Circulation. 2009;119: ; originally published online January 12, 2009; doi: /CIRCULATIONAHA

Transcription

1 Impact of a Prescription Copayment Increase on Lipid-Lowering Medication Adherence in Veterans Jalpa A. Doshi, Jingsan Zhu, Bruce Y. Lee, Stephen E. Kimmel and Kevin G. Volpp Circulation. 2009;119: ; originally published online January 12, 2009; doi: /CIRCULATIONAHA Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX Copyright 2009 American Heart Association, Inc. All rights reserved. Print ISSN: Online ISSN: The online version of this article, along with updated information and services, is located on the World Wide Web at: Data Supplement (unedited) at: Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: Subscriptions: Information about subscribing to Circulation is online at: Downloaded from by guest on February 27, 2014

2 Health Services and Outcomes Research Impact of a Prescription Copayment Increase on Lipid-Lowering Medication Adherence in Veterans Jalpa A. Doshi, PhD; Jingsan Zhu, MBA; Bruce Y. Lee, MD, MBA; Stephen E. Kimmel, MD, MSCE; Kevin G. Volpp, MD, PhD Background In February 2002, the Department of Veterans Affairs (VA) increased copayments from $2 to $7 per 30-day drug supply of each medication for many veterans. We examined the impact of the copayment increase on lipid-lowering medication adherence. Methods and Results This quasiexperimental study used electronic records of 5604 veterans receiving care at the Philadelphia VA Medical Center from November 1999 to April The all copayment group included veterans subject to copayments for all drugs with no annual cap. Veterans subject to copayments for drugs only if indicated for a non service-connected condition with an annual cap of $840 for out-of-pocket costs made up the some copayment group. Veterans who remained copayment exempt formed a natural control group (no copayment group). Patients were identified as adherent if the proportion of days covered with lipid-lowering medications was 80%. Patients were identified as having a continuous gap if they had at least 1 continuous episode with no lipid-lowering medications for 90 days. A difference-in-difference approach compared changes in lipid-lowering medication adherence during the 24 months before and after copayment increase among veterans subject to the copayment change with those who were not. Adherence declined in all 3 groups after the copayment increase. However, the percentage of patients who were adherent (proportion of days covered 80%) declined significantly more in the all copayment ( 19.2%) and some copayment ( 19.3%) groups relative to the exempt group ( 11.9%). The incidence of a continuous gap increased significantly at twice the rate in both copayment groups (all copayment group, 24.6%; some copayment group, 24.1%) as the exempt group (11.7%). Compared with the exempt group, the odds of having a continuous gap in the after relative to the before period were significantly higher in both the all copayment group (odds ratio, 3.04; 95% confidence interval, 2.29 to 4.03) and the some copayment group (odds ratio, 1.85; 95% confidence interval, 1.43 to 2.40). Similar results were seen in subgroups of patients at high risk for coronary heart disease, high medication users, and elderly veterans. Conclusion The copayment increase adversely affected lipid-lowering medication adherence among veterans, including those at high coronary heart disease risk. (Circulation. 2009;119: ) Key Words: coronary disease medication adherence fees, pharmaceutical statins veterans With the rate of spending growth for prescription medications far outpacing other areas of medical care, health plans and employers have increased patient cost sharing for prescription drugs in an attempt to control escalating costs in recent years. 1 As in the private sector, the Department of Veterans Affairs (VA) has faced rapid increases in pharmaceutical spending. In the late 1990s, VA outpatient prescription expenditures were increasing at a 19% annual rate and totaled almost $3 billion in fiscal year To combat these rising expenditures, the VA also increased patient cost sharing through the Veterans Millennium Health Care Act of 1999 (Public Law ), which increased copayments for outpatient medications from $2 to $7 per 30-day supply of each medication starting February Editorial p 365 Clinical Perspective p 397 Numerous studies in the private sector have confirmed that such policies generally result in significant savings in medication costs by reducing use. 3,4 Some studies, however, have also found reductions in the use of essential medications and disruptions in treatment for chronically ill patients. 3,4 Little is known about how changes in copayment policy affect essential medication use among VA patients, a low Received April 8, 2008; accepted October 17, From the Center for Health Research and Promotion, Philadelphia Veterans Affairs Medical Center, Philadelphia, Pa (J.A.D., J.Z., K.G.V.), the Department of Medicine (J.A.D., J.Z., S.E.K., K.G.V.) and Department of Biostatistics and Epidemiology and Center for Clinical Epidemiology and Biostatistics (S.E.K.), University of Pennsylvania School of Medicine, Philadelphia; Center for Health Incentives, Leonard Davis Institute of Health Economics (J.A.D., K.G.V.), and Department of Health Care Systems, The Wharton School (K.G.V.), University of Pennsylvania, Philadelphia; Department of Medicine, University of Pittsburgh School of Medicine, Pittsburgh, Pa (B.Y.L.); and Department of Epidemiology, University of Pittsburgh Graduate School of Public Health, Pittsburgh, Pa (B.Y.L.). The online-only Data Supplement can be found with this article at Correspondence to Jalpa A. Doshi, PhD, 1222 Blockley Hall, Philadelphia, PA jdoshi@mail.med.upenn.edu 2009 American Heart Association, Inc. Circulation is available at DOI: /CIRCULATIONAHA Downloaded from by guest on February 27, 2014

