2 preventing potentially inappropiate medication use in hospiutalized older patients

ORIGINAL INVESTIGATION

LESS IS MORE

Preventing Potentially Inappropriate Medication Usein Hospitalized Older Patients With a ComputerizedProvider Order Entry Warning SystemMelissa L. P. Mattison, MD; Kevin A. Afonso, BSBA; Long H. Ngo, PhD; Kenneth J. Mukamal, MD

Background: Potentially inappropriate medication(PIM) use in hospitalized older patients is common.Our objective was to determine whether a computer-ized provider order entry (CPOE) drug warning systemcan decrease orders for PIMs in hospitalized olderpatients.

Methods: We used a prospective before-and-after de-sign among patients 65 years or older admitted to a large,urban academic medical center in Boston, Massachu-setts, from June 1, 2004, throughNovember 29, 2004 (forpatients admitted before the warning systemwas added),and fromMarch 17, 2005, through August 30, 2008 (pa-tients admitted after the warning systemwas added).Weinstituted a medication-specific warning system withinCPOE that alerted ordering providers at the point of carewhen ordering a PIM and that advised alternative medi-cation or dose reduction. Themain outcomemeasurewas

the rate of orders for PIMs before and after the warningsystem was deployed.

Results:Themean (SE) rate of orderingmedications thatwere not recommendeddropped from11.56 (0.36) to 9.94(0.12) orders per day after the implementation of a CPOEwarning system (difference, 1.62 [0.33]; P .001), withno evidence that the effect waned over time. There wereno appreciable changes in the rate of ordering medica-tions for which only dose reduction was recommendedor that were not targeted after CPOE implementation.These effects persisted in autoregressive models that ac-counted for secular trends and season (P .001).

Conclusion: Specific alerts embedded into a CPOE sys-tem, used in patients 65 years or older, can decrease thenumber of orders of PIMs quickly and specifically.

Arch Intern Med. 2010;170(15):1331-1336

O LDER PEOPLE ADMITTEDto the hospital are espe-cial ly vulnerable toadverse drug events(ADEs),1 which occur inup to 40% of hospital admissions.2 Ad-verse drug events increase the length ofstay, the cost of caring for patients admit-ted to the hospital, and the risk of death.3

Some medications may predispose vul-nerable older patients to ADEs. Fick et al4

proposed a list of drugs identified by apanelof geriatricmedicine experts as beingmedi-cations that should be avoided in older per-sons. Despite the publication of the Beersmedications, the prescription of poten-tially inappropriate medications (PIMs) toelderly patients remains common.5,6

Up to 60% of ADEs during hospitaliza-tion occur at the time of ordering3,6,7; theremainder occur downstream, during de-livery or omission (not giving a medica-tion as prescribed). Computerized pro-vider order entry (CPOE) systems providean opportunity for intervention to changeprescribing practices before PIMs are or-dered. However, to our knowledge, no

CPOE system has previously been de-scribed that uses a warning system builtaround PIMs in older, hospitalized adults.

The purpose of this study was to de-termine whether the number of orders forPIMs in hospitalized patients 65 years orolder could be decreased using a comput-erizedwarning system linked toCPOE.Westudied the ordering patterns before andafter the implementation of such a sys-tem for 3 groups of drugs: a larger groupof drugs included in the original list ofdrugs compiled by Fick et al according tothe Beers criteria (Beers medications) thatwere flaggednot to be used, a second groupof Beers medications that were flagged tobe used at reduced doses, and a third groupof Beers medications not flagged.

METHODS

PATIENT POPULATION

We studied all inpatients 65 years or older hos-pitalized at a single urban academicmedical cen-ter in Boston, Massachusetts. The hospital pro-videsprimaryand tertiarycare,with621 inpatientbeds and approximately 40000 inpatient admis-

Author Affiliations: Divisionsof Gerontology (Dr Mattison)and General Medicine andPrimary Care (Drs Mattison,Ngo, and Mukamal),Department of Medicine, andDepartment of Clinical SystemsDevelopment (Mr Afonso),Beth Israel Deaconess MedicalCenter, Boston, Massachusetts.

