Insulin Fills Declined as Costs Rose, With Persistent Gaps by Income: Elise S. Tremblay, MD, MPH

In part 1 of an interview, Elise S. Tremblay, MD, MPH, lead investigator of "Trends in Insulin Out-of-Pocket Costs and Use Disparities, 2008-2021," published in the August 2025 issue of The American Journal of Managed Care®, shares the study's inspiration, objectives, and key findings.

Tremblay is a pediatric endocrinologist at Boston Children's Hospital and Boston Medical Center.

Transcript

How did the rise of high-deductible health plans with savings options influence your decision to study insulin out-of-pocket costs and usage trends over time?

Insulin is an essential life-saving, life-sustaining medication for people who have type 1 diabetes, and it often becomes a critical medication for people with type 2 diabetes, as well. When you have a medication like that that is essential, and a medication like that that has risen in price so substantially, there's an argument to be made that there might be changes in people's use patterns based on how much they have to pay for the medication.

Given that we know that's a trend overall, we identified the fact that there were some changes in health plan structure that made it such that people might be paying more out of pocket, meaning what their insurance doesn't cover for particular medications, such as insulin. This becomes particularly impactful when a medication is particularly expensive.

Recognizing that there were changes in health plan structures and the amount of costs that people would be facing based on how their employer structured their health plans, we thought it would be important to look at how those differently structured plans might impact people's use of certain life-sustaining medications like insulin.

In particular, when you have a high-deductible health plan, that means that individuals are responsible for a higher proportion of their health plan costs, but they typically pay a lower premium. Then, a high-deductible health plan with a savings option suggests that individuals are paying even more out of pocket.

When we could narrow down on that particular health plan structure, we could see if paying more out of pocket impacted their use of certain medications.

Could you further explain the primary objectives of your study and the methods used to investigate them?

We used a database that [includes] all privately insured individuals across all 50 states and all of the medical claims for those individuals, in particular, [for] calendar years. Using the claims-based data, we were able to ask the questions, how much are individuals paying for their medications? How many fills of those medications are they getting per year? We could use those 2 questions to help us understand patterns over time in both the number of fills and the cost.

In order to do that, what we did is we first identified a population of individuals in the claims database that we believe, based on their preponderance of claims, had diabetes. We ensured that they were covered by the plan for a washout period to make sure we were capturing all of their fills of medications and true time with the diagnosis. Then, we annualize the fills, so we looked at the number of fills that would cover a 30-day supply per year. A year has 12 subsets of 30 days, so if you're only filling nine 30-day supplies per year, for instance, you could imagine that you're short [on] medication for a total of about 25% of the year.

By annualizing the fills and looking at the number of 30-day supplies per year, we were able to get a sense of people's adherence to treatment plans and ability to fill the medications. At the same time, in the same patients, we looked at their out-of-pocket costs per year. We couldn't draw direct conclusions about the impact of out-of-pocket costs on the fills. We could only provide the data points for each question, but we did notice patterns in terms of rises in the out-of-pocket costs and then changes in the number of fills per year.

What were the main findings? Were there any that surprised you?

We had a few primary findings from our study. The first was that the out-of-pocket costs per year, ranging in the data period that we studied, which was 2008 to 2021, initially underwent a relatively steep rise between about 2008 and 2014. That rise was more apparent among individuals who had savings option plans than among those who did not.

Following that time period, we experienced a relative stabilization of costs followed by a minimal decline. When we looked at the same comparison but among 30-day fills rather than cost, we saw that during the period of time in which the costs were rising, there was a slight decline in the number of fills by the end of that period. Then, when the cost stabilized out, and even decreased, we noticed an increase in the number of fills per year.

We also broke this question down by poverty level. In the claims-based data, we don't have individual demographic characteristics, but we have zip code-level characteristics. We use a data source called the American Community Survey to look at what the population-level poverty is in particular zip codes. We look at the percentage of households that are experiencing certain income levels to define poverty.

When we split the population up that way, between higher-income and lower-income individuals, we saw consistently across time that higher-income individuals with diabetes filled more 30-day supplies of insulin per year than their lower-income counterparts; there's a consistent disparity in those.

For me, the most significant and somewhat surprising finding was how consistent the disparity was between higher-income and lower-income individuals. Although both of their rates over time in terms of the number of 30-day fills per year did vary over the study period, the gap between the 2 of them was highly consistent.