3 Doshi et al Copayments and Lipid-Lowering Drug Adherence 391 socioeconomic-status group with a high rate of comorbidities. Only 2 published studies to date have assessed the effect of the copayment increase under Public Law One study examined changes in the number of 30-day medication supplies for all chronic medications and for broad categories of chronic medications defined as low- versus high-cost chronic medications and over-the-counter versus prescription-only chronic medications 5 ; the second study focused only on patients with schizophrenia. 6 Evaluation of the effect of this policy change in veterans with other chronic conditions and on the use of specific essential medication classes is needed to inform ongoing policy debates because further increases in VA copayments have been part of recently proposed presidential budgets. 7,8 Statins (3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors), the most commonly used lipid-lowering agents, have been shown to significantly reduce the risk of future coronary events and cardiovascular mortality in patients at high coronary heart disease (CHD) risk. 9 Yet studies have shown that treatment and adherence to such lipidlowering drugs are suboptimal among patients in all age groups Given that the prevalence and burden of heart disease in the VA population have been found to be higher than the non-va population 13 and that efforts have been made to improve adherence with anticholesterol medications in VA patients, 14 identifying whether increases in copayments may worsen adherence to these essential medications in this population is critical. This study examines the impact of the VA copayment increase from $2 to $7 on adherence to lipid-lowering medications in veterans through the use of prescription refill data. Methods Data Source and Study Sample The study used administrative data and electronic medical record information entered, processed, and stored in the Veterans Health Information Systems Technology and Architecture (VistA) database, which is the automated environment that supports day-to-day operations at local VA healthcare facilities. This system provides data on patient demographics, copay-exempt status, medical diagnoses, laboratory results, and prescription orders written and filled within the VA system. Our study used the Philadelphia Veterans Affairs Medical Center s VistA database. The Philadelphia Veterans Affairs Medical Center is a VA tertiary referral center for the eastern half of the VA Stars & Stripes Healthcare Network, VISN 4, and provides health care for all veterans living in Philadelphia and 7 surrounding counties in Pennsylvania, New Jersey, and Delaware, a large number of whom are black and the majority are men. The study sample frame consisted of patients receiving care at the Philadelphia Veterans Affairs Medical Center who filled at least 1 lipid-lowering medication in the 24 months before the copayment increase because patients put on lipid-lowering medication are generally intended to stay on it indefinitely. However, to ensure that patients had not already discontinued lipid-lowering medication use before the copayment policy change, we excluded patients without any evidence of use of these medications in the 3-month period before February 1, Other exclusion criteria included the following: death during the study period, history of dementia or cancer, missing information on copay exemption status, and changes in copay exemption status during the study period. Study Design A quasiexperimental design was used to compare changes in lipid-lowering medication use over time for veterans who experienced increases in prescription copayments versus veterans who did not. This pre-post intervention study with a contemporaneous control estimates the impact of the VA copayment increase by comparing the magnitude of the difference in outcomes before and after the increase for veterans subject to the copayments with the magnitude of the difference in outcomes before and after the increase for veterans not subject to copayments. We determined whether veterans were subject to the prescription copayment depending on their VA priority category (1 through 8, with 1 being the highest priority). Priority groups were established to help the VA manage access in relation to its available resources and to ensure that resources are allocated to veterans with the highest priority for enrollment. 