(REPRINTED) ARCH INTERN MED/VOL 170 (NO. 15), AUG 9/23, 2010 WWW.ARCHINTERNMED.COM1331

2010 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ on 03/27/2014

sions annually. The medical centers institutional review boardapproved this study.

CPOE WARNING SYSTEM

The CPOE system at the medical center was developed byprogrammers at the institution and is not commercially avail-

able. All medications prescribed to inpatients are orderedthrough the CPOE system, although dispensed medicationsare tracked through a separate pharmacy program. With hos-pital Pharmacy and Therapeutics Committee feedback, we de-veloped medication-specific alerts that were built into thehospitals CPOE system. Two screen shots illustrating thewarnings encountered when a geriatric precaution is found

Table 1. Specific Beers Medications Targeted and the Warning Received by the Ordering Physician

Drug

Warninga

Explanation for WarningbDegree

of SeverityConditions That Put Patients

at Increased Risk

Not-Recommended MedicationsAmitriptyline hydrochloride Because of its strong anticholinergic and sedation properties,

amitriptyline should be used rarely in the elderly.High Hepatic, cardiovascular, or

neurologic/psychiatric impairmentChlordiazepoxide This drug has a long half-life in elderly patients (often several days),

producing prolonged sedation and increasing the risk of falls andfractures. Short- and intermediate-acting benzodiazepines arepreferred if a benzodiazepine is required. .

High Renal, hepatic, orneurologic/psychiatric impairment

Clonidine Potential for orthostatic hypotension and CNS adverse effects. Low Renal or neurologic/psychiatricimpairment

Clorazepate dipotassium This drug has a long half-life in elderly patients (often several days),producing prolonged sedation and increasing the risk of falls andfractures. Short- and intermediate-acting benzodiazepines arepreferred if a benzodiazepine is required.


Cyclobenzaprine hydrochloride Most muscle relaxants and antispasmodic drugs are poorly toleratedby elderly patients, since these cause anticholinergic side effects,sedation, and weakness. Additionally, their effectiveness at dosestolerated by elderly patients is questionable.

High Renal, hepatic, cardiovascular, orneurologic/psychiatric impairment

Diazepam This drug has a long half-life in elderly patients (often several days),producing prolonged sedation and increasing the risk of falls andfractures. Short- and intermediate-acting benzodiazepines arepreferred if a benzodiazepine is required.

High Hepatic or neurologic/psychiatricimpairment

Diphenhydramine May cause confusion and sedation. Should not be used as a hypnotic,and, when used to treat emergency allergic reactions, it should beused in the smallest possible dose.

High Neurologic/psychiatric impairment

Doxazosin mesylate Potential for hypotension, dry mouth, and urinary problems. Low Renal, cardiovascular, orneurologic/psychiatric impairment

Fluoxetine hydrochloride Long half-life of drug and risk of producing excessive CNS stimulation,sleep disturbances, and increasing agitation. Safer alternatives exist.

High Hepatic or neurologic/psychiatricimpairment

Hydroxyzine All nonprescription and many prescription antihistamines may havepotent anticholinergic properties. Nonanticholinergic antihistaminesare preferred in elderly patients when treating allergic reactions.

High Neurologic/psychiatric impairment

Ketorolac tromethamine Immediate and long-term use should be avoided in older persons,since a significant number have asymptomatic GI pathologicconditions.

High Renal, hepatic, or cardiovascularimpairment

Naproxen Has the potential to produce GI bleeding, renal failure, high bloodpressure, and heart failure.


Oxybutynin Most muscle relaxants and antispasmodic drugs are poorly toleratedby elderly patients, since these cause anticholinergic side effects,sedation, and weakness. Additionally, their effectiveness at dosestolerated by elderly patients is questionable.


