REVIEW ARTICLE Prl<>'moooECO<lO<I'Wcs 7 ( 3): 221 ·22&, 1'105 1170-J09O{9:ji'OOJJ..(l22 \ 1W4.00/0

Defining the Patient Group for Cost-Effective Withdrawal of Antihypertensive Therapy Lawrellce R. Krakoff 1,2 and Sylvia WassertheiJ-Smoller3

Mount Sinai School of Medicine, New York, New York, USA 2 Englewood Hospital and Medical Center, Englewood, New Jersey, USA 3 Albert Einstein College of Medicine, Bronx, New York, USA

Contents Summory 1. Well Controlled Hypertension .

221 222 223 223 224 225 225 226 226 227

2. Studies of Withdrawing Drug Treatment 2.1 The Dietary Intervention Study of Hypertension . 2.2 The Hypertension Control Program .

3. Economic Aspects of Withdrowing Antihypertensive Medication. 3.1 Reduction of Drug Costs . 3.2 O ther Potential Benefits of Drug Withdrawal 3.3 Costs of Nonpharmocologicollntervention

4. Conclusions .

Summary Many people receiving drug treatment for hypertension have had good control of their blood pressure for several years. It is feasible. and might be desirable. to discontinue drug treatment in selected groups of these patients if their long term outcome was not adversely affected. A few small series and 2 larger trials indicate that withdrawal of antihypertensive medication can be followed by 1 10 4 years of normal blood pressure. especially for those with mild hypertension (pretreat­ment diastolic blood pressure < lOOmm Hg) and who have no evidence of target organ damage.

In the 2 randomised trials of withdrawing medication emerging from the High Blood Pressure Detection and Follow-up Program. the addition of nutritional interventions (bodyweighl reduction ancVor dietary salt restriction) was shown to enhance the likelihood that antihypertensive drugs cou ld be deferred for long intervals. Sustained bodyweight reduction for overweight patients with mild hy. pertension was the most successful approach.

Short term nonpharmacological programmes have been evaluated for their economic effects on this process through cost-effectiveness and cost-benefit anal­yses. The available studies imply that reduced costs and decreased adverse effects associated with withdrawal of antihypertensive medication are offset by cost of the nonpharmacological interventions needed to maximise continued blood pres­sure control. Other influences, such as willingness to pay for nonpharmaeological

222 Krokoff & Wasserlilci/-SI1JQlIer

management or change in sense of well-being when on or off medication. need consideration for comprehensive cost-effectiveness analyses. Development of highly acceptable. effective and low cost alternative interventions that focus on those most likely to respond and remain compliant arc necessary before with­drawal of anti hypertensive medication becomes widely accepted for management of most drug treated hypertensive patients.

Effec ti ve reduction of arterial blood pressure in hypertensive patients decreases future li kel ihood of fatal and nonfatal cardiovascu lar disease. This view was origi nally predicted by a large and con­sistent body of epidemiological ev idence and has since been con fi rmed by many cli nical trials. II ·31

The largest fractio n of patients considered to be hypertensive. by current crite ri a, have pretreatment pressures minimally above the upper limit of ac­cepted normal r anges. When treated, thei r pressure oftcn fa ll s quickly to lower levels and remains so for long periods of time.

T here has been a general reluctance to discon­ti nue anti hypertensive drug treatment because of an assum ption that the underlying hypertensive process is irreversible. However. it has a lso been suggested that, on occasion, those patients receiv­

ing antihypertensive drugs whose blood pressures have remai ned in normal or low risk ranges fo r several years might be considered for withdrawal of drug treatment with reasonable expectat ion that their blood pressure wou ld remain be low that ne­cessitating d rug treatment. If a substantial fract ion o f hypertensive patients could be successfully withdrawn from medication, there might be a sig­nificant effect on the cost effectiveness o f treat­

ment because purchase of drugs contributes so much tooverall expenditures. 14.71 This brief review exami nes the uption of withdrawal of drug treat­ment and also explores the potential economic e f­fects of strategies for successful drug wi thdrawal.

