The American Journal of Managed Care July 2011
Adherence to Medication Under Mandatory and Voluntary Mail Benefit Designs
Objective: To compare adherence rates under voluntary and mandatory mail benefit designs.
Study Design: Matched retrospective cohort.
Methods: Adherence rates in the first year of therapy were compared between voluntary and mandatory mail cohorts composed of individuals who initiated statin, angiotensin-converting enzyme (ACE) inhibitor, angiotensin receptor blocker (ARB), platelet aggregation inhibitor, metformin, glitazone, or sulfonylurea therapy at a retail pharmacy between January 1 and March 31, 2009. Initiators in mandatory mail plans were matched on therapeutic class, age, sex, prospective risk, and cost of initial prescription with those in voluntary mail plans. Logistic regression models of optimal adherence were constructed to adjust for measured confounders.
Results: Persistence rates were similar through the first 60 days of therapy. The mandatory mail cohort had a notable drop in persistence by day 90 (63.3% vs 56.3%, P <.001), with a more pronounced drop among those without previous mail-service pharmacy use (50.5%). Median medication possession ratio (49.2% vs 57.4%) and optimal adherence (33.6% vs 36.1) were also lower. In the multivariable models, mandatory mail participants were less likely to achieve optimal adherence overall (odds ratio [OR] 0.70; 95% confidence interval [CI] 0.67-0.74) and in the metformin (OR 0.55), sulfonylurea (OR 0.72), ACE inhibitor (OR 0.74), ARB (OR 0.69), and statin (OR 0.69) classes. Participants with no prior use of mail-service pharmacy had significantly lower odds of achieving optimal adherence in all therapeutic classes.
Conclusions: Mandatory mail appears to cause some members to discontinue therapy prematurely, particularly those without previous mailservice pharmacy experience.
(Am J Manag Care. 2011;17(7):e260-e269)
Pharmacy benefit designs that mandate mail pharmacy use interfere with prescription drug access that could be particularly important for individuals with chronic conditions.
- By restricting pharmacy choice and access, mandatory mail appears to cause some members to discontinue therapy prematurely.
- When members choose to discontinue rather than switch pharmacy channels, the unintended consequence is a reduction in medication adherence and the potential for increased medical expenses.
- Individuals without previous use of mail-service pharmacy are particularly sensitive to this plan design and are an important population to target for interventions to support adherence.
The majority of plan sponsors offer “voluntary mail”: a pharmacy benefit that allows prescriptions to be dispensed from either retail or mail-service pharmacy but that discounts out-of-pocket costs to encourage mail-service pharmacy use. In contrast, nearly 19% of employers implement “mandatory mail,” a design that requires use of mail-service pharmacy for maintenance medications by ending benefit coverage for retail-dispensed prescriptions after a predetermined limit is reached. By requiring mail-service pharmacy use for all chronic medications, plan sponsors can expect additional cost savings,4 but they limit access to retail pharmacies, a channel through which the vast majority of prescriptions are dispensed.5
Little is known about how voluntary and mandatory mail benefit designs affect adherence to essential medications. Plan sponsors should be aware of the trade-offs in adherence that may be related to limiting pharmacy choice or other barriers to drug access imposed by mandating a specific pharmacy channel. This trade-off can be analogous to the choice that payers must make when assigning copayments to medications: cost savings through elevated copayments can come at the expense of nonadherence to essential medications.6 Accordingly, we compared voluntary and mandatory mail pharmacy benefit designs with respect to their impact on medication adherence. Payers can use this information to appropriately balance cost management with beneficiary medication adherence when designing pharmacy benefits.
We conducted a matched retrospective cohort study of medication adherence among initiators in their first year of therapy. The sampling frame was employersponsored pharmacy insurance plans that offered either mandatory (n = 85) or voluntary (n = 240) mail pharmacy benefits managed by CVS Caremark between July 1, 2008, and March 31, 2010. Within this sampling frame, we identified potential study subjects among members who were enrolled in a pharmacy insurance plan with a mandatory mail benefit (mandatory mail participants) or a plan with a voluntary mail benefit (voluntary mail participants). In addition, eligible study participants met the following criteria: (1) valid demographics (age, sex, zip code); (2) valid Pharmacy Risk Group score; (3) a paid retail 30-day pharmacy claim (index prescription) between January 1 and March 31, 2009, for a prescription drug in 1 of the following therapeutic classes: HMG CoA reductase inhibitors (statins), angiotensin receptor blockers (ARBs), angiotensin-converting enzyme (ACE) inhibitors, platelet aggregation inhibitors (PAIs), sulfonylureas, metformin, or glitazones; (4) no previous paid pharmacy claim for a drug in the same class within the previous 6 months; and (5) continuous eligibility for pharmacy benefits for the 6 months before and 12 months subsequent to the index prescription.
Mandatory mail participants who met these qualifications were individually matched (1:1 match without replacement) to voluntary mail participants on the following criteria: age (±5 years); sex; index therapeutic class; out-of-pocket cost for the index prescription (±$0.20 per day of supply); and Pharmacy Risk Group category. Therapeutic class definitions were derived from Medi-Span’s Generic Product Identifier. The Pharmacy Risk Group score uses a proprietary algorithm based on pharmacy claims to provide an estimate of each individual’s predicted healthcare costs; this score can be used to adjust for risk in the absence of medical utilization data. The Pharmacy Risk Group score was derived from claims filled prior to the index prescription date. The final analytical sample was limited to the mandatory and voluntary mail participants successfully matched.