Piroxicam Has the potential to produce GI bleeding, renal failure, high bloodpressure, and heart failure.


Propoxyphene Offers few analgesic advantages over acetaminophen, yet has theadverse effects of other narcotic drugs.

Low Renal, hepatic, orneurologic/psychiatric impairment

Thioridazine Greater potential for CNS and extrapyramidal adverse effects. Otherantipsychotic agent might be more appropriate.


Dose-Reduction MedicationsLorazepam Because of increased sensitivity to benzodiazepines in elderly patients,

smaller doses may be as effective as well as safer. Total daily doseshould rarely exceed the suggested maximum of 3 mg.


Ferrous sulfate Doses greater than 325 mg/d do not dramatically increase the amountabsorbed but greatly increase the incidence of constipation.

Low NA

Unflagged MedicationsAmiodarone hydrochloride NA NA NADigoxin NA NA NADisopyramide phosphate NA NA NAIndomethacin NA NA NA

Abbreviations: CNS, central nervous system; GI, gastrointestinal; NA, not applicable.aAll warnings also state, Precaution is necessary for use of [drug name] in geriatric patients. All warnings provide a reference to the updated Beers criteria in

the article by Fick et al.4bThe exact wording of the explanations is given.




are available in a supplementary eAppendix (http://www.archinternmed.com).

From the larger list of PIMs listed in the article by Fick etal,4 we identified 3 primary classes of medications for study apriori: medications that were flagged as not recommended foruse in older patients (not-recommended medications), thoseforwhichonly a reduceddosewas advised (dose-reductionmedi-cations), and those that were not flagged because no safer al-ternative was considered equally efficacious (unflagged medi-cations [amiodarone hydrochloride, digoxin, disopyramidephosphate, and indomethacin]); the last group represented con-trols in our analyses. Table 1 shows the targeted drugs andthe wording of the alerts used. A geriatrician (M.L.P.M.) and apharmacist proposed the specific groups of medications, usingthe literature where possible to support their decisions, and thePharmacy and Therapeutics Committee at the medical center,composed of senior hospital pharmacists and clinicians, re-vised and approved the list.

We did not target 2 other general categories of medicationsfrom the original list of Beers medications: (1) drug classes forwhich individual drugs were not consistently in the formularythroughout the study or that were extremely infrequently usedamong elderly inpatients and (2) drug classes with very broadandheterogeneoususe (eg, nonsteroidal anti-inflammatorydrugsand calcium channel blockers) that were left unflagged tomini-mize the number of warnings encountered by users. We in-cluded these latter 2 classes as controls in secondary analyses.

For all flaggedmedications, theorderingproviderhad theop-tion tobypass thewarning andorder themedication;noprior ap-proval was required. Each time, however, the ordering providerhad to choose a reason. FromMarch 17, 2005, through October1,2006,therewere3possiblereasonsfromwhichthecliniciancouldchoose: (1)Patient stabilizedonregimen;willmonitorappropri-ate drug levels or laboratory values; (2) Interactionnoted, regi-men clinically indicated, will closely monitor; or (3) Other. OnOctober 2, 2006, a fourth choicewas added: (4) Warningnoted,will use smaller dose and monitor for side effects.

Thewarning systemapplied to all patients admitted to thehos-pital who were 65 years or older at the time of the order regard-less of their location within the hospital or admitting service,although meperidine hydrochloride and promethazine hydro-chloride were part of a fixed postanesthesia care unit order setthat was not flagged and thus are not included. There were noother concurrent efforts made to educate providers about medi-cation safety, and thewarnings suggested no specific alternatives.

