Potentially Inappropriate Prescribing in Elderly Veterans: Are We Using the Wrong Drug, Wrong Dose, or Wrong Duration?

Mary Jo V. Pugh, PhD, wz B. Graeme Fincke, MD,wz Arlene S. Bierman, MD, MS, Bei-Hung Chang, ScD,wz Amy K. Rosen, PhD,wz Francesca E. Cunningham, PharmD, k z Megan E. Amuan, MPH,w Muriel L. Burk, PharmD, k z and Dan R. Berlowitz, MD, MPHwz

OBJECTIVES: To identify the extent of inappropriate prescribing using criteria for proper use developed by the Agency for Healthcare Research and Quality (AHRQ) and dose-limitation criteria dened by Beers, as well as to describe duration of use and patient characteristics associated with inappropriate prescribing for older people. DESIGN: Retrospective national Veterans Health Administration (VA) administrative database analysis. SETTING: VA outpatient facilities during scal year 2000 (FY00). PARTICIPANTS: Veterans aged 65 and older having at least one VA outpatient visit in FY00 (N 5 1,265,434). MEASUREMENTS: Operational definitions of appropriate use were developed based on recommendations of an expert panel convened by the AHRQ (Zhan criteria). Inappropriate use was identied based on these criteria and inappropriate use of drugs per Beers criteria for dose-limitations in older people. Furthermore, duration of use and patient characteristics associated with inappropriate use were described. RESULTS: After adjusting for diagnoses, dose, and duration, inappropriate prescribing decreased from 33% to 23%. Exposure to inappropriate drugs was prolonged. Pain relievers, benzodiazepines, antidepressants, and musculoskeletal agents constituted 61% of inappropriate prescribing. Whites, patients with psychiatric comorbidities, and patients receiving more medications were most likely to receive inappropriate drugs. Women were more likely to

receive Zhan criteria drugs; men were more likely to receive dose-limited drugs CONCLUSION: For the most part, the Zhan criteria did not explain inappropriate prescribing, which includes problems related to dose and duration of prescriptions. Interventions targeted at prescriptions for pain relievers, benzodiazepines, antidepressants, and musculoskeletal agents may dramatically decrease inappropriate prescribing and improve patient outcomes. J Am Geriatr Soc 53:12821289, 2005. Key words: pharmacoepidemiology; patient safety; prescribing; appropriateness; geriatrics

From the Veterans Evidence-based Research Dissemination and Implementation Center, Audie L. Murphy Division /South Texas Veterans Health Care System, San Antonio, Texas; wDepartment of Veterans Affairs Hospital, Bedford, Massachusetts; zSchool of Public Health, Boston University, Boston, Massachusetts; Inner City Health Research Unit and University of Toronto, St. Michaels Hospital, Toronto, Ontario, Canada; k Pharmacy Benets Management Strategic Health Group, Department of Veterans Affairs, Hines, Illinois; and zCollege of Pharmacy, University of Illinois at Chicago, Chicago, Illinois. Address correspondence to Mary Jo V. Pugh, PhD, VERDICT, STVHCS/ ALMD, 7400 Merton Minter Blvd (11C6), San Antonio, TX 78229. E-mail: pughm@uthscsa.edu DOI: 10.1111/j.1532-5415.2005.53402.x

tudies using the explicit, expert consensusbased Beers criteria1,2 consistently nd that 20% to 30% of community-dwelling older Americans receive drugs identied as potentially inappropriate.312 Many of these drugs are known to cause confusion, altered consciousness, and increased propensity to fall, and they may lead to diminished independence and death. Thus use may represent a serious threat to patient safety.1316 Data limitations have restricted analyses of potentially inappropriate prescribing in the elderly (PIPE) primarily to drugs considered inappropriate regardless of diagnosis (disease-independent) in the Beers criteria and excluded examination of drugs identied as having dose limitations in older people (dose-limited). Controversy thus ensued, with clinicians arguing that the Beers criteria are too broad and that some drugs are appropriate for specific patients in certain circumstances. To address this issue, an expert consensus panel was convened at the Agency for Healthcare Research and Quality (AHRQ) to identify appropriate indications for use of the disease-independent Beers drugs.6 The group identied 11 drugs that should always be avoided (always avoid), eight drugs that are appropriate on rare occasions (rarely appropriate), and 14 drugs that have some indications but are often misused (some indications). The panel further identied situations in which use of rarely appropriate and some-indications drugs are proper (hereafter, Zhan criteria; information available on request). Using data from the