2 Veterans with priority 1 have a serviceconnected disability rated 50% disabling and are exempt from all prescription drug copayments. These copay exempt veterans formed a natural control group given that they did not face copayments in the preincrease or postincrease period. Next, we defined 2 intervention groups that were subject to copayment increases since February The some copay group included veterans in priority groups 2 through 6 who have service-connected disabilities rated 50% disabling, lower incomes, or other recognized statuses that confer higher priority than those without (eg, veterans awarded the Purple Heart, former prisoners of war). They are subject to copayments for drugs only if indicated for a non service-connected condition. Because service-connected disability refers to a disability that was incurred or aggravated in the line of duty in the active military, naval, or air service (eg, amputations, burns, posttraumatic stress disorder, and traumatic brain injuries), hyperlipidemia is unlikely to be a service-connected disability; hence, most veterans in this group were subject to copayments for their lipid-lowering medications. Nevertheless, an annual maximum of $840 (increased to $960 in 2006) for out-of-pocket medication costs has been established for these veterans (ie, no copays for additional fills after 120 prescription months per year). The all copay group included veterans in priority groups 7 and 8 who have incomes, net worth, or both above the VA means test threshold and are required to pay copayments for all prescription drugs. An annual medication copayment cap has not been established for veterans enrolled in priority groups 7 and 8. For each group, the 2 comparison time periods were a 24-month pre period (November 1, 1999, to October 31, 2001) before the copayment increase and a 24-month post period (May 1, 2002, to April 30, 2004) after the copayment increase. The 3-month period before and after the copayment increase date was treated as a washout period to account for potential stockpiling of medications and to allow depletion of existing stores of medications purchased in the preincrease period. Outcome Variables Lipid-lowering medication use was measured from the VistA Prescription File, which contains records of all outpatient medications dispensed, including those processed by the VA mail-order pharmacy. Medication use outcomes were calculated separately for the 24-month preincrease and 24-month postincrease period. The preincrease period observation window for new users (identified as those without any lipid-lowering medication fill in the 12 months before the preincrease period) was adjusted to initiate from the date of the first prescription fill if later than November 1, Our first outcome was measured as the proportion of days covered (PDC) in each period with lipid-lowering medications. 15 The PDC was calculated as the number of days with lipid-lowering drug supply on hand divided by the number of days in the specified time period. Information from the VistA prescription file such as the prescription fill date and the days supply was used to designate each day in the period as covered or not. Filled prescriptions were evaluated using a set of rules to avoid double counting covered days. The portion of the days supply from prescriptions filled before the start date of the time period that would run into the follow-up period was used to assign covered days. When a prescription was issued before the previous prescription should have run out, use of the new prescription was assumed to begin the day after the end of the old

4 392 Circulation January 27, 2009 prescription, and days supply for the new prescription was appended to the end of the previous fill. Any surplus days supply remaining on the last day of the postincrease period was truncated. Thus, the PDC could have a value only between 0 and 1 and was multiplied by 100 to yield a percentage. On the basis of their PDC measure, patients were classified as adherent in each period if their PDC was 80%. 11,16 Although the PDC indicates the extent of availability of medication, it does not provide information on whether any gaps in use were continuous or were dispersed over the entire period; each of these scenarios has different clinical implications. Hence, a second set of measures for continuous gaps was used to indicate whether the patient had at least 1 continuous episode with no lipid-lowering medication for a minimum of 90 days in the study periods. Sensitivity analyses were performed with a minimum of 30-, 60-, and 120-day continuous gap periods. Independent Variables The independent variables included 2 dummy variables indicating whether the veteran was in one of the groups that paid copayments (ie, some copay group or all copay group) versus being in the copayment exempt group, a dummy variable indicating whether the observation occurred in the postincrease period, and terms of each copayment group and the post period variable. The terms were the key independent variables of interest providing an estimate of the policy effect. Other covariates included baseline patient demographics such as age and gender. Because patient-level income was not complete and reliable in VistA, we used the patient s residence area level median household income information derived from (5-digit) ZIP code matched census data as a proxy measure. We also included covariates