OUTCOME MEASURES

From June 1 through November 29, 2004, the 6 months be-fore the deployment of the CPOE warning system, the num-ber of orders in hospitalized patients 65 years or older wererecorded for the selected medications. All orders, whether as-needed or standing orders, were included; the hospital does nothave an electronicmedication administration record, and hencewe were not able to record the number or dosage of medica-tions actually given to the patient. We excluded the period be-tween November 30, 2004, and March 16, 2005, the period ofbeta testing of the warning system. We then recorded all or-ders after the warning system was deployed, from March 17,2005, through August 30, 2008.

STATISTICAL ANALYSES

We computed 2 measures of the rate of prescribing of Beersmedications: the daily number of medications in each class di-vided by either the total number of hospitalized patients 65 yearsor older or the number of newly admitted hospitalized pa-

tients 65 years or older each day. Because medications are dif-ferentially more likely to be prescribed on the first hospital day,these denominators represent complementary estimates of thenumber of patients at risk for inappropriate prescriptions.

We first plotted the daily rate of each outcome measureagainst calendar time and fit separate smoothed splines for theperiods before and after implementation. In initial analyses, wecalculated the mean daily rates of each of the 3 classes of drugsbefore and after the warning system was instituted and com-pared thesewith unpaired, 2-tailed t tests. Because the smoothedsplines indicated that the underlying trend of the outcome rateover time was linear, we assumed linearity in time series mod-els and fit autocorrelative regressionmodels that accounted forthe serial correlation in themeasurement errors of the daily out-come rates. These models included calendar time, period (be-fore vs after implementation), the product (or interaction) ofperiod time (ie, change in the secular trend after the imple-mentation), and season. Regression analyses were performedin SAS statistical software, version 9.2 (SAS Institute Inc, Cary,North Carolina), using the PROC AUTOREG procedure.

RESULTS

During the period of study, there was a secular trend inthe number of patients in the hospital 65 years or olderand in the mean number of all orders, resulting in largernumbers of orders over time.

TheFigure shows the temporal trends in the rateof or-dersof the3classesofmedications studied.After thewarn-ing systemwasdeployed, therewas an immediate and sus-taineddecreaseintherateofordersforthenot-recommendedmedications. There was a modest secular trend resultingindecreaseduseofunflaggedmedicationsthatdidnotchangeappreciably after thewarning systemimplementation, andthere was no change in the dose-reduction medications.

In before-and-after comparisons (Table 2), the mean(SE) rate of prescribing not-recommended medicationsdroppedfrom11.56(0.36)to9.94(0.12)ordersperday(dif-ference,1.62[0.33];P .001).Therewasamodestdecreasein the use of unflaggedmedications that was of borderlinestatistical significance, consistent with the observed secu-lar trend,andweobservednochange in therateofprescrib-ingmedications inwhichonlyadosereductionwasadvised.

Autoregressivemodelsyieldedsimilar results.Therewasa highly significant and immediately observed drop in therateofuseofnot-recommendedmedications(P.001),withno change in the secular trend after implementation of theintervention(P=.11).Therewerenosignificant changes intheabsoluterateofprescribingor inthesecular trendofpre-scribing for the other 2 classes of medications after imple-mentation of the intervention in these models.

In secondary analyses, we also examined the rates ofprescribing of all unflagged medications on the originallist of Beers medications rather than the 4 medicationsselected a priori. In autoregressive models, there was nosignificant effect of the intervention on the daily rate ofprescribing in absolute terms (P=.44) or on the seculartrend of prescribing (P=.17).

Among the not-recommendedmedications, the mostcommonly prescribed was diphenhydramine, which ac-counted for approximately one-third of all prescrip-tions in that group before the implementation. Both itsuse and the use of other targeted medications droppedmarkedly after implementation of the warning system,




although we had insufficient power to examine othermedications individually. For example, the daily rate oforders for not-recommended medications per new ad-

mission (SE) dropped by 0.070 (0.008) (P .001) afterimplementation; the corresponding drops were 0.043(0.004) (P .001) for diphenhydramine alone and 0.027(0.006) (P .001) for other targeted medications. Thedrops after implementation were also significant in au-toregressivemodels for diphenhydramine (P .001) andfor other targeted medications (P=.001).

