlukast.RESULTS: Zafirlukast was prescribed to patients on avariety of baseline medications, with 72% already receiv-

Abstracts 407

ing inhaled corticosteroids. Within 3 months of startingzafirlukast, outpatient visits, inpatient stays, emergency de-partment visits, and albuterol prescriptions were all signifi-cantly reduced compared with the previous 3 months (p ,0.002). In the 3 months before starting zafirlukast, 55%of patients received oral corticosteroids, compared with,15% in the 3 months after commencing zafirlukast.CONCLUSIONS: Retrospective data analysis may bevaluable in determining how therapies are used in clinicalpractice and in measuring the effectiveness of asthma ther-apies outside of controlled clinical trials. This study showsthat the addition of zafirlukast to existing asthma regi-mens was associated with reduced healthcare utilisationand reduced need for oral corticosteroids and albuterol,which may translate into reduced direct healthcare costs.

PTH10

EVALUATING PHARMACEUTICAL COSTS OF PATIENTS WITH CHRONIC LUNG DISEASE IN GERMAN IN- AND OUTPATIENTS SETTINGHessel F, Wasem JDepartment for Health Care Management, University of Greifswald, Greifswald, Germany

OBJECTIVES: Costs for medication are beside productiv-ity loss most relevant for patients with chronic lung diseaselike asthma and COLD. In Germany hospital prices forpharmaceuticals are not fixed by the government and notaccessible through administrative data. Official sales-prices(Rote Liste) do not represent opportunity costs.METHODS: To improve the evaluation of medicationcosts we calculated the resources used by measuring theunits of 100 fully hospitalized patients of a rehabilitationclinic for chronic lung diseases and validated with unitprices from Rote Liste. These costs were compared withaverage expenses for medication per day according to theclinic’s routine calculation. To assess medication costs inan outpatient setting we used a cost diary, where patientsnote all health care resource uses, and a standardized ret-rospective questionnaire.RESULTS: Medication costs in inpatient rehabilitationsetting were 11,4 DM per day (SD 5,0) validated by RoteListe. Actual average expenses for the clinic were 2,8 DM(Rate: 1:4). Medication costs in outpatient setting were260 DM per month according to the cost diary (n 5 28)and 270 DM using the questionnaire (n 5 41). Dropoutswere remarkably higher using the diary (44%), comparedto the questionnaire (18%).CONCLUSIONS: Investigators should be aware of thefact, that the difference between medication costs vali-dated by sales prices and hospital expenses can be crucialfor the result of an economic evaluation especially forchronic diseases with mainly pharmaceutical therapy. Inan acute hospital setting this difference might even behigher, further research is necessary. To assess medicationcosts in an outpatient setting a retrospective questionnaireseems to be the most reliable and practicable way.

PTH11

APPLYING DECISIONAL ANALYSIS TO METERED DOSE INHALER MONITORING AND EDUCATIONAL OUTCOMESSampson JM, Schmitt JBVA Medical Center, Alexandria, LA, USA

OBJECTIVES: We currently have over 2,000 active pre-scriptions for albuterol inhalers. Improper use could rendermedication not effective and result in over prescribing. Atan average wholesale price of $25.00 per inhaler the cost isnot insignificant. We wanted to know if patients with air-way obstruction were able to deliver medication throughtheir inhaler correctly and whether monitoring would becost effective in terms of training and education cost.METHODS: We sampled thirty-eight consecutive patientsin pulmonary clinic with chronic obstructive pulmonary dis-ease being treated with albuterol metered does inhalers.Assessment was performed using an aerosol inhalationmonitor and observation. Critical steps evaluated in themaneuver: 1. Shaking canister, 2. Begin inspiration betweenfunctional residual capacity and residual volume, 3. Firingof inhaler and synchrony with inspiration, and 4. Breathhold time. Inhaler performance, costs of medication, andtraining costs were evaluated using Decisional Analysis soft-ware (Data 3.5).RESULTS: Only thirty three percent of patients could per-form the inhaler technique correctly on screening. Sixtyseven percent failed and underwent training; of this groupwhich failed ninety seven percent passed after training.CONCLUSIONS: We found this process of inhaler mon-itoring to be cost effective for our patient population.