capturing a history of medical conditions and/or risk factors that could affect the likelihood of using lipid-lowering medications such as coronary artery disease, diabetes, peripheral vascular disease, cerebrovascular disease, and hypertension (see the Appendix in the Online-only Data Supplement for International Classification of Diseases, ninth edition, clinical modification, codes). In addition, the total number of distinct medication classes (except lipid-lowering agents) filled per year by the veteran in the preincrease period was included as a measure of the complexity of medication regimen and because the out-of-pocket burden for medication expenses resulting from the copayment increase would be higher among those using several drugs. Finally, an indicator for a new user of lipid-lowering medications also was included as a control variable. Analysis A difference-in-difference analytic approach was used. Descriptive results compared the before-after difference in outcomes in each of the 2 groups subject to copayment increases relative to the copayment exempt group. In multivariate analysis, the impact of the copayment increase was assessed using the copayment status (all copay group, some copay group) by time (postincrease) term for each copayment group. Generalized estimating equations for binary data with logit link function were used to estimate the adherent (ie, PDC 80%) and the continuous gap outcomes. SEs accounted for the clustering caused by repeated observations for each patient. We also examined our outcomes within vulnerable patient subgroups such as those at high risk for CHD (ie, history of coronary artery disease, atherosclerosis, diabetes, peripheral vascular disease, or cerebrovascular disease), elderly veterans ( 65 years of age), and veterans with a high medication burden (ie, 5 distinct medication classes 5 in the preincrease period). Although the primary analyses excluded data from the 3-month preincrease and postincrease period (washout period), we included this period in our sensitivity analysis. Sensitivity analysis also was conducted by reducing the observation window from 24 months before and 24 months after the policy change to 12 months before and after the policy change. Furthermore, this analysis of 12 months of data before and after policy change was conducted with and without the washout period. All analyses were conducted with SAS version (SAS Institute Inc, Cary, NC). Table 1. Sample Characteristics by Copayment Status Copay Exempt Group Some Copay Group All Copay Group Total, n Age, mean (SD), y 70.1 (11.3) 73.4 (9.8)* 76.5 (6.6)* Male, % * Median household income, * 52.6* $1000 Cardiovascular risk factors, % Coronary artery disease * Atherosclerosis * 4.1* Diabetes mellitus * 0.9* Peripheral vascular disease * Cerebrovascular disease * Hypertension * Distinct medication classes, median, n * 3.0* *Significantly different from copay exempt group at P The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written. Results A total of 5604 patients met the inclusion criteria for the study. About 96% of these patients used statins as the only lipid-lowering agent; 0.3% of the patients used only a nonstatin lipid-lowering agent; and the remainder had prescriptions for both during the study period. As Table 1 shows, 9% of the sample was copay exempt (n 495), almost half belonged to the some copayment group (n 2793), and the remainder were in the all copayment group (n 2316). The mean age of the patients in each group was 70 years, with patients in the all and some copayment groups being significantly older than patients in the copayment exempt group. The median ZIP code level household income was lowest for the some copayment group, followed by the copayment exempt group. As expected per the VA enrollment criteria for the priority groups included in the all copayment group, income was the highest in this group. Except for diabetes mellitus and atherosclerosis, the prevalence of other cardiovascular risk factors in the some copayment group was similar to that in the copayment exempt group. On the other hand, the all copayment group had significantly lower prevalence of all cardiovascular risk factors. The number of different medications prescribed per month was significantly higher in the copayment exempt group than in the all and some copayment groups. The Figure displays plots of the mean PDC in each month in the 24-month periods before and after the copayment increase for the 3 groups. Adherence levels were high and relatively stable in the majority of the preincrease period across all groups, with the highest PDC observed within the all copayment group (mean PDC, 0.89). The PDC values in the preincrease period for both the some copayment and the copayment exempt groups were similar and slightly lower (mean PDC, 0.83 in both groups). The declining slope of the