All orders recorded in this study on flagged medica-tions reflect orders in which the ordering provider by-passed the warning; the CPOE does not track prescrip-tions that are started but not completed. In our study,users provided Interaction noted, regimen clinically in-dicated, will closely monitor as the reason for overrid-ing the warning about half the time. Patient stabilizedon regimen; will monitor appropriate drug levels or labo-ratory values was given as the secondmost common rea-son for overriding the warning (Table 3). A third op-tion that indicated the prescriber intended to use a lowdose was instituted onOctober 2, 2006; as intended, thisoptionwas usedmore frequently for dose-reductionmedi-cations (19%) than for not-recommended medications(13%; P .001 for heterogeneity across categories).

COMMENT

In this quasi-experimental study of a large urban medicalcenter, the rate of orders for PIMs in older patients wasmarkedly decreased by the use of a CPOEwarning systemtargeting a subset of Beers medications. The interventionshowed no signs of users growingweary of repeatedwarn-ings and ignoring them (alert fatigue), and other medi-cations that were not flagged or flagged only for dose ad-justment continued to be prescribed at unchanged rates.

SPECIFIC FEATURES AND FINDINGS

After our alert system was implemented, the rate of or-dering of the targeted medications declined immedi-ately in the study population. Others have found similarresults in the outpatient setting.8,9 This may reflect ourrestriction to only a subset of medications with legiti-mate alternatives in a vulnerable patient population.10 Inthis regard, the specificity and immediacy of the drop inthe use of flagged medications is reassuring and sug-gests that local systems can be effectively tailored tomeetlocal standards of care. Indeed, previous research sug-gests that drug alerts created in internally developedCPOEsystems tailor-made to an individual institution or ser-vice can reducemedication errors or orders for PIMs.11-13

We did not observe a substantial learning effect, inwhich onemight hope to see a further reduction over timein the rate of ordering the PIMs, perhaps because of theannual turnover of house staff who order medications.On the other hand, the effect of our warning system ap-peared to be durable over time with no sign that usersgrew weary of repeated warnings.

CPOE IN AN AGING POPULATION

Many clinicians have not received formal education aboutthe unique medical needs of elderly patients and, de-

0.5

0.4

0.3

0.2

0.1

0.0

6/1/04 11/29/04 3/17/05 8/30/08

A

Preintervention periodPostintervention period

0.5

0.4

0.3

0.2

0.1

0.0

Daily

Rat

e of

Med

icat

ion

Orde

rs

B

0.5

0.4

0.3

0.2

0.1

0.0

C

Figure. Daily rate of medication orders among hospitalized older patients(65 years) for each of 3 medication classes before and after implementationof a computerized provider order entry alert system. The daily rate wasdetermined by calculating the number of medications ordered on a daily basisfor patients 65 years or older divided by the number of older adults admitted onthe corresponding day. The x-axis shows the beginning and ending dates of theobservation period before implementation (June 1, 2004, through November29, 2004), the date that the alert system was implemented (March 17, 2005),and the date that follow-up concluded (August 30, 2008). A, Medications thatwere flagged by the alert system as being not recommended to order(not-recommended medications). B, Medications that were flagged with arecommendation for dose reduction (dose-reduction medications).C, Medications that were not flagged (unflagged medications). The lines indicatesmoothed splines fit separately for the preintervention and postinterventionperiods. We found a significant change in the rate of ordering after theintervention for the not-recommended medications (P .001).




spite the fact that more people are living longer, thereare no geriatric medicine-specific performance stan-dards for US medical students.14 This may explain whyPIMs continue to be prescribed for hospitalized patients65 years or older and provide a rich target for CPOE in-tervention.15