JAGS 53:12821289, 2005 r 2005 by the American Geriatrics Society

0002-8614/05/$15.00

JAGS

AUGUST 2005VOL. 53, NO. 8

WRONG DRUG, WRONG DOSE, WRONG DURATION

1283

1996 Medical Expenditure Panel Survey, they found that 21.3% of older Americans received at least one potentially inappropriate drug, and 2.6% received an always-avoid drug,6 but data limitations prevented determination of the extent to which observed use may be appropriate, and as with previous research, they also prevented assessment of appropriateness of Beers dose-limited drugs. Data limitations have also restricted the scope of research identifying patient characteristics associated with PIPE. National databases often restrict the number of diagnoses or medications included in the le.7,17 Smaller studies including richer clinical data have been restricted geographically to small regions of the country.18,19 Regional variation in prescribing20,21 may explain differences in the most commonly used PIPE drugs across studies. These variations may then lead to different assessments of patient characteristics associated with PIPE. Moreover, extant research has not examined patient characteristics associated with use of dose-limited drugs. Using comprehensive national databases from the Veterans Health Administration (VA)Fthe largest integrated healthcare system in the United StatesFthe VA Study of Potentially Inappropriate Prescribing in the Elderly described the multifaceted nature of PIPE. The availability of data describing diagnoses, dosage information, and the duration of prescriptions allowed for the development of algorithms that operationalize the Zhan criteria.6 Proper use of PIPE drugs was then identied, the extent to which older veterans received drugs exceeding dose recommendations for older people was examined,1 and the duration of inappropriate prescriptions was described. Furthermore, the comprehensive clinical data available allowed a more complete assessment of patient characteristics associated with use of Zhan criteria and dose-limited drugs.

in which 50% or fewer older veterans received medications from the VA. These numbers are inconsistent with other networks, where 75% to 90% of patients received medications. Thus, patients from these networks were excluded. Institutional review boards of the Bedford and Hines VA medical centers approved this study.

METHODS Data Sources and Population This study was conducted with national VA inpatient and outpatient administrative and outpatient pharmacy data (pharmacy benets management). Administrative data included demographic and diagnostic information (International Classication of Diseases, Ninth Revision (ICD-9) codes22) for each visit. Because there are up to 10 ICD-9 codes per visit, these administrative data have been found to be accurate in capturing active problems.2325 Pharmacy data include the drug and dose prescribed, number of days supply, and prescription duration. Veterans aged 65 and older on October 1, 1999, and who had at least one outpatient clinic visit during FY00 (October 1, 1999, to September 30, 2000) were included. Encrypted patient identiers were linked to the national pharmacy database to identify drug information for each individual. Pharmacy data for drugs listed in Table 1 for FY99 and FY00 were extracted. Consistent with previous studies,4,6,18 prevalence was assessed during a 1-year period (FY00). Data from FY99 were used to assess appropriateness and duration of prescriptions. To assure the completeness of pharmacy data, all patients who received at least one medication in the national PBM database were identied. There were three geographically contiguous networks (in the northeast United States)

Identifying Appropriate Indications The study team developed operational definitions for appropriate indications using transcripts from the AHRQ panel (AB, DB were panel members). These definitions were framed as diagnoses to facilitate matching them to ICD-9 codes (available upon request). This was straightforward for diagnoses such as gout but more difcult for diagnoses such as neuropathic pain, in which the condition can be associated with different diagnoses. One team member (BGF) developed a list of ICD-9 codes for each indication using key words to electronically search the index of the ICD-9 coding manual. In the most complex caseFneuropathic painFkey words used were neuropathy, neuralgia, neuritis, pain, neuropathic, radiculitis, radiculopathy, compression, polyneuropathy, mononeuropathy, myelopathy, myelitis, sympathetic dystrophy, nerve root plexus, intervertebral disc, spondylosis, spondylitis, zoster, sciatica, and spinal stenosis. The resulting ICD-9 codes were compared with coding definitions in the manual, related codes were reviewed for relevance, and new codes were added if appropriate. Finally, to identify additional codes, key words were entered into the problem list module of the VA computerized patient record, which uses the National Library of Medicine metathesaurus to map key words to ICD-9 codes. Clinical research team members reviewed the list of codes for each indication, determined which should be kept or discarded, and sought missing codes. Members agreed about the disposition for 71% of the codes. The group discussed the remaining codes until consensus was obtained. A neurologist was consulted to determine whether codes for neuropathy, back pain, and muscle spasm were appropriate and to ensure that appropriate codes had not been omitted. Because of potential coding variation for back pain and alcohol withdrawal, broad and narrow interpretations were identied. For instance, narrow criteria for back pain included acute conditions such as sprains and strains, whereas broad criteria also included chronic conditions such as disorders of the lumbar region. Using the most liberal interpretation (e.g., ICD-9 codes for broad criteria if relevant), patients were classied as having appropriate use if codes for these indications were present; others were categorized as having inappropriate use. In addition to diagnostic criteria, the AHRQ panel6 recommended a maximum duration of 14 days for longacting benzodiazepines, antispasmodics, and antihistamines. Use of methyldopa, reserpine, and disopyramide was considered appropriate if it was not a new prescription and if the patient had no adverse effects. Because a 30-day equivalent is used to compare usage patterns in the VA prescription database, use of a single prescription for 30 days or less was classied as appropriate. Prior use could only be ascertained for those with medication data from the previous year; consequently, appropriate use of methyldopa, reserpine, and disopyramide was assessed only for

1284

PUGH ET AL.

