Objective: To measure the impact of medication copayment level and other predictors on compliance with antihypertensive medications, as measured by the medication possession ratio.
Study Design: Retrospective observational analysis.
Methods: We used claims data from a large managed care organization. The identification of subjects was based on a diagnosis of hypertension and a filled prescription for antihypertensive medication between January 1999 and June 2004. Multivariate logistic regression models were used to evaluate copayment level and patient characteristics as predictors of medication compliance.
Results: Analysis of data for 114 232 patients filling prescriptions for antihypertensive medications revealed that compliance was lower for drugs in less preferred tiers. Relative to medications with a $5 copayment, the odds ratio (95% confidence interval) for compliance with drugs having a $20 copayment was 0.76 (0.75, 0.78); for drugs requiring a $20 to $165 copayment, the odds ratio for compliance was 0.48 (0.47, 0.49). Medication compliance also differed by patient age, morbidity level, and ethnicity, as well as by medication therapeutic class—with the best compliance observed for angiotensin receptor blockers, followed by calcium channel blockers, ß-adrenergic receptor antagonists (ß-blockers), angiotensin-converting enzyme inhibitors, and last, thiazide diuretics.
Conclusion: Copayment level, independent of other determinants, was found to be a strong predictor of compliance with antihypertensive medications, with greater compliance seen among patients filing pharmacy claims for drugs that required lower copayments. This finding suggests that patient use is sensitive to price. The potential impact on compliance should be considered when making pricing and policy decisions.
(Am J Manag Care. 2006;12:678-683)
Poor blood pressure control raises the risk of stroke, myocardial infarction, and heart failure for the 65 million Americans with hypertension.1 Efforts to control hypertension through the use of antihypertensive medications are considered to be among the most efficacious. Because hypertension often is asymptomatic, however, poor patient compliance with pharmacologic treatment has consistently limited the effectiveness of these interventions.2 Other factors, including patient forgetfulness, the number of daily doses, side effects, and/or class of agent, also affect patient compliance.3-6 Substantial economic costs are associated with noncompliance.7,8
In previous studies, increasing levels of patient cost-sharing have been associated with lower medication compliance and persistency.8-14 However, our study is the first comprehensive analysis of the effect of copayment level on medication compliance for all classes of antihypertensive drugs. Our analysis also links data from medical claims to derive an estimate of patient morbidity level, from surveys to include a measure of self-reported ethnicity for a subset of members, and from enrollment files for information on age and sex and type of coverage.
The study sample was drawn from members of a managed care organization covering approximately 650 000 members. Study eligibility required participants (1) to have received a medical diagnosis of hypertension on any type of claim, professional or facility, and (2) to have filled at least 1 antihypertensive medication prescription with at least a 15-day supply between January 1999 and June 2004. We did not require continuous enrollment; instead, we excluded days without coverage from the compliance calculation.
Data Sources and Variable Definitions
Patient information such as age, sex, and type of coverage (HMO, preferred provider organization [PPO], Medicare cost contract) was obtained from administrative data. The diagnosis of hypertension and the data to determine comorbidity level were obtained from medical claims databases. Patient morbidity level was determined by using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes according to the Johns Hopkins Adjusted Clinical Group methodology; levels of 4 or 5 on the 5-point scale were considered high morbidity.15
Prescription data on medication names, fills, and number of days supply were obtained from pharmacy claims databases. Compliance was assessed for each specific therapeutic class. Hence, if a patient switched drugs within a therapeutic class, the number of days supply for the 2 drugs would be added together. If they switched to a new therapeutic class, the days supply would be counted toward the new class. From these data, we calculated a medication possession ratio for each prescription based on the number of days supply from the index prescription fill, divided by the number of days to the last prescription fill (calculated as the fill date of the last prescription claim minus the fill date of the first prescription claim).10 This approach, in which a medication possession ratio of 0.8 was deemed compliant, has been used in other compliance studies.10,16,17 The overall medication possession ratio was obtained by summing the total days of compliance per year across all prescriptions and dividing by the number of days of drug coverage within the year.
Formulary tier was used as an indicator of copayment level. Three copayment levels were possible: $5 for tier 1, generic agents; $20 for tier 2, preferred branded agents; and a variable copayment for nonpreferred branded agents (tier 3) that reflected the difference between the price of the nonpreferred brand and the preferred brand, ranging from $20 to $165.