1. Well Controlled Hypertension

With in 6 months of the initiation of drug treat­ment for hypertension, it has been noted that a substant ial fract ion of patients become normo­tensive (defined as having cl in ic blood pressu res < 140/90mm Hg) and remain so for long periods. In

the High Blood Pressure Detection and Follow-Up Program (HDFP) 40% of patients were controlled

on a diuretic alone, and 24% requ ired an additional

agent {either methy ldopa or reserpine).I HI Clini­

cians will often attribute this desirable response to

the effect of active antihypertensive drug treat­ment. However. there are a lternative explanations. as shown i n table I.

Small numbers of measurements over a short

interval are often used for primary and secondary screening for hypertension. As such. there is a high like lihood that regression dilution will result in

misclassification of some patients as hypertensive,

when their usual pressures. based on more record­

ings. would be far lower.19.111 This fa lse positive

diagnosis is partly due to the posit ion of the cut-off

point for diagnosis on the down-slope of the d istri­buti on curve of blood pressures in the general pop­ulation. A small error in either d irec ti on will result in many more normotensive people being d iag­nosed as hypertensive than hypertensive people be­ing d iagnosed as normotensive. With time and

more blood pressure meas urements the actual blood pressure will be approached. This g ives the

appearance of a decrease in blood pressure due to treatment when the real effect is due to regression

to their mean. 1121

In addi tion, some patients with clin ic blood

Tabte I. Factors cO!1tributing to we ll CO!1trolied hyper1enSion. delined as average syslolic blood pressure <140mm Hg and diaslolic blood pressure <9Omm Hg lor 3 10 5 years

Misclass~ied patients (false positive diagnosis) wilh lower usual pressures. due to regression di lution at baseline or 'white coat" hyper1ension

True placebo responders

True act .... e drug responders with reversal 01 hypertens .... e palhophysiology (possible reversal 01 vascular structural abnormalitjes)

Withdrawing Antihypertensive Therapy

pressure readings in the hypertensive range will repeatedly have lower or normal pressures when measured duri ng ambulatory blood pressure mon­itoring, a p henomenon known as 'white coat' hy­penension.IIJ. 161 This pattern may persist for many years l171 and is therefore dis tinct from the misclas­sification of hypertension due only to regression dilution. The combination of these 2 facto rs might account for the misdiagnosis of20 to 40% of those initially thought to have mild hypertension.l14-161

In nearly all placebo-controlled tri als of long term antihypertensive treatment. patients receiv ing placebo have had an average reduction in blood pressure. While part of the apparent placebo effect may be due to regression to the mean. a true p la­cebo effect (i.e. a response based on/yon the belief that the medication is effective) may still resu lt in decreased blood pressure in some patie nts. The magnitude of a true placebo effect on arterial pres­sure remains unknown. Recent studies usi ng non­invas ive ambulatory blood pressure mon itoring demonstrated that when adequate numbers of measurements are made during ordi nary activity before treatment and on placebo, no reduction in pressure is apparenLl 18•191 Nonetheless. in the large scale prevention trials of drug therapy. an inde­pendent t rue placebo effect can not be fully ex­cluded in some patients.l101

A substantial decrease in blood pressure is as­sumed to be the di rect result of active drug therapy for most treated patients with hypertension . After several years of a sustained reduct ion in blood pressure, reversal of the processes that caused the initial elevation in pressure might occur. Recent studies have emphasised the importance of altered structure in resistance vessels of hypertensive pa­tients due to increased wall thickness relative to reduced lumi nal di ameter. 1211

No doubt vasoconstriction by smooth muscle still plays an important role in the increased sys­temic vascular resistance of hypertension. How­ever, the decreased vasoconstriction caused by an­tihypertensive drugs would be expected to reverse quick ly if therapy were disconti nued. Alterna­tively. normalisation of the structural alterations in

() Acis In!emOlionoll.i'Ti!eIl. AI righl$ r€"$erved .

223

hy pertensive resistance vessel s by prolonged treat­ment might be fo llowed by a sustained period of normal blood pressure. Gluteal biopsy studies of resistance vessels in a small number of patients observed over 12 to 14 months suggest that such vascular changes may take place.ln.nl Limited as these reports are, they provide a rationale fo r the hypothesis that long term blood pressure reduction by drug treatment may allow intervals duri ng which treatment can be withdrawn with only a low risk of hypertension recurring.