AUGUST 2005VOL. 53, NO. 8

JAGS

patients who also received care in the VA during FY99. Because whether patients on methyldopa, reserpine, or disopyramide experienced adverse effects could not be ascertained, the liberal assumption that continued use reected lack of adverse effects was used.

Dose-Limited Drugs The Beers criteria identify digoxin and short-acting benzodiazepines/hypnotics as having maximum geriatric doses lower than usual adult doses (Table 1).1 Average daily doses were calculated by multiplying the strength of the medica-

Table 1. Prevalence and Duration of Inappropriate Prescribing (N 5 1,265,434)

Prevalence of Inappropriate Medications Variable Any always-avoid Barbiturates Flurazepam Meprobamate Chlorpropamide Meperidine Pentazocine Trimethobenzamide Belladonna alkaloids Dicyclomine Hyoscyamine Propantheline Any rarely appropriate Chlordiazepoxide Diazepam Propoxyphene Carisoprodol Chlorzoxazone Cyclobenzaprine Metaxalone Methocarbamol Any some indications Amitriptyline Doxepin Indomethacin Dipyridamole Ticlopidine Methyldopa Reserpine Disopyramide Oxybutynin Chlorpheniramine Cyproheptadine Diphenhydramine Hydroxyzine Promethazine Any dose-limited (maximum dose) Digoxin (0.125 mg, except in atrial brillation) Lorazepam (3 mg) Oxazepam (60 mg) Alprazolam (2 mg) Temazepam (15 mg) Zolpidem (5 mg) Triazolam (0.25 mg)

Inappropriate After Adjusting for Diagnosis, Dose, and Duration %

Prescription Received in FY99 and FY00

0.8 o0.1 o0.1 o0.1 0.1 0.1 o0.1 o0.1 o0.1 0.4 0.1 o0.1 8.0 0.4 1.2 4.1 0.1 0.1 1.2 o0.1 1.6 13.3 2.6 0.6 1.4 0.4 0.4 o0.1z o0.1z o0.1z 1.6 1.8 0.2 3.0 2.7 0.2 4.6 3.4 0.1 o0.1 0.1 1.0 0.2 o0.1

w w w w w w w w w w w w

70.6z 78.8z 82.4z 96.1z 13.1 56.5 17.0 60.4 48.2k 42.9 67.3k 80.0z 72.6z 51.9 59.4k 43.0 42.9 28.6 56.7k 69.5z 75.7z 58.1 78.0z 89.4z
zk z z zk

97.0 86.0
w

86.6 77.2 72.6 75.0 82.7 75.8 82.4 83.2 81.3

w

10.5 26.3 13.3 74.2 91.9 95.9 94.1 94.6 40.4 33.5 7.4 2.6 6.0 38.5 56.3 11.4

53.0k 51.4k 44.5k 51.3k 56.6k 43.0k 85.7z 73.3z 83.3z 86.3z 64.8z 45.6k 81.7z

Butabarbital, secobarbital, pentobarbital. No appropriate indications identied. z Assessed only for those who also received care in the Veterans Health Administration in FY99 and FY00 (n 5 971,607). New use is dened as inappropriate; therefore, no patients had previous use. Days supply: k 90179, z 180.

JAGS

AUGUST 2005VOL. 53, NO. 8

WRONG DRUG, WRONG DOSE, WRONG DURATION

1285

tion (mg per pill) by the total pills received and dividing by the total day supply.

Inappropriate Use Patients who received Zhan-criteria or dose-limited drugs were rst identied as having potentially inappropriate drug use. For the purposes of this study, patients who received these drugs without indications of proper use based on diagnosis, dose, or duration were classied as having inappropriate use. Duration Duration of use was determined only for patients who received drugs inappropriately. First, persistent use of these drugs was assessed by identifying individuals who received each drug during both FY99 and FY00. Second, the number of days of medication received for each drug in FY00 was summed to create the total day supply. To ensure that all patients were eligible during the entire study period, individuals who rst received care after September 30, 1999, were excluded from this aspect of the analysis. Patient Characteristics Associated with PIPE Prior research identied demographic and clinical characteristics as significant predictors of potentially inappropriate prescribing;46,9,18,26 therefore, these attributes were included in models predicting inappropriate prescribing. Demographic information (age, sex, race) was obtained from inpatient and outpatient data (FY98FY00). Missing race data were supplemented using self-reported race from the 1999 Large Health Survey of Veterans, a nationally representative survey of VA enrollees (July 1, 1999, to January 1, 2000; details elsewhere.)27 Race categories included white, black, Hispanic, other, and unknown. Hispanic patients include all patients of Latino or Hispanic origin, including Mexican Americans. It was hypothesized that the youngest veterans (aged 6569) would be most likely to receive inappropriate drugs because there is greater variation in health status in this group.28 It was also hypothesized that the oldest veterans ( 85) would be least likely to receive inappropriate drugs because providers would become increasingly cautious with these generally frail patients. Thus veterans were categorized into three groups: aged 65 to 69, 70 to 84, and 85 and older. To ensure that these groups were suitable, the distribution of inappropriate prescribing in the 70- to 84-year-old group was examined for within-group variation by age. Clinical characteristics of the patient population were obtained from inpatient and outpatient administrative data (FY98FY00) and PBM data (FY00). Because psychiatric and physical conditions may be differentially associated with inappropriate prescribing for older people, Selims27 psychiatric and physical comorbidity indices developed for the veteran population were used to assess disease burden. The psychiatric comorbidity index consists of six groups of psychiatric diagnoses, and the physical comorbidity index consists of 30 chronic physical diseases. To assure the identication of relevant diagnoses, comorbidity status was assessed for 3 years (FY98FY00). In addition to diagnostic information, clinical status was assessed by counting the number of unique oral, topical, and injectable medications

