Comment on “Association Between the Part D Coverage Gap and Adverse Health Outcomes”

To the Editor: We read with great interest the article by Polinski et al.1 which adds important information to the growing body of literature on the effect of the Medicare Part D coverage gap. The study concludes that older adults facing the coverage gap are at greater risk of temporarily discontinuing their medications than persons with full insurance coverage, although the study results show no significant effect of medication discontinuation on the risk of adverse health outcomes. Despite the use of robust adjustment and sensitivity analysis techniques to evaluate the latter association, a number of methodological limitations of the study should be carefully considered. Studying the effect of the coverage gap on adverse health outcomes through medication discontinuation as a mediator requires sound relations between the coverage gap and drug discontinuation on one hand and between drug discontinuation and health outcomes on the other (Figure 1). We believe that the latter association was unlikely to be present in the Polinski et al. study because of the short follow-up period and the absence of a direct link between individuals who discontinued their medications and those who experienced adverse health outcomes. The effect of the coverage gap on health outcomes through drug discontinuation could have been estimated using mediation analysis.2 Mediation analysis in the framework of counterfactuals3-6 can test whether the greater risk of adverse health outcomes caused by the coverage gap is entirely mediated through medication discontinuation and, if not, what proportion of increased risk is directly caused by pathways other than medication discontinuation. The method of causal mediation analysis using propensity scores has been described and applied in previous studies.7, 8 Other study design features may have contributed to masking the effect of the coverage gap on health outcomes. The authors considered the discontinuation of several medication classes that treat several diseases. Among them were medications to treat cardiovascular diseases that can directly affect death rates and cardiovascular outcomes. Cardiovascular drug classes include a range of drugs that vary considerably in pharmacological mechanism and price. It is more likely that an individual will discontinue an expensive medication than a cheaper one. Discontinuation of an expensive medication (e.g., statins) may not lead to short-term adverse health outcomes, whereas skipping doses of less-expensive medications (e.g. diuretics) may be more likely to elicit immediate effects. By measuring discontinuation of any medication, the observed effect on health outcomes may have been attenuated. A more-valid approach would be to examine health outcomes of those who discontinue specific classes of medications—especially those with short-term adverse effects. Furthermore, the exposure period to the coverage gap was brief for most participants. To evaluate the long-term effect of medication discontinuation, exposure should be redefined as a cumulative or continuous measure of the total number of days in the gap over subsequent years. Finally, by conducting subanalyses for disease-specific outcomes, it is not clear whether the study was sufficiently powered to detect significant differences in outcomes over the short coverage gap exposure period. To summarize, the study demonstrates a significant increase in medication discontinuation rates due to the coverage gap, but its effect on short- and long-term health outcomes remains unclear. Methodological aspects such as the ability to link those who discontinued to health outcomes, the study of specific classes of medications that can cause short-term adverse effects if discontinued, and the redefinition of the exposure into a cumulative measure need to be addressed to discern the true clinical effect of the coverage gap. Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper. Author Contributions: All authors contributed equally in the preparation of manuscript. Sponsor's Role: None.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.