2. Studies of Withdrawing Drug Treatment

Short term cessation of drug treatment for 2 to 4 weeks has often been assessed in studies of new antihypertensive agents. For example, in the enrol­ment phase of I such report, approx imately 25% (424 out of 1666) of the mild to moderate hyper­tensive patients initially screened failed to meet the entry criterion of a seated diastolic blood pressure ~5mm Hg during withdrawal of pri or treatment and on placebo.1241 Most had blood pressures be­low the entry criteria.

A few reports of very small series have shown that it is possible to discontinue drug therapy in some hypertensive patients and that blood pressure remains at acceptable levels fo r months to years without specific measures,l25.261 However, 2 sub­stantial randomised intervention tr ials (see sec­tions 2.1 and 2.2) were conducted in groups of pa­tients who had been well controlled on medications during the 5 years of drug treatment of the HDFP study. Both studies were designed to explore the va lue of li festyle changes (non pharmaco logica l therapies) for maintenance of lowered blood pres­sure that would preve nt or delay resumption of drug treatment.

2.1 The Dietary Intervention Study of Hypertension

The Dietary Intervent ion Study of Hyperten­sion (DISH)1271 selected well controlled hyperten­sive patients from 3 of the HDFP special care clin­ics. 496 patients (57%) whose hypertension was

Pha,mocoEC()I"IO(I""oiC$ 7 (J) 1995

Tlble II. FractiOIl ot patients in various Intervention groups remaining ott drug treatment after 56 weeks In the Dietary Intervention Study of Hypertertsiofll21J

Patient gfO\Jp and interventiOf1

Overweight - saM restriction

Overweight - calorie reduction

Non-overwetght - no ontefVelltion

Non-overwetghl - saM restriction

Fraction (% ) remaining oH medication overllVmikl

.... "'''' .. " "'"

well controlled with drug therapy were recru ited from 865 of those being actively fo llowed-up. based on stringent crileri:! incl uding no diastolic blood pressures ~95mm Hg on drug treatment (lnd last 2 clini c diastolic blood press ures readi ngs <90mm J-1g. Eligible patients were divided into 2 groups. the overweight (~ 1 20% above ide:tl body­weight) or non-overweight « 120% ideal body­weight). The overweight were allocated to 3 inter­ventions: observation only. diet:!ry salt restriction onl y or a programme of body weight reduction (cal­orie restriction) only. The non-overweight group were :!lIocated 10 either a nonintervention status or to the low salt strategy. A comparison group was randomiscd to remain on their m cdicll tion.

The pattern of those remaini ng normotensive during a follow-up interval of :tbout I year without the need for ant ihypertensive medication is s hown in table 11 . There were nonsignificant trends favour­ing bodywe ighl reduction for the overweight or salt restrict ion in the non -overweight mild hyperten­sive patients comp:ued with no dietary interven­tion. However. the individual groups were sm:tll (33 to 41 patients). Nonetheless. more than 50% of those with mild hypertension at entry to the HDFP cou ld fo rego at least I year of drug therapy after the 5-year period of good control.

Plasma renin activity (PRA) was measured in a subset of DISH participant s 4 months after begin ­ning the st udy. At thi s time they remained normotensive and were undergoing the various in­terventions of the trial. 75 such patients we re eva l­uated. and the med ian PRA at Ihis time was 0.53 nglmllh. a relative ly low level. Over 56 weeks. sig-

Kmkoff & WasS('fthril-Smolil'f

nificantly more of those with PRA above thi s me­dian remained off drug treatment compared with those below this value; the adjusted risk ratio for return to medication was calculated at nearly 3.0 for e:tch log unit of decreased PR A.!2~)

This finding is consistent with resu lts of a the Trial of Antihypertcnsive Intervention and Man­agement (TAIM ),1 291 a random ised clinical trial of combinmions of diet and drug therapy for mild hy­penension. The invest igators of this trial found that patients with higher pl:tsma ren in act ivi ty levels h:td:t greater blood pressure lowering response 10 dietary thempy than did patients with lower renin levels. Other studi es. however. classifying pat ients by blood pressure responses to hi gh and low salt intake have reported opposi te findings. namely that salt sensi tive hypenensive patients arc more likely have low plasma renin levels.lJO) The low renin state is v iewed by some as adaptation to a sustained e levation in arterial pressure.J3 tl If so. perhaps those with higher PR A in DISH or TAIM were less li kely 10 have sustained hypertension at entry 10 the these trials.