received by the patient in FY00.29 Use of a single medication at different doses was considered one unique medication. Finally, the number of outpatient clinic visits was counted for each patient in FY00 as a proxy for the number of opportunities available for prescribing. Categorical variables were developed for physical and psychiatric comorbidities, the number of medications, and outpatient clinic visits based on the variable distributions. Physical comorbidities were divided into ve groups (0, 12, 35, 67, and 8 comorbidities), psychiatric comorbidities were divided into three groups (0, 1, and 42 different diagnoses), and number of medications (13, 46, 79, and 10) and outpatient clinic visits (12, 34, 59, 10) were divided into four groups.

Analysis This study rst assessed the prevalence of any use of potentially inappropriate drugs. The Zhan and Beers doselimited criteria were then used to identify proper use of PIPE drugs, and the duration of inappropriate drug use was described. Finally, for those who received at least one medication during the year, logistic regression analyses were used to identify patient characteristics associated with inappropriate use. All patient demographic and clinical characteristics described above were included in the model based on a priori hypotheses, and the C-statistic was used to assess general model t.30 Collinearity between variables included in analyses was also examined. Because patient characteristics in bivariate analyses differed for the Zhan criteria and dose-limited drugs, separate analyses were conducted for these groups. All analyses were conducted using SAS 8.2 (SAS Institute, Inc., Cary, NC) RESULTS This cohort of older veterans (N 5 1,265,434) was predominantly male (98%), with a mean age standard deviation of 73.5 5.6. Of those for whom race data were available (72%), 81% were white, 12% black, 6% Hispanic, and 1% other. Eighty-three percent (n 5 1,053,652) of the cohort received at least one medication in FY00, and 77% received VA care in both FY99 and FY00 (n 5 971,607). Prevalence: Potentially Inappropriate Prescribing In this cohort, 0.8% received one or more always-avoid drugs, 8.9% received at least one rarely appropriate drug, and 15.5% received at least one some-indications drug. Propoxyphene, amitriptyline, diphenhydramine, and hydroxyzine were each used by more than 2.5% of older veterans. In addition, 16.1% received at least one dose-limited drug, with digoxin being most commonly used (10% of cohort). Combining all sources, 33% of older veterans received a potentially inappropriate medication. Prevalence: Inappropriate Prescribing After adjusting for the Zhan criteria, inappropriate use decreased from 8.9% to 8.0% for rarely appropriate drugs and from 15.5% to 13.3% for some-indications drugs. With the exception of medications with previous use as the only criterion, inappropriate use was still high for most medications

1286

PUGH ET AL.

AUGUST 2005VOL. 53, NO. 8

JAGS

(e.g., chlordiazepoxide, 96.9%) after adjustment (Table 1). Of patients receiving dose-limited drugs, 28.7% (4.6% of the cohort) received inappropriate doses. The median of the average daily dose was twice the recommended daily dose for digoxin (0.25 mg; recommended 0.125 mg without diagnosis of atrial brillation) temazepam (30 mg; recommended 15 mg), zolpidem (10 mg; recommended 5 mg), and triazolam (0.5. mg; recommended 0.25 mg). Including Zhan-criteria and dose-limited drugs, 23% received at least one inappropriate drug. Of those, 17.5% received two, and 4.4% received three or more during the year (4% and 1% of the overall cohort, respectively). Pain relievers, musculoskeletal agents, antidepressants, and benzodiazepines/hypnotics constituted 61% of all inappropriate drug use.