Medication adherence was measured for each therapeutic class separately using the following metrics: medication possession ratio (MPR) and persistence. The MPR was defined as the total days of supply divided by 365 days (the follow-up period), and an MPR greater than or equal to 80% signified optimal adherence. Beginning with the index prescription, we catalogued each prescription filled during the follow-up period and aligned them chronologically. Using the days of supply associated with each prescription, we determined the expected availability of drug therapy on a given day. For example, if the index prescription had a 30-day supply and the refill was dispensed on the 32nd day after the index fill date, day 31 would be designated as a day without available drug therapy. Days with available drug therapy were deemed as persistent days, and we calculated the proportion of patients persistent on each day during the follow-up period. Daily persistence was then graphed in 5-day increments and stratified by baseline mail pharmacy use. For all calculations, days of supply from early refills were counted from the beginning of the date of exhaustion of the previous prescription, and the days of supply that extended beyond the follow-up period were truncated.
We constructed multivariate logistic regression models to estimate odds of achieving optimal adherence. Variables used in the matching process were included in the multivariable analysis to provide adjusted estimates for the other variables. The following predictors were included in the models: age (in years, as of the index date; <50, 50 to <58, 58 to <65, and 65 ); sex; geographic region; out-of-pocket costs per day of supply of therapy from the index prescription (categorized by quartile); median household income based on zip code of residence (categorized by quartile); previous mail use (yes/no for a mail prescription in the 6-month period prior to their index prescription); and pharmacy benefit design (mandatory mail vs voluntary mail).
Of the 1,238,194 potential study participants, approximately 7% were excluded because of missing or incomplete data, and 34% were excluded because of insufficient eligibility for pharmacy benefits (Table 1). Of those who were eligible for pharmacy benefits and who had valid data, nearly 81% were excluded because they had a previous paid pharmacy claim(s) for a drug in the index class and thus were not therapy initiators. Of the remaining participants, approximately half were ineligible because their index prescription was not for a 30-day supply dispensed at a retail pharmacy, though this proportion varied substantially between the 2 cohorts. From the remaining qualified participant pool, 95% of the mandatory mail initiators were successfully matched to voluntary mail participants.
The final study cohorts included 27,828 (13,914 mandatory and 13,914 voluntary mail) participants. Table 2 presents the comparison of demographic (eg, age, sex, region, etc) and utilization (eg, Pharmacy Risk Group, previous mail use) measures overall and by therapeutic class. The cohorts had similar distributions for the matched characteristics and were similar with regard to household income and geographic region of residence. As expected, the mandatory mail cohort had significantly more previous mail pharmacy use.
Across all therapeutic classes, the voluntary mail cohort achieved higher MPR and optimal (MPR >80%) adherence rates than the mandatory mail cohort during the first year of therapy (Table 2). In the voluntary mail cohort, 36.1% achieved and maintained optimal adherence compared with only 33.6% of those in the mandatory mail cohort, resulting in significantly different median MPR scores (57.4% vs 49.2%; P <.001). These results varied by therapeutic class, however. Among participants initiating antidiabetic therapy, voluntary mail participants had higher adherence than mandatory mail participants for the metformin and sulfonylurea classes, but lower adherence for glitazones. Among participants initiating cardiovascular therapy, voluntary mail participants had higher adherence than mandatory mail participants for the ACE inhibitor, ARB, and statin classes, but lower adherence for PAIs (though adherence in general for this class was substantially greater).
Across all therapeutic classes combined, persistence was similar through the first 60 days (Figure 1). By the 90th day after initiating therapy, the mandatory mail cohort had a substantial drop in the percentage of members who remained on therapy (63.3% vs 56.3%; P <.001), a difference that continued through to day 365 (51.6% vs 46.4%; P <.001). By therapeutic class, these persistence differences were similar to the MPR scores, with voluntary mail participants having higher adherence than mandatory mail participants for sulfonylureas, ACE inhibitors, statins, ARBs, and metformin but lower adherence for glitazones and PAIs (data not shown).
When stratified by previous mail use, the voluntary mail participants exhibited similar persistence during the first year of therapy regardless of previous mail-service pharmacy use (Figure 2). However, mandatory mail participants without previous mail use had significantly lower persistence than mandatory mail participants with previous mail use. By the 90th day, previous mail users had a persistence rate that was 12 percentage points higher than the rate for those with no previous mail use (62.5% vs 50.5%) and this difference expanded to more than 15% by day 365 (55.4% vs 39.8%).
In the multivariable models (Table 3) increasing age, male sex, higher income, and lower out-of-pocket costs were typically associated with increased odds of achieving optimal adherence. Absence of previous mail use was a significant predictor of adherence overall (adjusted odds ratio [OR] 0.51, 95% confidence interval [CI] 0.48-0.54) and was significant in each therapeutic class, with adjusted ORs ranging from a low of 0.36 (ARBs) to a high of 0.65 (ACE inhibitors). After adjusting for previous mail use, voluntary mail was associated with significantly higher adherence than mandatory mail both overall (adjusted OR 0.70; 95% CI 0.67-0.74) and in 5 of the 7 classes: metformin (adjusted OR 0.55; 95% CI 0.46-0.64), sulfonylureas (adjusted OR 0.72; 95% CI 0.57-0.91), ACE inhibitors (adjusted OR 0.74; 95% CI 0.66-0.82), ARBs (adjusted OR 0.69; 95% CI 0.59-0.81), and statins (adjusted OR 0.69; 95% CI 0.63-0.76). Results for glitazones and PAIs did not reach statistical significance.