2.2 The Hypertension Control Program

A 4-year trial of former HDFP pmients. the Hy­pertension Control Program (HCP).I32I enrolled 189 patients who had been effectively treated by criteria si milar to those in the DISH study and al­located them to 3 groups. One received no inter­vention. A second received nutritional counse lling designed 10 reduce bodyweight (for Ihe over­weight). salt intllke (a ll ) and alcohol (ethanol) in­gesti on to :526g. eq uivalent to 120z (360m!) beer. 4.5oz (135m]) wine o r 1.50z (451111) distilled liq­uor. The thi rd comparison group con tinued to take antihypertensive medication.

The 4-year experience of the HCP i s presented in table III . As with DISH. det:tiled analysis sug­gested Ihat those who entered HDFP with mild hy­pertension were more likely to remain without the need for resumpti on of drug trealment during the Hep. In :tddilion. the ability to sustain a reduction in relative bodywcight predicted a sign ificant like­lihood that blood pressure wou ld stay withi n the

Withdrawing Anlihypert('nsive Th('rapy

Table m. Fraction (%) 01 pat ients remaining off drug treatment in the Hypertension Control Program after each year 01 intervention or observationl:):!1

Year 01 Nutritional No interventiO<l

intervent iO<l intervention"'

69 50

2 " 33

3 " " , 39 5 a Coonselling wrth respect to bodyweight reduction, sa~ intake

and alcohol (ethanol) consumption

normotensive range without drug treatment. When adjusted for other faclO rs, sustained reduction of sa lt intake alone was not sign ifica nll y associated with a pattern of continued drug withdrawal.

Taken together, the resu lts of DISH and Hep trials imply that a substantial proportion of patients with mild hypertension controlled on antihyper­tensive medication for several years might be able 10 forego drug therapy forthe next I to 4 years with the proviso that a defi ned nutritional programme would replace prescriptions. In general, these pa­tients would have the characteristics described in table IV. The economic implications of a strategy to explore this option are therefore worth considering.

3. Economic Aspects of Withdrawing Antihypertensive Medication

Cessation of antihypertensive medications might affect the overall cost effectiveness of management for a population. Factors to be assessed in such cal­culations might include: (i) reduced expenditures for medication; (ii) reduced costs attributab le to reduced adverse drug reactions; (i ii) increased quality-years because of cessat ion of subjective symptoms formerly due to medication; (iv) risk of increased blood pressure after drug withdrawal; (v) possible change in non hypertensive risk fac­tors after drug withdrawal; and (vi) cost of non­pharmacological interventions or li festyle changes as alternative antihypertensive therapy and unfore­seen changes in cost or quality indices. Only a few of these have been eva luated even in prelimi nary form, wh ich limits firm conclusions. Some compo­nents of the problem will be briefly discussed.

C A~ tnla<notionolLimilod AJ rights reserved

225

3.1 Reduction of Drug Costs

For those groups in the DISH and Hep studies who were withdrawn from med icat ion, the cost of drug treatment (including potassium supplements for those on diuret ics) decreased during the period that patients remained drug-free. Since the special care clin ics of HDFP employed a diuretic-bascd stepped care st rategy with addition of reserpine, methyldopa or proprano lol as the second step. the annual cost of drug treatment was relat ively low, in the range of S$US250/year (1990 dollars). Us­ing the 4-year trends from the Hep study to deter­mine the frac tion of untreated patients each year, the calculation of reduced costs due to discontinu­ation of ant ihypertensive medication (as used in HOFP) can be made. as shown in fi gure I. The total savi ngs fo r lOOO patients over 4 years would be about $US503 000 for the nutritional intervention group and $US260 000 for the nonintervention group. The difference represents the gain projected for successful nonpharmacological interventi on or $US243 000.