Duration Table 1 also illustrates that patients received these prescriptions for long durations. Patients receiving drugs for chronic disease management (e.g., chlorpropamide, ticlopidine, dipyridamole, digoxin) and benzodiazepines (e.g., chlordiazepoxide, diazepam, oxazepam, alprazolam, triazolam) were most likely to receive prescriptions during both years. Similar results were found by examining the days supply received in FY00. Patients receiving drugs used for chronic disease/symptom management (chlorpropamide, amitriptyline, doxepin, dipyridamole, ticlopidine, methyldopa, reserpine, disopyramide, oxybutynin, and digoxin) received a median 210 days supply. Patients receiving benzodiazepines received a median 180 days supply, and patients receiving drugs used to treat conditions with acute exacerbations (e.g., carisoprodol, diphenhydramine, methocarbamol) received a median 90 days supply. Patient Characteristics Associated with PIPE Characteristics of patients who received any inappropriate Zhan-criteria drug, any dose-limited drug, and no inappro-

priate drugs are shown in Table 2. Bivariate analyses suggest that, in general, those aged 65 to 69 were most likely to receive an inappropriate drug, and those aged 85 and older were least likely to receive an inappropriate drug. Although women were more likely to receive inappropriate Zhan drugs, they were less likely to receive dose-limited drugs. Whites were more likely to receive all inappropriate drugs than blacks and Hispanics. Clinical characteristics of patients receiving inappropriate drugs suggest that patients with the greatest disease burden were most likely to receive inappropriate drugs. Results of multivariable logistic regression models for Zhan-criteria and dose-limited drugs are shown in Table 3. C-statistics indicated adequate model t (Zhan criteria c 5 0.74; dose-limited c 5 0.73). Although both models were statistically significant, they predicted relatively little variance in inappropriate prescribing (Zhan criteria pseudo coefcient of determination (R2) 5 0.18; dose-limited drug pseudo R2 5 0.08). Consistent with previous research,6,9,10 the number of unique medications prescribed was the strongest predictor of inappropriate prescribing. Psychiatric comorbidity was another strong predictor of inappropriate prescribing. Subjects having multiple psychiatric diagnoses were more likely to receive inappropriate medications than those with a single diagnosis. In addition, a greater number of outpatient visits was associated with a greater likelihood of receiving Zhan-criteria drugs but a lower likelihood of receiving dose-limited drugs. Differences were also found in demographic characteristics between Zhan-criteria and dose-limited drugs. First, the youngest old were more likely to receive Zhan-criteria drugs than the oldest old, but differences between these groups were not significant for dose-limited drugs. Second, although women were more likely to receive inappropriate Zhan-criteria drugs, they were less likely to receive inappropriate dose-limited drugs. Finally, Hispanics were more likely than whites to receive Zhan-criteria drugs, and they were less likely to receive dose-limited drugs.

Table 2. Patient Characteristics by Inappropriate Prescribing Status

Any Inappropriate Zhan Criteria Drug (n 5 249,566) 73.2 5.4 23.6 29.6 26.4 25.5 25.3 21.8 15.8 0.49 0.87 5.1 2.9 10.7 5.6 9.8 9.5 Inappropriate Dose-Limited Drug (n 5 40,677) 73.7 5.3 3.9 2.5 4.0 2.4 2.9 2.9 4.3 0.38 0.79 4.3 2.8 9.3 4.7 6.7 7.7 No Inappropriate Drug (n 5 763,409) 73.4 5.5 72.5 67.9 69.6 72.2 71.9 75.4 79.9 0.24 0.62 3.7 2.5 6.4 4.2 6.5 7.2

Variable Age, mean SD Sex, % Male Female Race, % White Black Hispanic Other Unknown Psychiatric comorbidity, mean SD Physical comorbidities, mean SD Unique drugs, mean SD Outpatient visits, mean SD

Note: All comparisons with the no-inappropriate-drug group significant; Po.001.

JAGS

AUGUST 2005VOL. 53, NO. 8

WRONG DRUG, WRONG DOSE, WRONG DURATION

1287

Table 3. Logistic Regression Model Predicting Inappropriate Medication Use

Any Inappropriate Zhan Criteria Drug Versus No Inappropriate Characteristic Inappropriate Dose-Limited Drug Versus No Inappropriate

Odds Ratio (95% Condence Interval)

Demographic Age 6569 1.3 (1.21.3) 7074 1.1 (1.11.2) 85 1.0 F Sex Female 1.3 (1.21.3) Male 1.0 F Race Black 0.9 (0.90.9) Hispanic 1.1 (1.01.1) Other 0.9 (0.80.9) Unknown 0.9 (0.90.9) White 1.0 F Clinical Psychiatric comorbidity 0 1.0 F 1 1.4 (1.31.4) 2 1.7 (1.71.7) Physical comorbidity 0 1.0 F 12 0.74 (0.710.76) 35 0.77 (0.750.80) 67 0.80 (0.770.83) 8 0.87 (0.840.90) Unique medications 13 1.0 F 46 2.5 (2.42.5) 79 4.5 (4.44.6) 10 9.3 (9.19.5) Outpatient visits 12 1.0 F 34 1.1 (1.11.1) 59 1.1 (1.11.2) 10 1.1 (1.11.2)

