TOPLINE:

Implementation of a $35 monthly out-of-pocket cap for insulin in 2021 and 2023 was associated with significant reductions in out-of-pocket spending on insulin, modest increases in insulin access, and slightly lower A1c levels in a population of 4.8 million US Medicare beneficiaries with type 2 diabetes (T2D).

METHODOLOGY:

• Medicare capped out-of-pocket insulin costs at $35 per month for some Part D beneficiaries in 2021 and later extended the cap to all beneficiaries in 2023. However, independent evaluations of the cap’s impact on clinical outcomes are lacking.
• Researchers conducted a cohort study using data spanning 2019-2023 to evaluate whether the $35 monthly cap was associated with changes in out-of-pocket spending on insulin, insulin use, glycemic control, and other clinical outcomes among Medicare beneficiaries with T2D.
• They analyzed 4.8 million beneficiaries with at least 1 Medicare-paid pharmacy claim during the quarter, with a median quarterly cohort size of 1,393,402 patients; in the first quarter of 2019, 51.1% were men, and 51.3% were aged 65-74 years.
• The outcomes were evaluated in three periods: the baseline pre-cap period in January 2019, when the cap applied to some beneficiaries in January 2021, and when it applied to all beneficiaries in January 2023.
• The mean levels of A1c were evaluated in a subset of 207,197 patients with linked data from electronic health records.

TAKEAWAY:

• The mean quarterly out-of-pocket spending on insulin among users was $192.66 at baseline, which was reduced by $47.90 after the 2021 cap and by further $58.59 after the 2023 universal cap (P < .001 for both). The share of insulin users paying more than $35 per month fell from 62.14% at baseline to only 0.11% by end of 2023.
• Among existing insulin users, the mean daily insulin use was 57.38 units at baseline, which increased by 1.36 units after 2021 and by a further 1.16 units after 2023; among all patients with T2D, the mean daily insulin use rose by 0.23 units after 2023 (P < .001 for all).
• The mean A1c level decreased by 0.06% from a baseline of 7.28% after 2023 (P < .001), with corresponding decreases in the proportion of patients with poorer glycemic control (A1c level > 9%) and increases in those with better glycemic control (A1c level < 7%).
• The quarterly rates of severe hypoglycemic events rose modestly during the study period; no changes were observed in the incidence of diabetic ketoacidosis or hyperosmolar hyperglycemic state.

IN PRACTICE:

“Ultimately, reducing out-of-pocket costs for medications is a laudable goal but the consideration of multiple other factors is essential to assess the full potential of these important policy initiatives, to understand both the advantages and disadvantages of copay caps, and to maximize their benefit at a population level,” the experts wrote in an editorial accompanying the journal article.

SOURCE:

This study was led by Dongzhe Hong, PhD, Brigham and Women’s Hospital and Harvard Medical School, Boston. It was published online on April 06, 2026, in JAMA Internal Medicine.

LIMITATIONS:

The lack of a comparison group of patients unaffected by the $35 cap policies restricted drawing causal interpretations. Concurrent events, such as the voluntary list prices announced by the three major insulin manufacturers in March 2023, may have influenced out-of-pocket costs. The database represented a large convenience sample of US pharmacy and medical claims, limiting generalizability of the results to the entire Medicare population.

DISCLOSURES:

This study was supported by grants from Arnold Ventures to Brigham and Women’s Hospital and by internal funding from IQVIA. Several authors disclosed having financial relationships including employment; stock ownership; and receipt of personal fees, grants, consulting fees, and royalties with various institutions and pharmaceutical companies, including the funding agency. One author also reported serving as an expert witness in a litigation concerning pharmacy benefit managers and insulin, outside the submitted work.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.