While HDFPand other trials such as the Systolic Hypertension in the Elderly Program (S HEP)133) used low cost drugs. more expensive agents have become widely used in recent years, namely the angiotensin converting enzyme (ACE) inh ibitors and calcium antagonists. 134) The reduction in drug costs depends on which drugs would be withdrawn. Would th is amount enti rely support the costs of providing the successful li festyle intervention?

Tabte IV. Candidates lor withdrawat 01 antihypertensive drug

treatment

Diastolic blood pressure (DBP) <9Omm Hg for 3-5 years on medication

Pretreatment DBP ";100mm Hg

Bodyweight <: ltO% 01 Kleal (body mass index :?:27.5 1Of men Of <:26.4 for women)

Likelihood 01 compHar.ce w~h txldyweighl reduction programme

Lack of target organ damage {i.e. lell ventricular en largement,

p!"oteinuria. elevated sel\Jm creatinine levels (<:1.3 mgldl Of 100 Ilmol/L). cerebrovascutar or peripheral vascular diseasel

Normal (not low) plasma renin activity

226

• Nutritional intervention

• No intervention

~

a'L,-----~----~----_=r_ 2 Yea, 3 4

Fig_ 1. Predicted reduction in costs fordrug treatment after with­drawal of medication based on results of the Hypertension Control Program trial lor 1000 patients in either the nutritional intervention or nonintervention groups. Ill) The annual cost of medication was assumed to be $US250.

3.2 Other Potential Benefits of

Drug Withdrawal

Antihypertensive medications cause adverse effects. Some are subjective symptoms reducing quality of li feP5J while others represent potential increased long term risk, i.e. the proposed impa ired glucose tolerance and/or changes in serum lipid levels that have been linked 10 diuretic use.l 361

There is a large literature base dealing with these issues. which are as yet unresolved and contro­ve rsial.

Concern about the metabolic adverse effects of diuretic-based antihypertensive strategies may have led to switching to the newer lIgents. How­

ever. experience from the SHEP tdal lEI and the recently concl uded Trial of Mild Hyperte nsion Study (TOMHS)131 suggest Ihat low but effective dosages of a diuretic, f}-blocker, ACE inhibitor, calcium antagonist or an a-blocker arc remarkably well tolerated and have minimal effects on serum glucose or lipid fractions in highly compliant pa­tients. Thu s. it is uncertain whether withdrawal of commonly used medications at the low dosages. as often prescribed for mild hypertension. will sub­stantially alter either quality of life (i.e . specific

symptoms) or nonhypertensive risk.

() Adis InternolionolUm/led. AI rights reserved.

Krakoff & Wasserthei/-Smol/er

3.3 Costs of Nonpharmacological Intervention

The actual costs for development and mainte­nance of the lifestyle intervent ion s used in the DISH and HCP trials are not available for review. As these were research programmes and included a grealer range of activities than the intervention s alone. estimates from these trials might be greater than costs for deli very of care withoutlhe research component. However. a comparative study has been conducted in Swede n evaluating a non­pharmacological intervention and parallel drug treatment (using a f}-blocker- based strategy) that included 400 patients in 8 cent res.l37] Lifesty le management was undertaken by trained nurses who visited patients monthly. Physician visits oc­curred every 6 months. The programme lasted 2 years. and there was an additional 2-year period of follow-up assessment.

The cost of the nonpharmacological programme was estimated at 5300 Swedish kronor (SEK) [$US833] per patient higher than for drug treat­ment with a calculated benefit of SEK3200 ($US533). Thi s resulted in a net loss of SEK2100 ($US350) per patient when nonpharmacologicai methods were compared with drug treatment. A second. smaller study of 64 overweight men has been conducted by 10hannesson and Fagerbergl38 1

comparing e ither drug or diet treatment for I year. Changes in both blood pressure and serum lipid levels were taken into account in calc ulation of benefit. The computer anal ysis included life years gained and wi lli ngness to pay in 5 separate cost-ef­fectiveness analyses. Drug treatment was the pre­ferred option in 3 of the 5 simulations. A cost-ben ­efit analysis found no difference between the 2 groups of the study.