1.0 (1.01.1) 1.2 (1.11.3) 1.0 F 0.6 (0.60.7) 1.0 F 0.6 (0.60.6) 0.8 (0.80.9) 0.8 (0.70.9) 1.2 (1.11.2) 1.0 F 1.0 F 1.1 (1.11.2) 1.6 (1.51.6) 1.0 F 0.72 (0.670.76) 0.69 (0.650.74) 0.70 (0.660.75) 0.75 (0.700.80) 1.0 F 4.3 (4.14.5) 9.2 (8.89.7) 17.6 (16.718.5) 1.0 F 0.72 (0.700.74) 0.51 (0.490.52) 0.39 (0.370.40)

DISCUSSION Consistent with previous work, this study found inappropriate prescribing to be a relatively common phenomenon, although the availability of comprehensive databases representing a nationally distributed patient population presented the unique opportunity to explore additional facets of the problem. Not only were potentially inappropriate medications commonly used, most use was considered improper based on the Zhan criteria. These data fail to support the idea that most potentially inappropriate medication use is actually proper based on the Zhan criteria, although the data indicate that most dose-limited drug use is appropriate. Consistent with previous studies, the most commonly prescribed drugs included propoxyphene, amitriptyline, and benzodiazepines.4,6,7,9,18,26 Pain relievers, psychotrop-

ic medications, and muscle relaxants constituted 61% of inappropriate drug use, suggesting that many of these patients experience chronic pain. Thus, it may be useful to explore the role of chronic pain in inappropriate prescribing and focus on developing interventions to improve adherence to pain-management guidelines for older people that do not include these agents.31 The data from the current study also indicate that duration of prescriptions is prolonged. Although certain drugs are used to treat chronic illness and require long-term treatment, the duration of use for long-acting benzodiazepines and other psychotropic drugs is of considerable concern because of their association with falls, altered cognition, depression, new institutionalization, and trafc accidents.13,3237 Although some patients who receive these drugs for long durations may not experience obvious adverse effects, many adverse effects may be subtle and arise after prolonged use. Consequently, additional research using longitudinal data is needed to determine the effect of continued and intermittent use on health status for community-dwelling older people and to provide a stronger foundation to support (or dispute) these controversial, consensus-based criteria. In addition to describing additional facets of PIPE, this study also examined the association between inappropriate prescribing and patient demographic and clinical characteristics. An important contribution is the nding that similar patient clinical characteristics predicted use of Zhan-criteria and dose-limited drugs, but demographic characteristics differed. Although the direction of the effect of physical comorbidities reversed when controlling for number of medications, the association with psychiatric diagnoses remained significant, suggesting that all comorbidities are not equal. Further research is needed to determine whether the effects of psychiatric diagnoses are related to receiving only inappropriate psychiatric drugs or all inappropriate drugs. It was found that Hispanics were marginally less likely to receive dose-limited drugs than whites but more likely to receive Zhan-criteria drugs after controlling for clinical characteristics. A previous study found that older Mexican Americans were less likely to receive Zhan-criteria drugs than blacks or whites38 and hypothesized that this was due to less healthcare use by Mexican Americans. Hispanics in the current study received more outpatient clinic visits than whites, which may help account for the nding for Zhancriteria drugs, although both studies demonstrate that the prevalence of PIPE in Hispanics is different from that in other race/ethnicity groups and should not be combined. Variation in predictors for sex and age also existed for Zhan-criteria and dose-limited drugs. Women were more likely than men to receive Zhan-criteria drugs and less likely to receive dose-limited drugs. The youngest old were more likely to receive Zhan-criteria drugs than the oldest old, but differences between these groups were not significantly different for dose-limited drugs. The fact that women and the youngest old were less likely to receive digoxinFthe most commonly prescribed dose-limited drugFthan were men and the oldest old may help explain these ndings. The variations in demographic predictors of Zhan-criteria and dose-limited drugs further illustrate that inappropriate prescribing is a heterogeneous entity. Thus research on each

1288

PUGH ET AL.

AUGUST 2005VOL. 53, NO. 8

JAGS

aspect of inappropriate prescribing is needed to fully understand how best to improve care. Although these data provided an opportunity to better understand the multifaceted nature of PIPE, limitations exist. First, 20% of the cohort was new to the VA during the year of the study. These patients received fewer medications than patients who received care the previous year. Consequently sensitivity analyses were conducted to determine the effect of including these patients in the analyses. A slightly higher prevalence of inappropriate prescribing was found for those who received at least one medication (24%) and similar results if the cohort was restricted to those who had two visits in FY00 or at least one visit each in FY99 and FY00 (21%). Furthermore, the most commonly used drugs did not vary in these sensitivity analyses, and predictors of inappropriate prescribing were unchanged. Second, only VA data were used. Although the overall ndings were similar to those from previous studies,6,9,12,18 the predominantly male population is not representative of the geriatric population, and results must be interpreted accordingly. Third, some older veterans may have had diagnoses for appropriate indications that were not documented in administrative databases, and some may have received additional medications because most are eligible for Medicare and others have private insurance or Medicaid. In summary, the availability of a comprehensive national data source allowed a closer examination of inappropriate prescribing in older people. The data portray inappropriate prescribing as a multifaceted, persistent problem including not only types of drugs, but also doses and duration of use. The data suggest that future research may benet from examining patterns of inappropriate prescribing to determine how underlying issues such as psychiatric disease or chronic pain contribute to this phenomenon. Moreover, the data suggest that, despite many previous studies, we are only beginning to understand the phenomenon of inappropriate prescribing. The multivariable models do not explain a substantial amount of variance; thus, it is likely that other patient characteristics, characteristics of providers, and characteristics of the healthcare system (e.g., formulary restrictions, geographic variation in prescribing) also contribute to inappropriate prescribing.39 Future research that examines interactions between patient, provider, and the healthcare system is needed to provide a more comprehensive understanding of how to reduce prescribing the wrong drug, wrong dose, or wrong duration of medications for older people.