5 Doshi et al Copayments and Lipid-Lowering Drug Adherence 393 Figure. Mean monthly PDC for lipid-lowering medications by copayment status and time. PDC plots reflects a reduction in adherence for all 3 groups. Relative to the adherence patterns in the copayment exempt group, clearly a larger decline occurred among the all and some copayment groups after the copayment increase. Table 2 summarizes the results on unadjusted outcomes by copayment group and time period. Both outcome measures were similar in the copayment exempt group and the some copayment group in the preincrease period. Between 71% and 73% of the patients in these 2 groups were adherent (PDC 80%) to lipid-lowering medications in the preincrease period, and between 20% and 24% of the patients in these 2 groups had at least 1 continuous gap of 90 days in the preincrease period. On the other hand, and as reflected in the Figure, adherence was significantly higher in the all copayment group, with 83% of the patients identified as adherent and only 9% having a continuous gap of 90 days in the preincrease period. After the increase in copayments, the percentage of patients adherent to lipid-lowering medications declined in all 3 groups. However, the percent of patients who were adherent in the postincrease period declined significantly more in the all ( 19.2%) and some ( 19.3%) copayment groups relative to the copayment exempt group ( 11.9%; P 0.05 for both comparisons). Similarly, the incidence of a continuous gap of 90 days increased at twice the rate in both the all copayment group (24.6%) and some copayment group (24.1%) relative to the copayment exempt group (11.7%; P for both comparisons). These results were confirmed in the multivariate analyses as illustrated by the adjusted odds ratios (ORs) on the copayment group by postincrease period terms Table 2. Unadjusted Outcomes by Copayment Status and Time Period Outcome by Copayment Status Adherent (PDC 80%) Preincrease Period, % Postincrease Period, % Difference (Post Pre), % Difference in Difference, % Copay exempt Some copay group All copay group Continuous gap ( 90 d) Copay exempt Some copay group All copay group P

6 394 Circulation January 27, 2009 Table 3. Logistic Regression Results: Adherent and Continuous Gap Adherent (PDC 80%) Continuous Gap ( 90 d) Adjusted OR 95% CI P Adjusted OR 95% CI P Some copay group 0.87 ( ) ( ) 0.24 All copay group 1.44 ( ) ( ) Postincrease period 0.58 ( ) ( ) Some copay group by postincrease 0.73 ( ) ( ) period All copay group by postincrease period 0.61 ( ) ( ) Age 1.03 ( ) ( ) Female 0.64 ( ) ( ) ZIP code level household income 1.01 ( ) ( ) Medical conditions Coronary artery disease 0.99 ( ) ( ) 0.99 Atherosclerosis 1.03 ( ) ( ) 0.47 Diabetes 0.89 ( ) ( ) 0.44 Peripheral vascular disease 0.98 ( ) ( ) 0.82 Cerebrovascular disease 1.03 ( ) ( ) 0.74 Hypertension 1.01 ( ) ( ) 0.45 No. of distinct medication classes 1.01 ( ) ( ) New user 0.91 ( ) ( ) (Table 3). After other factors are controlled for, compared with the control group of copayment exempt veterans, the odds of being adherent in the postincrease period relative to the preincrease period were significantly lower among the all copayment group (OR, 0.61; 95% confidence interval [CI], 0.47 to 0.80) and the some copayment group (OR, 0.73; 95% CI, 0.57 to 0.95). Similarly, the odds of having a continuous gap of 90 days in the postincrease period relative to the preincrease period were almost 3 times as high in the all copayment group (OR, 3.04; 95% CI, 2.29 to 4.03) and twice as high in the some copayment group (OR, 1.85; 95% CI, 1.43 to 2.40) compared with the copayment exempt group. Subgroup analysis indicated that the impact of the copayment increase was felt similarly among the 3 vulnerable groups (Table 4). Even among the subgroup of veterans at high CHD risk, the adjusted odds of having a continuous gap of 90 days in lipid-lowering medication use in the postincrease period relative to the preincrease period were almost Table 4. Logistic Regression Results: Subgroup Analyses Adherent (PDC 80%) Continuous Gap ( 90 d) Adjusted OR 95% CI P Adjusted OR 95% CI P High CHD risk* Some copay group by postincrease All copay group by postincrease High medication users Some copay group by postincrease All copay group by postincrease Age 65 y Some copay group by postincrease All copay group by postincrease *Includes patients with history of coronary artery disease, atherosclerosis, diabetes mellitus, peripheral vascular disease, and cerebrovascular disease. Includes patients using 5 distinct medication classes.

7 Doshi et al Copayments and Lipid-Lowering Drug Adherence 395 Table 5. Logistic Regression Results: Sensitivity Analysis by Outcome and Study Period Definitions Adjusted OR Continuous Gap Adherent (PDC 80%) 30 d 60 d 90 d 120 d 2-y Pre-post analysis with washout period Some copay group by postincrease 0.73* All copay group by postincrease y Pre-post analysis without washout period Some copay group by postincrease All copay group by postincrease y Pre-post analysis with washout period Some copay group by postincrease All copay group by postincrease y Pre-post analysis without washout period Some copay group by postincrease All copay group by postincrease *Significant at P Significant at P twice as high in the some copayment group and 3 times as high in the all copayment group compared with the copayment exempt group. Sensitivity analysis with the outcome of continuous gaps defined by 30, 60, or 120 days resulted in similar findings (Table 5). Furthermore, including data from the washout period, reducing the observation window from 24 to 12 months for the