These calculations suggest thai a strategy de­pendent on a programme of nonpharmacological intervention to increase the fraction of those who could remain withdrawn from antihypertensive drug treatment might not reduce overall healthcare expenditures. In fact. such a programme cou ld in­crease the cost of care relative to the option of maintaining patients on medication.

PharmocoEconomics 7 ( 3) 1995

Withdrawing Antihypertensive Therapy

II might be assumed that nearl y all patients tak­ing antihypertensive drugs would welcome the op­portun ity to have them discontin ued . There is. however. little documentati on as to the freque ncy of thi s preference withi n treated groups. A clue is provided, however, from the DISH study as most of the patients were given a mood questionnaire at entry to and concl usion of the triaJ.l391 When those randomised to remain on medication were com­pared with those who were withdrawn. a slight, but significant improvement in mood score occurred in the drug treated group. However, among the groups withdrawn fro m medication. those who were successfu l in achiev ing sustained bodyweight reduction had a small improvement in m ood.

These results s uggest that withdrawal of medi­cation may be a more complex phenomenon than ori ginally conceived. Perhaps many patients be­come convinced that taking thei r drugs confers benefit s and have a sense of decreased protection (reduced we ll-being) without their daily tablets un­less an alternate behaviour (i .e. successful body­we ight reduction) can replace the p rescript ion. Cost-effecti veness project ions for drug withdrawal strategies will need to take th is effect on quality of lifc into account.

4. Conclusions

There is li ttle doubt that the entire process of screening for hypertension. detection of those with elevated blood pressure and therapy by antihyper­tensive medication is beneficial by reducing car­diovascular mortality a nd morbidity. Nonetheless, withdrawal fro m drug treatment for selected groups is appeali ng. Characteristics of those more suitable for a trial of discontinuing medication with likelihood of success can be del ineated. Such patients include those who arc compliant and had mild hypertension at baseli ne without target organ damage and who have had blood pressu res in the normal range for several years on treatment.

For those highly compliant with a nutritional programme, nearly half would be able to remai n drug-free for 2 years. Nearly one-thi rd of the less compliant patients could be drug-free for that in-

227

terval. Currently avai lable estimates suggest that an act ive intervention programme focusing on a nonpharmaco[ogica l, nu tritional intervention wou ld actually add to the cost of treat ment. Such a programme might also yield a positive cost-effec­tiveness or cost-benefi t projection for keeping those withdrawn from drug treatment at low blood pressures without medication .

Theseconsiderations suggest that low cost strat­egies be explored . These strategies should be de­signed to promote patient choice to achieve the de­sired nutritional goals both for ini ti al treatment and for the sustained reduction in blood pressure after drug withdrawal. Considerations of cost effective­ness imply the need to identify patients who would both respond to and be compl iant wi th pro­grammes in li festyle change before inst ituting such interventions.

References I. MacMahon SW. CUller JA. Furberg CD. el al. Tile effects of

drug treatment for hype"cnsion on morbidity and monality from cardiova.<;cular disease: a review of randomized clinica l lrials. Prog CardiovJsc Dis 1986: 24: 99-118

2. Collins R. Peto R. MacMahon S. tt 31. Blood pressure. stroke. and coronary hean disease. Pan 2. shon· lerm reduclions in blood pressure: ovcrview of randomized drug trials in tlleir epidemiologicol contexl. Lancel 1990: 335: 821-38

3. Neaton JD. Grimm Jr RH. Prineas RJ . et al. Trealment of Mild Hype"cnsioll Study: final resuhs. JAMA 1993: 201: 713·24

4. Johanncsson M, Borgquist L. Jonsson B. The costs of lreating hype"en .• ion in Sweden: an empirical investigation in pri. mary health care. Scand J Primary Health Care 1991: 9: 155·60

5. Kawachi I. Malcolm LA. The cost·effectivcnes<oflre3ting mild to moderale hype"cnsion: reappr.!isal. J Hypeflen. 1991: 9: 199-208

b. Flelcller A. Pressure 10 lreat and pressure to cost: 3 review of eost ·cffecti'·c n~ss analysis. J Hype"ens 1991: 9: 193-8