ford, MA, and the Veterans Evidence-based Research, Dissemination, and Implementation Center (VERDICT), Audie L. Murphy VA Hospital, San Antonio, TX. The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs. We also acknowledge Janice E. Knoefel, MD, for consultation in developing operational denitions for appropriate indications, and Lewis Kazis, ScD, for providing access to data from the 1999 Large Health Survey of Veteran Enrollees. The authors have no conicts of interest with the manufacturers of any drug evaluated in this paper. Financial Disclosure: This study was funded by VA Health Services, Research and Development Service, Veterans Health Administration (IIR 02-076-1). The authors have no conicts of interest with the manufacturers of any drug evaluated in this paper. Author Contributions: Mary Jo V. Pugh: study design, conduct of study, interpretation of data, and preparation of manuscript. B. Graeme Fincke: conduct of study, interpretation of data, and preparation of manuscript. Arlene Bierman: study design, conduct of study, interpretation of data, and preparation of manuscript. Bei-Hung Chang: statistical analysis, interpretation of data, and preparation of manuscript. Amy K. Rosen: interpretation of data, and preparation of manuscript. Francesca Cunningham: study design, interpretation of data, and preparation of manuscript. Megan Amuan: data collection, data analysis, preparation of manuscript. Muriel Burk: data collection, interpretation of data and preparation of manuscript. Dan R. Berlowitz: study design, conduct of study, interpretation of data, and preparation of manuscript. Sponsors Role: The sponsor had no role in the design, methods, subject recruitment, data collection, analysis, or paper preparation.

REFERENCES
1. Beers MH. Explicit criteria for determining potentially inappropriate medication use by the elderly. Arch Intern Med 1997;157:15311536. 2. Beers MH, Ouslander JG, Rollingher I et al. Explicit criteria for determining inappropriate medication use in nursing home residents. Arch Intern Med 1991;151:18251832. 3. Aparasu RR, Mort JE. Inappropriate prescribing to the elderly: Beers criteriabased review. Ann Pharmacother 2000;34:338345. 4. Piecoro LT, Browning SR, Prince TS et al. A database analysis of potentially inappropriate drug use in an elderly medicaid population. Pharmacotherapy 2000;20:221228. 5. Fick DM, Cooper JW, Wade WE et al. Updating the Beers criteria for potentially inappropriate medication use in older adults. Results of a US consensus panel of experts. Arch Intern Med 2003;163:27162724. 6. Zhan C, Sangl J, Bierman AS et al. Potentially inappropriate medication use in the community-dwelling elderly: Findings from the 1996 Medical Expenditure Panel Survey. JAMA 2001;286:28232829. 7. Goulding MR. Inappropriate medication prescribing for elderly ambulatory care patients. Arch Intern Med 2004;164:305312. 8. Gurwitz JH. Improving the quality of medication use in elderly patients. Arch Intern Med 2002;162:16701672. 9. Hanlon JT, Schmader KE, Boult C et al. Use of inappropriate prescription drugs by older people. J Am Geriatr Soc 2002;50:2634. 10. Curtis LH, Ostbye T, Sendersky V et al. Inappropriate prescribing for elderly Americans in a large outpatient population. Arch Intern Med 2004;164: 16211625. 11. Simon SR, Chan KA, Soumerai SB et al. Potentially inappropriate medication use by elderly persons in U.S. Health Maintenance Organizations, 200001. J Am Geriatr Soc 2005;53:227232. 12. Stuart BC, Kamal-Bahl SJ, Briesacher BA et al. Trends in prescribing inappropriate drugs for the elderly between 1995 and 1999. Am J Geriatr Pharmacother 2003;1:6174.