Understanding this limit in training,we created aCPOEwarning system to decrease the use of PIMs in older pa-tients. These CPOE systems change the way cliniciansordermedications and provide newopportunities to guidebehavior. Although less than 10% of US hospitals cur-rently use CPOE, the Institute of Medicine report call-ing for universal adoption of CPOE heralds an increas-ing reliance on this technology.16

Designing CPOE systems to shape best practice is anevolving field. Research suggests that CPOE systemswith-out any decision support aroundmedication ordering areassociated with high rates of ADEs.7 Furthermore, gen-eral drug alerts within CPOE systems are overridden upto 90% of the time.17 Initial efforts to reduce ADEs withCPOE systems have focused on reducing medication er-rors, such as drug allergy and drug-drug interac-tions.11,18 Our results suggest that specific drug alerts formedications that place older patients at particular riskfor ADEs could be an especially attractive addition to suchsystems.4,19,20

NEXT STEPS AND IMPLICATIONS

Our study and some studies of outpatient drug warningsystems8,9 have found a clear reduction in the use of PIMs.

As such, our findingsby showing that these drugs areindeed amenable to targeted change by a straightfor-ward ordering systemprovide the first necessary stepin determining whether reducing the use of these medi-cationwill ultimately improve patient outcomes. It is notyet clear whether any differences in patient outcomes canbe attributed to this change in behavior, but our resultsprovide optimism that this important research questioncan be addressed in the near future.

As CPOE is more widely adopted, it seems likely thatmost institutions will rely on commercially available(rather than internally developed) systems. Such sys-temswill rarely be sufficientlymalleable to allow the fine-tuned and circumscribed type of intervention we de-scribe herein. As such, designing commercially availableCPOE systems to guide clinicians at the local level to ad-here to the best care is challenging. To be most effec-tive, systems shouldminimize generalizedwarnings and,like this warning system, use focused alerts to target spe-cific patient populations where alternate treatmentexists. We encourage developers of commercial CPOEsystems to build in the flexibility to implement point-of-care warnings appropriate to local circumstances.

LIMITATIONS

There are several limitations to our data. For lorazepamand ferrous sulfate, the warning advised a dose reduc-tion. The warning did not advise against the use of thesemedications and, consistent with this advice, the rate ofordering thesemedications did not change after the imple-

Table 2. Number of Orders per Day in 3 Groups of Drugs Before and After Start of CPOE Warning Systema

Not-RecommendedMedications

PValue

Dose-ReductionMedications

PValue

UnflaggedMedications

PValueBefore After Before After Before After

Orders per day 11.56 (0.36) 9.94 (0.12) 6.09 (0.20) 6.87 (0.08) 5.40 (0.20) 5.55 (0.07)Orders per total No.

of patients per day0.070 (0.002) 0.054 (0.001) .001 0.037 (0.001) 0.037 (0.001) .71 0.033 (0.001) 0.030 (0.001) .03

Orders per No. of newpatients per day

0.333 (0.001) 0.263 (0.003) .001 0.182 (0.006) 0.186 (0.002) .51 0.158 (0.005) 0.148 (0.002) .08

Abbreviation: CPOE, computerized provider order entry.aData are presented as mean (SE) unless otherwise indicated.

Table 3. Reasons for Overrides of Not-Recommended Medications and Dose-Reduction Medicationsa

Reason

No. (%)b

Not-Recommended Medications Dose-Reduction Medications

3/1/05 to 10/1/06 10/2/06 to 8/30/08 3/1/05 to 10/1/06 10/2/06 to 8/30/08

Patient stabilized on regimen; will monitor appropriate drug levelsor laboratory values

1863 (38) 2673 (40) 1147 (33) 1564 (31)

Interaction noted, regimen clinically indicated, will closely monitor 2828 (58) 3015 (45) 2204 (62) 2326 (46)Warning noted, will use smaller dose and monitor for side effects NA 859 (13) NA 948 (19)Other 215 (4) 214 (3) 177 (5) 173 (3)

Abbreviation: NA, not applicable.aThree override options were available until October 1, 2006; 4 were available October 2, 2006, and afterward.bBecause of rounding, percentages may not total 100.




mentation of the warning systemsuggesting that thesystem could provide adequate specificity. However, welack information on the dose of medication prescribedand hence cannot be certain whether the targeted dosereductions were achieved.