Ethnicity and education data were drawn from self-reported satisfaction surveys available for 33% of the study population. These mail surveys, which used the 17 ethnic categories developed by the Hawaii Department of Health's Hawaii Health Surveillance Program, asked health plan members to indicate each of the categories that applied to them. Data from these respondents were analyzed to determine medication compliance for the following 6 main ethnic groups: Japanese, Chinese, Caucasians, Hawaiians (this group includes all those who claim Hawaiian ethnicity regardless of other ethnicity selections, because of the small number of pure Hawaiians), Filipinos, and Koreans. Those who claimed membership in more than one of these groups, except Hawaiian, were categorized as "Mixed" ethnicity. Those who indicated membership in any other ethnic group were categorized as "Other."
Patient characteristics were analyzed according to copayment level; if members switched drugs over time, they could appear in several categories. The likelihood of compliance with antihypertensive medications was estimated as a logistic function of copayment level, patient age, sex, race/ethnicity, morbidity level (low/high), type of insurance coverage, and therapeutic class of medication. All analyses were conducted using Stata, version 8 (StataCorp, College Station, Tex). Results were considered significant at the P < .05 level.
Descriptive Analysis of the Study Population
A total of 114 232 patients met study inclusion criteria. The mean age of the study sample was 64 years (SD = 14 years; range, 18 to 107 years). Approximately 50% of all patients were female, and 25% had a high morbidity level. Approximately 70% of all patients were enrolled in the PPO, 28% in the HMO, and 2% in the Medicare cost contract plan. Of the 37 697 patients for whom ethnicity data were available, 42% were Japanese, 14% Hawaiian, 13% Caucasian, 12% Filipino, 8% Chinese, 4% Mixed, and 5% Other. A summary of patient characteristics, broken out by copayment level, is shown in Table 1.
The number of patients submitting pharmacy claims for antihypertensive medications was as follows: 58 809 (31.6%) for medications with a $5 copayment (tier 1); 66 486 (35.8%) for medications with a $20 copayment (tier 2); and 60 553 (32.6%) for medications with a $20 to $165 copayment (tier 3).
Characteristics of patients at each copayment level differed significantly (P < .05); however, the magnitude of these differences was small. Patients with pharmacy claims for tier 1 medications were slightly older (64.5 years vs 63.6 and 64.0 years for tiers 2 and 3, respectively), more likely to be female (51.3% vs <50% for the other 2 tiers), and more likely to have a high morbidity level (26.4% vs 25.6% and 21.1% for tiers 2 and 3, respectively). Patients filling prescriptions at the 3 copayment levels varied slightly by ethnic group. The percentage of Caucasian patients receiving tier 1 medications was slightly higher than the percentage of Caucasian patients in the overall study population (15.3% vs 13%). Filipino patients were more likely than other ethnic groups to have received tier 3 medications (13.4% vs 12%).
Unadjusted Medication Compliance. Overall compliance for antihypertensive agents was 66.8% in tier 1, 66.1% in tier 2, and 54.6% in tier 3. Compliance by therapeutic class of agent is shown in Figure 1. The therapeutic class with the highest compliance was the angiotensin receptor blockers, whereas thiazide diuretics had the lowest compliance. Therapeutic classes with the greatest disparities in compliance between the least ($5) and greatest ($20 to $165) copayment levels were β-adrenergic receptor antagonists (β- blockers) (70% vs 48%) and thiazide diuretics (57% vs 36%).
Compliance was also significantly associated with all patient characteristics examined, including age, sex, morbidity level, type of coverage, and ethnicity (Table 2). Groups that may need to be targeted because of low compliance include members under age 40 years (42.5% compliance), members with Filipino ethnicity (58.7% compliance), and members with HMO coverage (59.7% compliance).
Adjusted Odds Ratios for Medication Compliance. Adjusted odds ratios and 95% confidence intervals (CIs) for medication compliance by copayment level with respect to studied variables are summarized in Table 3. These odds ratios were adjusted to account for differences in tier, therapeutic class, and patient characteristics.
Relative to medications in tier 1, the adjusted odds ratio for compliance with medications in tier 2 was 0.76 (95% CI = 0.75, 0.78), and for medications in tier 3 it was 0.48 (95% CI = 0.47, 0.49) (Figure 2).