7. Strasser 1. Rclal;-'c costs of anlihype"cnsi"e drug treatment. J Hum Hype"cns 1992: 6: 489_94

8. The Hypertension [)election and Follow-up Program Coopera­live Group. Mild hypertensives in lhc Hypertension Detection and Follow·upProgram. Ann N Y Acad Sci 1978: 304: 254-66

9. MacMahon S. Pcto R. CUller J. et 31. Blood pressure. slroke. and coroMry Ilean disease. Part I. prolonged differences in blood pressure: prospecti,'c observational <ludics correcled for the regression dilution bias. Lancet 1990: 335: 765_74

10. Pcrry Jr HM. Milkr JP. DifflCultie. in diagnosing hype"cnsion: implicalions and altemalivcs. J Hypeflcns 1992: 10: 887-96

II. Schcchter CB. Adkr RS. Bayesian analysis of diastol ic blood pressure mCaSuren1<!nt. Med Decis Making 1991 ; 8: 182-90

12. Pearce KA. Grimm RH. RaoS. ct al. Population-derived com­pari«)ns of ambulatory and office blood p<c,surcs: implica· tion, for the delermination of uSlial blood presgure and the

228

oon<:ept of white cnal hypenension. Art'h Jrllcm Med 1992: 1 ~2: 750·6

13. M~nciaG. Benini.: .; G. Grassi G,tl al. EffcC1S0fblood pressu"" nlcasun:menl by the doctor on patient's blood p"'Ssu,," and llearl rale. Lancet 1983: 2: 695·8

14. Pickering TG. J~mcs GO. Boddie C, ('I al. How common is whi te coat hypertension? JA MA 1988; 259: 225·8

15. Hoegllolm A. Kristensen KS. MadSC'n NIt. ('I al. White coal hypertension diagnosed by 24-h ambulatory moniu)ring. Am J HYJ'C'nens 1992: 5: 64·70

16. Eison fl . Phillips RA. Ardcljan M. ('I aJ. Diffc",nas in ambu­la10ry blood prcssu«: bel,,-un men and women wilh mild hypencnsiQn. J Hum Hypencns 1990: 4: 400-4

17. Martin K. Phillips RA. Krakoff LR. PcniSlcnl white Coal hy_ pertension. Am J Hypencns 1994: 7: 368-70

18. Weber MA. Cheung 00. Gmening.:r WF. Ct al. Characterization of antihypertensive therapy by whQle .day blood pressure moni loring. l AMA 1988: 259: 3281·5

19, Zanchcll i A. Bianchi L. Bozza M.el ~I. Anl ihyperlcnsivec fTe<:IS o f n'fedipine GITS on dinical and ambulalory blood pres· sures in c~nl,al hypcnensives. High Blood Pfcss I~: 3: 45·56

20. Unlrealed mi ld hypcnension: a reporl by lhe m~nage""nl com· mi ll« of the AUSlral,an lllcrapculic Trial in Mi ld Hypcnen· ,ion. Lancet 1982: I: 185·9 1

21. Heagcny AM. AalkjacrC Bund Sl .el a1. 5rn;l1l anery ~rucl ure

in hypenension. HypcTtcnsion 1993: 21: 391·1 22. Hcageny AM. Bund 51. AalkjlOe1' C Effe<:ls of drug lreal .... m

on human resislance ancriole morphology ill encnlial hyper· Icn.tion: direct cvidcn« for structural remoo:lelling of resis· lance vessels. Lancel 1988: 2: 1209.12

23. Aalkj.er C. Eiskjaer H. Mul vany MJ. CI al. Aboormal siruciure and function of isola led subculaneous resist ~nce vessel.' fro m usential hypertensive palienls despile anlihypo.: nensive Ircal· men!. J Hypcnens 1989: 7: 305·10

24. KrnkofT LR. Br~"o EL. Tuck ML. cl 31. Mode", ApproKh '0 lhe Trea' .... m of Hypenension (MATH) Siudy Group. Nifed­,pine gaslroinlcslillal lhernpculic system in lhe lreal .... nl of hypencIIsion: results of a nlu ll kenler lrial. Am J Hypcncns 1991 : 3 Suppl.: 3 18S·25S