preincrease and postincrease period, or both in combination did not substantially change the significance or magnitude of the results. Discussion The VA copayment increase from $2 to $7 adversely affected lipid-lowering medication adherence among veterans. The likelihood of being adherent, as measured by evidence of possession of lipid-lowering medications for 80% of the observation period, significantly declined among veterans subject to copayments for some or all drugs compared with veterans exempt from copayments. This decline in adherence was not just a result of short gaps in use interspersed between prescription refills. In fact, the copayment increase was accompanied by a significant increase in the likelihood of having continuous gaps of 120 days in lipid-lowering medication use among veterans subject to copayments relative to those who were exempt. Gaps in use for such extended periods clearly have clinical implications and in fact have been called discontinuation of therapy in many medication adherence studies using prescription refill data. 16,17 Of even greater concern was our finding that a similar adverse effect of the copayment increase was observed in groups at higher risk for CHD who were using these medications for either primary or secondary prevention. Our study is the first to examine the impact of the VA copayment increase on lipid-lowering medication adherence in veterans. Our findings on worsening adherence to lipidlowering medications are supported by previous studies that have evaluated the impact of increased cost sharing on statin adherence in other patient populations Most of these studies examined patients with employer-sponsored health insurance, and the increase in their cost-sharing requirements was considerably higher than the VA copayment increase from $2 to $7. Nonetheless, we found a comparable or even larger negative impact on lipid-lowering medication adherence in our sample of veterans. Given that copayments in the private sector were much higher during this time period, with mean copayments from 2001 to 2004 increasing for generic drugs from $7 to $10 (42.9%), for preferred brand-name drugs from $13 to $21 (61.5%), and for nonpreferred brand-name drugs from $17 to $33 (94.1%), 1,3 policy makers who proposed this VA copayment increase may have thought that an increase of $5 and the new copayment of $7 were small enough to be unlikely to adversely affect use. However, the $2 to $7 increase more than tripled the out-of-pocket costs for veterans, who were likely have lower incomes than the patients affected by the changes in copayments in much of the private sector, and our study indicates that it did have a significant negative impact on adherence for at least 1 class of essential medications that has been shown to significantly reduce cardiovascular morbidity and mortality. Our findings have several important implications for VA policy makers. First, price sensitivity varies across different populations, and small absolute changes in copayments may still have a significant impact on medication use in populations with lower socioeconomic status, as had been reported previously with even a 50-cent copayment increase among Medicaid patients. 22 Hence, policy makers need to be cautious before implementing abrupt increases in copayments within low socioeconomic-status populations. Second, the benefits of a uniform VA copayment requirement that does not vary with the brand/generic status, cost, or clinical benefits of the drug could be debated. 23 Although the VA followed the private sector s prescription cost-containment strategy of increased cost sharing, it has not embraced tiered copayments, which are widely prevalent among employer-

8 396 Circulation January 27, 2009 sponsored plans. This is particularly relevant in the case of statins. Two statins (simvastatin and pravastatin) have been available as generics since 2006 and are available at significantly lower prices to the VA. Presumably, the VA could charge veterans lower copayments for such medications and thereby facilitate higher adherence to drugs from such essential medication classes. Furthermore, with the increasing availability of $4 generic drug programs from large discount department stores such as Walmart and Target since 2006, future increases in VA copayments that apply to both generics and brands will further drive veterans to fill some of their generics at such stores. This will limit reliable tracking of medication adherence among veterans using VA pharmacy refill records for both clinical and research purposes and may have important implications in terms of monitoring and improving quality of care and outcomes in veterans with chronic conditions. Finally, our finding that the increased copayment had an equally large adverse impact among patients at high CHD risk suggests that policymakers need to pay particular attention to the fact that a one-size-fits-all approach to designing cost-sharing policies may adversely affect certain higher-risk patient groups. A more promising approach that has been proposed recently is to link copayments to individual patient need (specifically, lower patient copayments for higher expected therapeutic benefit and higher copayments for lower therapeutic benefit). 