Another limitation of this study is its generalizabil-ity. Our drug warning system was used at an academicmedical center where medical trainees or physician ex-tenders ordermostmedications.Wedonot knowwhethera similar result would be seen in a system where attend-ing physicians place most of the orders or in institu-tionswithout a firmly entrenched andmultipurposeCPOEsystem.We also lacked the ability to determine whetherADEs were prevented by the use of this warning system.

Similarly, we are only able to comment onmedicationsordered. Although we recorded all orders, including as-needed and standing orders, we lacked the ability to as-certain the number of medications actually given to pa-tients because our hospital does not have an electronicmedical administration record. Nonetheless, all medica-tions actually administered at themedical center necessar-ily were captured as orders, so our findings accurately re-flect a decline in the number of patients exposed to a subsetof potentially problematic medications.

Finally,without detailed clinical record review,we can-not determine whether the medications that were or-dered were clinically required. One important area of fu-ture study is a better understanding of the scenarios inwhich it is clinically appropriate and reasonable to pre-scribe the Beers medications even to older adults.

In summary, we found that a CPOE system with spe-cific, targeted, and straightforward warnings can dra-matically yet selectively reduce the prescription of PIMsin vulnerable hospitalized older patients. Such systemscan produce rapid and clinically significant change whileleaving unchanged the rate of prescribing unflaggedmedi-cations. Thismay represent a tool for improving the safetyof hospitalized older adults.

Accepted for Publication: January 18, 2010.Correspondence:Melissa L. P. Mattison, MD, Divisionsof Gerontology andGeneral Medicine and Primary Care,Department of Medicine, Palmer-Baker Span 2, Beth Is-rael Deaconess Medical Center, 330 Brookline Ave, Bos-ton, MA 02215 ([email protected]).Author Contributions: Study concept and design: Matti-son and Mukamal. Acquisition of data: Afonso. Analysisand interpretation of data:Mattison, Ngo, and Mukamal.Drafting of the manuscript: Mattison, Ngo, and Muka-mal. Critical revision of the manuscript for important in-tellectual content:Mattison, Afonso, Ngo, andMukamal.Statistical analysis:Ngo.Administrative, technical, andma-terial support: Mukamal. Study supervision: Mukamal.Financial Disclosure: None reported.Funding/Support:This studywas funded in part by grant1 UL1 RR025758-01, Harvard Clinical and Transla-tional Science Center, from the National Center for Re-search Resources (Dr Ngo).

Online-Only Material: A supplementary eAppendix isavailable at http://www.archinternmed.com.Additional Contributions: Rachel Murkofsky, MD, andLisa Saubermann, PharmD, BCPS, contributed to thedevelopment of the original alert system.

REFERENCES

1. Passarelli MC, Jacob-Filho W, Figueras A. Adverse drug reactions in an elderlyhospitalised population: inappropriate prescription is a leading cause.Drugs Aging.2005;22(9):767-777.

2. Fattinger K, Roos M, Vergeres P, et al. Epidemiology of drug exposure and ad-verse drug reactions in two Swiss departments of internal medicine. Br J ClinPharmacol. 2000;49(2):158-167.

3. Bates DW, Spell N, Cullen DJ, et al; Adverse Drug Events Prevention Study Group.The costs of adverse drug events in hospitalized patients. JAMA. 1997;277(4):307-311.

4. Fick DM, Cooper JW, Wade WE, Waller JL, Maclean JR, Beers MH. Updatingthe Beers criteria for potentially inappropriate medication use in older adults:results of a US consensus panel of experts [published correction appears inArch Intern Med. 2004;164(3):298]. Arch Intern Med. 2003;163(22):2716-2724.