PTH12

INTENSIVE CARE COSTS OF TREATING PEDIATRIC PATIENTS WITH ACUTE RESPIRATORY DISTRESS SYNDROME: HOW CAN THESE BE MEASURED?Sauriol L1, Lacroix J2, Lebel D2, Morneau S2

1Hoechst Marion Roussel Canada Research Inc., Montreal, Quebec, Canada; 2Sainte-Justine Hospital, Montreal, Quebec, Canada

OBJECTIVES: The objective of this study was to collectinformation about health care resources used in the In-tensive Care Unit (ICU) to treat pediatric patients withacute respiratory distress syndrome, and to calculate to-tal treatment costs. To collect this information, the use ofadministrative databases or hospital records was consid-ered. The latter method was used. Costs for each re-source were estimated through micro and gross costing.METHODS: A non-comparative retrospective cohort studywas conducted using hospital records. The study popula-tion was comprised of pediatric patients treated in theICU of the Sainte-Justine Hospital. Selected were ninety-six (96) patients who required ventilation support fol-lowing acute respiratory distress syndrome in 1996. Allhospital files were reviewed to collect data on patient

408 Abstracts

characteristics, diagnoses, hospitalizations, professionalfees, medical interventions, medication use, laboratorytests, and complications. These data were collected fromthe first day the patients were admitted in the ICU, until theday they were transferred to another ward. A trauma phy-sician, a pharmacist, and an inhalation therapist helped todevelop the Clinical Research Form (CRF). Two researchnurses completed the CRFs, and a clinical research assis-tant collected cost data.RESULTS: The use of micro costing is a very reliable wayof estimating treatment costs. However, it was found tobe a long and expensive process. In our study, researchnurses spent up to 8 hours per record to collect the infor-mation needed to complete the CRF.CONCLUSION: Given the extensive time needed to realizemicro costing studies, it is understandable that modeliza-tion has occupied a big place in pharmacoeconomic evalua-tions. However, in some cases, even modelization cannotreplace micro costing. Alternative ways, such as the use ofadministrative databases, should be considered. However,in Quebec, there is still a need to assess the reliability of us-ing administrative databases for economic evaluations.

PTH13

THE APPLICATION OF PORTFOLIO MANAGEMENT TECHNIQUES TO PHARMACEUTICAL R&D PROJECT SELECTION AND COST-EFFECTIVENESS ANALYSISTran G1, Pang F2

1The Lewin Group, Bracknell, Berkshire, UK; 2Centre for Health Economics, University of York, York, North Yorkshire, UK

OBJECTIVES: The objective of portfolio management isto create the most value from a portfolio of projects. It

attempts to ensure that the projects in the portfolio meetthe strategic goals of the business by determining whichprojects are to be funded and at what levels. This studyinvestigates the application of two portfolio managementtechniques involving Monte-Carlo simulation and thePearson Index to two diverse situations: pharmaceuticalR&D investment prioritisation decisions and health eco-nomic evaluations using cost-effectiveness analysis.METHODS: To assist the decision-maker to prioritisepharmaceutical R&D projects in which to invest, we con-structed a model involving Monte-Carlo simulation andthe Pearson Index. This model was applied to data on costof development and probability of success for each clinicaldevelopment phase (I–III), as well as overall return-on-investment (ROI) for each project in the drug portfolio. Toassist healthcare policy-makers to select the cost-effectivetreatment strategy, we constructed a model using Monte-Carlo simulation, to incorporate variability in medical re-source use and unit costs, and the bootstrap method to es-timate the 95% confidence interval for the incrementalcost-effectiveness ratio (ICER). This model was applied todata from a multinational health economic trial.RESULTS: For pharmaceutical R&D investment prioriti-sation decisions, we derived a probability distribution forthe Pearson Index for each project which was then used toprioritise project selection in light of uncertainty in devel-opment costs, success probabilities and overall ROI. In ad-dition, we derived an estimate of the ICER and its 95%confidence interval, which was then used to select the cost-effective treatment strategy.CONCLUSIONS: Portfolio management techniques cansuccessfully be used to support the decision-maker in mak-ing optimal decisions in pharmaceutical R&D project se-lection and to select the cost-effective treatment strategy inlight of the uncertainty/variability in input parameters.