23 In fact, a simulation-based study suggests that varying copayments for lipid-lowering therapy by therapeutic need would increase adherence and reduce hospitalizations and emergency department use, resulting in total savings of more than $1 billion annually. 21 Several limitations of this study should be acknowledged. First, the study was limited to veterans at the Philadelphia Veterans Affairs Medical Center, and the impact of the prescription copayment policy change may not be generalizable to other VA sites, although no obvious reason exits why that would be the case. Second, we used prescription refill patterns as a proxy for medication adherence. Third, we had access only to prescription and medical services received within the VA system. Hence, the occurrence of continuous gaps of 90 days in lipid-lowering medication use does not necessarily indicate that the veterans did not fill the prescription elsewhere, although it would be unlikely that many veterans could have obtained these prescriptions outside the VA system for less than $7 per month, 1,3 given that all statins except lovastatin were not available as generics during this time period. Fourth, the study was not a randomized experiment; hence, unobserved differences between the copayment groups and copay exempt group may have existed. Nevertheless, our pre-post study design with a contemporaneous control is considered a strong and valid observational study design. To the extent that the unobserved differences between the groups are time invariant, our difference-in-difference analysis controls for them. Finally, the study did not evaluate whether the reduction in adherence as a result of the copayment increase had an adverse impact on cardiovascular outcomes or total medical expenditures. The clinical ramifications and medical cost consequences of reductions in adherence with such preventive chronic medications may not be immediately apparent and require data from several years of follow-up, although other work has suggested that the increased costs ensuing from prescription cost sharing for beneficial medications may substantially offset any shortterm savings to payors. 24 Understanding the impact of this increase in copayments on health outcomes among veterans has clear policy relevance. In this era of large federal budget deficits, it is clear that there will be ongoing pressure to reduce or at least constrain growth of the VA budget, and one of the approaches that the Congress may take to cut costs is to increase VA prescription copayments. In January 2006, the VA prescription copayment was increased from $7 to $8. In fact, the president s fiscal year 2008 budget proposals and several previous years budget proposals had recommended increasing VA prescription copayments to $15 per 30-day supply. 7,8 Although these proposals were not incorporated into legislation, determination of the societal impact of increases in copayments requires careful assessment of the impacts on prescription use and health outcomes among patients, particularly those at high risk. Such understanding is imperative to consider in future policy reform initiatives. Sources of Funding We are grateful for funding support from the VA Center for Health Equity Research and Promotion; the American Heart Association Pharmaceutical Roundtable Award (grant N); the Commonwealth of Pennsylvania (grant ); the National Institute of Aging (R01 AG ); and the Penn Institute on Aging. The funding organizations or sponsors played no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript. Disclosures Dr Doshi has investigator-initiated research funding from Pfizer related to the use of cardiovascular medications and owns stock in Merck & Co, Inc. Dr Kimmel has received funding for research and consulted for several pharmaceutical companies, including Pfizer, all unrelated to statins, in the past 2 years. Dr Volpp has received investigator-initiated research funding from Pfizer related to the use of antihypertensive medications. References 1. Section 9: prescription drugs and mental health benefits. In: Employer Health Benefits: 2004 Annual Survey. Menlo Park, Calif: Kaiser Family Foundation and the Health Research and Educational Trust; 2004: VA Health Care: Expanded Eligibility Has Increased Outpatient Pharmacy Use and Expenditures. Washington, DC: US General Accounting Office; report to the Chairman, Committee on Veterans Affairs, House of Representatives; GAO Gibson TB, Ozminkowski RJ, Goetzel RZ. The effects of prescription drug cost sharing: a review of the evidence. Am J Manag Care. 2005; 11: Goldman DP, Joyce GF, Zheng Y. Prescription drug cost sharing: associations with medication and medical utilization and spending and health. JAMA. 2007;298: Stroupe KT, Smith BM, Lee TA, Tarlov E, Durazo-Arvizu R, Huo Z, Barnett T, Cao L, Burk M, Cunningham F, Hynes DM, Weiss KB. Effect of increased copayments on pharmacy use in the Department of Veterans Affairs. Med Care. 2007;45: Zeber JE, Grazier KL, Valenstein M, Blow FC, Lantz PM. Effect of a medication copayment increase in veterans with schizophrenia. Am J Manag Care. 2007;13: Pear R, Hulse C. Bush budget raises drug prices for many veterans. Available at:

9 Doshi et al Copayments and Lipid-Lowering Drug Adherence budget.html?pagewanted print&position. Accessed September 24, Spotswood S. Enrollment fees, copays, research budget still matters for contention in 08 va budget proposal. Available at: com/article.cfm?articleid 1523&issueID 97. Accessed February 25, Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Executive summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA. 2001;285: Avorn J, Monette J, Lacour A, Bohn RL, Monane M, Mogun H, LeLorier J. Persistence of use of lipid-lowering medications: a cross-national study. JAMA. 1998;279: Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein MC, Avorn J. Long-term persistence in use of statin therapy in elderly patients. JAMA. 2002;288: Jackevicius CA, Anderson GM, Leiter L, Tu JV. Use of the statins in patients after acute myocardial infarction: does evidence change practice? Arch Intern Med. 2001;161: Rogers WH, Kazis LE, Miller DR, Skinner KM, Clark JA, Spiro A 3rd, Fincke RG. Comparing the health status of VA and non-va ambulatory patients: the Veterans Health and Medical Outcomes Studies. J Ambul Care. 2004;27: Hiatt JG, Shamsie SG, Schectman G. Discontinuation rates of cholesterol-lowering medications: implications for primary care. Am J Manag Care. 1999;5: Peterson AM, Nau DP, Cramer JA, Benner J, Gwadry-Sridhar F, Nichol M. A checklist for medication compliance and persistence studies using retrospective databases. Value Health. 2007;10: Andrade SE, Kahler KH, Frech F, Chan KA. Methods for evaluation of medication adherence and persistence using automated databases. Pharmacoepidemiol Drug Saf. 2006;15: Schneeweiss S, Patrick AR, Maclure M, Dormuth CR, Glynn RJ. Adherence to statin therapy under drug cost sharing in patients with and without acute myocardial infarction: a population-based natural experiment. Circulation. 2007;115: Goldman DP, Joyce GF, Escarce JJ, Pace JE, Solomon MD, Laouri M, Landsman PB, Teutsch SM. Pharmacy benefits and the use of drugs by the chronically ill. JAMA. 2004;291: Huskamp HA, Deverka PA, Epstein AM, Epstein RS, McGuigan KA, Frank RG. The effect of incentive-based formularies on prescription-drug utilization and spending. N Engl J Med. 2003;349: Gibson TB, Mark TL, Axelsen K, Baser O, Rublee DA, McGuigan KA. Impact of statin copayments on adherence and medical care utilization and expenditures. Am J Manag Care. 2006;12:SP11 SP Goldman DP, Joyce GF, Karaca-Mandic P. Varying pharmacy benefits with clinical status: the case of cholesterol-lowering therapy. Am J Manag Care. 2006;12: Nelson AA Jr, Reeder CE, Dickson WM. The effect of a medicaid drug copayment program on the utilization and cost of prescription services. Med Care. 1984;22: Fendrick AM, Smith DG, Chernew ME, Shah SN. A benefit-based copay for prescription drugs: patient contribution based on total benefits, not drug acquisition cost. Am J Manag Care. 2001;7: Hsu J, Price M, Huang J, Brand R, Fung V, Hui R, Fireman B, Newhouse JP, Selby JV. Unintended consequences of caps on medicare drug benefits. N Engl J Med. 2006;354: CLINICAL PERSPECTIVE Statins, the most commonly used lipid-lowering agents, have been shown to significantly reduce the risk of future coronary events. Yet studies have shown that adherence to such lipid-lowering drugs is suboptimal among patients in all age groups. This study showed that an increase in prescription copayments from $2 to $7 per 30-day drug supply among veterans significantly reduced adherence to lipid-lowering medications. This decline in adherence was not just a result of short gaps in use interspersed between prescription refills. The copayment increase was accompanied by a significant increase in the likelihood of having continuous gaps of up to 120 days in lipid-lowering medication use. This seemingly small $5 increase, which more than tripled the copayments, adversely affected lipid-lowering medication adherence even among veterans at high risk for coronary heart disease.

10 SUPPLEMENTAL MATERIAL ICD-9-CM and CPT codes to identify medical conditions Medical condition Codes ICD-9-CM diagnosis codes: "410", "411", "412", "414", "4130", "4139", "4292", ICD-9-CM procedure codes: 3601, 3602, 3605, 361, 362 CPT codes: 33510, 33511, 33512, 33513, 33514, 33515, 33516, 33517, 33518, 33519, 33521, 33522, 33523, 33533, 33534, 33535, 33536, 92980, Coronary artery disease "92981", "92982", "92984", "92995","92996" ICD-9-CM diagnosis codes: "440", "4292", "4370", "4371", ICD-9-CM procedure codes: 360, 361, 362, 363, "3812", "3813", "3814", "3816", "3818", "3834", "3844", "3922", "3923", "3924", "3925", "3926", "3928", "3929", "3950", "9910" CPT codes: "35301", "35380", "35381", "35390", "33530", "33531", "33532", "33533", "33534", "33535", "33536", "35490", "35491", "35492", "35493", "35494", "35495", "33572", "35680", "35681", "35682", "35683", "92975", "92976", "92977", "92980", "92981", Atherosclerosis "92982", "92983", "92984", "92995", "92996" ICD-9-CM diagnosis codes: 250, "3572", Diabetes "36200", "36201", "36641" Peripheral vascular disease ICD-9-CM diagnosis codes: "443" ICD-9-CM diagnosis codes: 430, 431, 436, 4320, 4321, 4329, 43301, 43311, 43321, 43331, 43381, Cerebrovascular disease 43391, 43401, 43411, ICD-9-CM diagnosis codes: 401, 402, Hypertension 403, 404, 405