Switching to cheaper drugs or stopping use: Antipsychotic drug use changes and adverse events after entering the Medicare part d coverage gap

Background: The Medicare Part D program provides partial protection for antipsychotic drugs, i.e., mandated formulary inclusion but no restrictions on cost-sharing. The cost-sharing amounts can be considerable with standard Part D benefits including both multiple tiers and a coverage gap. Prior to 2008, second-generation antipsychotics (2GAs) also all were brand drugs, i.e., no generic availability. Therefore, beneficiaries using 2GAs could face substantial out-of-pocket costs upon entering the coverage gap, switch to cheaper generic first-generation antipsychotics (1GA), or discontinue use. Aims: We examined factors associated with discontinuing use or switching from a 2GA to generic 1GA after reaching the coverage gap threshold in 2007, and the association between changes in antipsychotic use and adverse events. Methods: We examined non-institutionalized beneficiaries enrolled in Medicare Advantage drug plans with at least one 2GA dispensed in 2006 and 2007 prior to reaching the coverage gap threshold; we examined beneficiaries reaching the threshold before October (n=8,275). Half did not have coverage for brand drugs during the gap, and half had coverage because of low-income subsidies. We used logistic regression models to examine factors associated with switching to 1GAs or discontinuing use, and Poisson regression models to examine the association between changes in drug use and adverse events (hospitalizations and emergency department visits) during the gap. Findings: After losing coverage in the gap, 3% of beneficiaries switched to 1GAs and 24% stopped using antipsychotics. Compared with beneficiaries with gap coverage, beneficiaries without coverage were more likely to stop using antipsychotics (OR=3.07;95%CI:2.67-3.53) or switch to 1GAs (OR=6.84;95%CI:4.31-10.86) after reaching the gap threshold. Among beneficiaries with a gap, those with bipolar disorder (vs. no mental health diagnoses), those using 2GAs only (vs. some 1GA use) in the pre-gap period, and those under vs. over age 65 (i.e., disabled beneficiaries) were more likely to discontinue antipsychotic use in the gap. Drug changes (discontinuing or switching) after losing coverage were associated with increased adverse events during the gap (RR=1.11;95%CI:1.03-1.18). Conclusions: Beneficiaries facing the prospect of high out-of-pocket costs because of the coverage gap frequently made changes, including switching to cheaper drugs or discontinuing use, compared with beneficiaries without the gap. Vulnerable subpopulations, including beneficiaries with bipolar disorder or permanent disabilities, were more likely to discontinue antipsychotic use. These changes in drug use were associated with increased adverse events. Part D drug protections should guard against all access restrictions, including financial barriers.