5. Onder G, Landi F, Cesari M, Gambassi G, Carbonin P, Bernabei R; Investigatorsof the GIFA Study. Inappropriate medication use among hospitalized older adultsin Italy: results from the Italian Group of Pharmacoepidemiology in the Elderly.Eur J Clin Pharmacol. 2003;59(2):157-162.

6. Gurwitz JH, Field TS, Avorn J, et al. Incidence and preventability of adverse drugevents in nursing homes. Am J Med. 2000;109(2):87-94.

7. Nebeker JR, Hoffman JM, Weir CR, Bennett CL, Hurdle JF. High rates of adversedrug events in a highly computerized hospital. Arch Intern Med. 2005;165(10):1111-1116.

8. Steele AW, Eisert S, Witter J, et al. The effect of automated alerts on providerordering behavior in an outpatient setting. PLoS Med. 2005;2(9):e255. doi:10.1371/journal.pmed.0020255.

9. Smith DH, Perrin N, Feldstein A, et al. The impact of prescribing safety alerts forelderly persons in an electronic medical record: an interrupted time seriesevaluation. Arch Intern Med. 2006;166(10):1098-1104.

10. Lin CP, Payne TH, Nichol WP, Hoey PJ, Anderson CL, Gennari JH. Evaluatingclinical decision support systems: monitoring CPOE order check override ratesin the Department of Veterans Affairs computerized patient record system. J AmMed Inform Assoc. 2008;15(5):620-626.

11. Bates DW, Teich JM, Lee J, et al. The impact of computerized physician order entryon medication error prevention. J Am Med Inform Assoc. 1999;6(4):313-321.

12. Terrell KM, Perkins AJ, Dexter PR, Hui SL, Callahan CM, Miller DK. Computer-ized decision support to reduce potentially inappropriate prescribing to older emer-gency department patients: a randomized, controlled trial. J AmGeriatr Soc. 2009;57(8):1388-1394.

13. Simon SR, Smith DH, Feldstein AC, et al. Computerized prescribing alerts andgroup academic detailing to reduce the use of potentially inappropriate medica-tions in older people. J Am Geriatr Soc. 2006;54(6):963-968.

14. Leipzig RM, Granville L, Simpson D, Anderson MB, Sauvigne K, Soriano RP.Keeping granny safe on July 1: a consensus on minimum geriatrics competen-cies for graduating medical students. Acad Med. 2009;84(5):604-610.

15. Rothberg MB, Pekow PS, Liu F, et al. Potentially inappropriate medication use inhospitalized elders. J Hosp Med. 2008;3(2):91-102.

16. Whiting SO, Gale A. Computerized physician order entry usage in North America:the doctor is in. Healthc Q. 2008;11(3):94-97.

17. van der Sijs H, Mulder A, van Gelder T, Aarts J, Berg M, Vulto A. Drug safetyalert generation and overriding in a large Dutch university medical centre. Phar-macoepidemiol Drug Saf. 2009;18(10):941-947.

18. Bobb A, Gleason K, Husch M, Feinglass J, Yarnold PR, Noskin GA. The epide-miology of prescribing errors: the potential impact of computerized prescriberorder entry. Arch Intern Med. 2004;164(7):785-792.

19. Bates DW, Miller EB, Cullen DJ, et al; ADE Prevention Study Group. Patient riskfactors for adverse drug events in hospitalized patients. Arch Intern Med. 1999;159(21):2553-2560.

20. EvansRS, Lloyd JF, StoddardGJ,Nebeker JR,SamoreMH.Risk factors for adversedrug events: a 10-year analysis. Ann Pharmacother. 2005;39(7-8):1161-1168.




2 preventing potentially inappropiate medication use in hospiutalized older patients

Documents