25. van den Ba;ch WJHM. Mol W. van Gcrwen W. et 31. Withdrawal o f anlihypcnensi,'C drug.. ill Sl:1e<:led p;!'~111$ [CQrTCspond· coccI, UIICCI 199.1: 343: 1157

26. Aylen M. KelChin S. Slopping Ircat .... nl in p<ll icnts Wilh hyper­lension (eorrcspondeocel. BMJ 1991: 303: 345

27, Langford fiG. Blaufox MD. Oberman A. e l 31. Dielary lherapy slows Ihe relurn of hypenension aflcr slopping prolonged med,cal ion, JAMA 1985: 253: 657-64

KrakQlf & Wasserlheil·Smolltr

28. Ho GYF. B laufo~ MD. WasSl:nheil·Smolicr S. e\ al. Plasma renin predicts success of anlihypcTtensivc drug wi lhdrawal. Am J Hypenens 1994 Aug: 7: 679-84

29. Blaufox MD. Lee HB. Davis B. el al. Reni n predicts diaslo lic press ure respon'\\: 10 nonph annacologie ~ nd pharmacologic Iherapy. JAMA 1992: 267: 1221-5

30. Kawasaki T. Oclu CS. Bantcr FC. el al. The efTe<:1 of high·w. dium and 10"'-sodium inlakes on blood pressure and OIher related variables in hUrn;ln subjttts with idiopathic hype nen­sion. Am J Med 1978: 64: 193-8

31. Sealey JE. Blu .... nfdd JO. Bell GM. et a!. On the rellal basis for essenlial hypcTtellsion: nephron helerogclICily wilh di s· cordant rellin secrelion and sodium CJ<crelion causing II hy· penensi"e vasoconstriction-"olume rdationship. J Hypencns 1988;6:763-17

32. Siamier R. Siamier l. Grimm R. el 31. Nulri lionallhernpy for h igh blood pres~ure: final repon of a four-year rnndomized conlrolled trial - the Uypen ension Conlrol Program, JA M A 1987; 257: 1484·9 1

33. SHEP CooperJlh'e Rc...:arch Group. Pfcvenlion of slrokc by anlihypencnsivcdrug trcallnenl ill o lder persons wilh isolaled syslolic hypertension: final rewll$ of the Syslolic Il ypcncn· s ion in lhe Elderly Progrnm(SHEP). JAM A 1991 : 265: 3255.t.4

34. Sinclair BL. Ashton T. Jackson R. CI al. Trellds in anlihypcncn· 5i"e medical1oncOSis inacohortof Aucklandcrs 1982-7. Au§! N Z J Med 1989: 102 (877): 521·3

35. Croog S H. Levine S. Tesla MA. CI al.lllc effecls of anlihypcr­lensi,'c therapy on lhequalityofl ifc. N Engl J Med 1986: 314: 1657-64

36. Grimm RH. Leon AS. Hunni nghakc DB. cl al. Effects of Ihia­zide diuretic~ on plasma lipids and lipoproteins in mi ldly hy_ pt'nensive patients: adoublc blind controll ed trial . Ann lntcm McdI98 1;94:7.11

37. Joh~nnesson M. Aberg H. Agreus L. el ai, Cosl·benefi l ana lysis of non·pharmacological Ireal menl of hypenellsion. J Inlern Me<! 199 1; 230: 307-12

38. Johanncsson M. Fagerberg B. A health-cconomic comparison of diel and drug lreatment in obese men with mild hypencn_ sion. J Hypencns 1992: 10: 1063-70

39. Thaler L. Wasscnheil -Smolier S. Blaufox MD. el 3.1. Effect o f wilhdra"'al on anlihypencnsive drug 011 depressi,'c mood. Am J Hypenens I~: 6: 1055-62

Correspondence and reprints: Dr LAwrmct R. Krako/f, Chief of Medicine, Englewood Hospilal and Medica! Center. 350 Engle S treet. Englewood. NJ 07631, USA.

Pho!rnocoEconom 7 (3) 1995