ACKNOWLEDGMENT This study was funded by the Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Service IIR-02-076 to Dan R. Berlowitz, MD, MPH, and a Merit Review Entry Program Award to Mary Jo V. Pugh, PhD (MRP-02-267). Dr. Berlowitz is the Director, Center for Health Quality, Outcomes, and Economic Research, Bedford, MA; Dr. Pugh is a Research Health Scientist at the Veterans Evidence-based Research, Dissemination, and Implementation Center, San Antonio, TX. We acknowledge the support of the Edith Nourse Rogers Memorial VA Medical Center, the Center for Health Quality, Outcomes, and Economic Research, Bed-

JAGS

AUGUST 2005VOL. 53, NO. 8

WRONG DRUG, WRONG DOSE, WRONG DURATION

1289

13. Marks R, Allegrante JP, Ronald MacKenzie C et al. Hip fractures among the elderly: Causes, consequences and control. Ageing Res Rev 2003;2:5793. 14. Lau DT, Kasper JD, Potter DEB et al. Hospitalization and death associated with potentially inappropriate medication prescriptions among elderly nursing home residents. Arch Intern Med 2005;165:6874. 15. Phillips J, Beam S, Brinker A et al. Retrospective analysis of mortalities associated with medication errors. Am J Health Syst Pharm 2001;58: 18351841. 16. Grymonpre RE, Mitenko PA, Sitar DS et al. Drug-associated hospital admissions in older medical patients. J Am Geriatr Soc 1988;36:10921098. 17. Huang B, Bachmann KA, He X et al. Inappropriate prescriptions for the aging population of the United States: An analysis of the National Ambulatory Medical Care Survey, 1997. Pharmacoepidemiol Drug Saf 2002;11:127134. 18. Fick DM, Waller JL, Maclean JR et al. Potentially inappropriate medication use in a Medicare managed care population with higher costs and utilization. J Manag Care Pharm 2001;7:407413. 19. Hanlon JT, Fillenbaum GG, Schmader KE et al. Inappropriate drug use among community dwelling elderly. Pharmacotherapy 2000;20:575582. 20. Hogan DB, Maxwell CJ, Fung TS et al. Regional variation in the use of medications by older CanadiansFa persistent and incompletely understood phenomena. Pharmacoepidemiol Drug Saf 2003;12:575582. 21. Luo X, Pietrobon R, Hey L. Patterns and trends in opioid use among individuals with back pain in the United States. Spine 2004;29:884890. 22. International Classication of Diseases, 9th Revision, Clinical Modication, 4th Ed. Los Angeles: U.S. Department of Health and Human Services, 1992. 23. Szeto HC, Coleman RK, Gholami P et al. Accuracy of computerized outpatient diagnoses in a Veterans Affairs general medicine clinic. Am J Manag Care 2002;8:3743. 24. Kashner TM. Agreement between administrative les and written medical records. A case of the Department of Veterans Affairs. Med Care 1998;36: 13241336. 25. Petersen LA, Wright SM, Normand S-LT. Positive predictive value of the diagnosis of acute myocardial infarction in an administrative database. J Gen Intern Med 1999;14:555558.

26. Willcox SM, Himmelstein DU, Woolhandler S. Inappropriate drug prescribing for the community-dwelling elderly. JAMA 1994;272:282286. 27. Selim AJ, Berlowitz DR, Fincke G et al. The health status of elderly veteran enrollees in the Veterans Health Administration. J Am Geriatr Soc 2004;52:12711276. 28. Rowe J, Kahn R. Successful Aging. New York: Pantheon Books, 1998. 29. Schneeweiss S, Seeger JD, Maclure M et al. Performance of comorbidity scores to control for confounding in epidemiologic studies using claims data. Am J Epidemiol 2001;154:854864. 30. Harrell FE, Lee KL, Califf RM et al. Regression modelling strategies for improved prognostic prediction. Stat Med 1984;3:143152. 31. Gloth FM. Pain management in older adults Prevention and treatment. J Am Geriatr Soc 2001;49:188199. 32. Ensrud KE, Blackwell T, Mangione CM et al. Central nervous system active medications and risk for fractures in older women. Arch Intern Med 2003;163:949957. 33. Hemmelgarn B, Suissa S, Huang AR et al. Benzodiazepine use and the risk of motor vehicle crash in the elderly. JAMA 1997;278:2731. 34. Gleason PP, Schulz R, Smith NL et al. Correlates and prevalence of benzodiazepine use in community-dwelling elderly. J Gen Intern Med 1998;13:243250. 35. Gray SL, Lai KV, Larson EB. Drug-induced cognition disorders in the elderly. Incidence prevention and management. Drug Saf 1999;21:101122. 36. Hogan DB, Maxwell CJ, Fung TS et al. Prevalence and potential consequences of benzodiazepine use in senior citizens: Results from the Canadian Study of Health and Aging. Can J Clin Pharmacol 2003;10:7277. 37. Wang PS, Bohn RL, Glynn RJ et al. Hazardous benzodiazepine regimens in the elderly: Effects of half-life, dosage, and duration on risk of hip fracture. Am J Psychiatry 2001;158:892898. 38. Raji MA, Ostir GV, Markides KS et al. Potentially inappropriate medication use by elderly Mexican Americans. Ann Pharmacother 2003;37:11971202. 39. Cabana MD, Rand CS, Powe NR et al. Why dont physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999;282: 14581465.