The Association Between Cost-Related Insulin Rationing and Health Care Utilization in U.S. Adults With Diabetes

Insulin is an essential medication for all people with type 1 diabetes and many with type 2 diabetes (1), but high costs lead to insulin rationing in 17–26% of adults (2–6). Several studies have linked poor adherence to insulin with increased rates of emergency department (ED) visits or hospitalizations among insured adults with diabetes (7–10), but cost-related insulin rationing and its consequences remain understudied. Understanding the association between cost-related insulin rationing and emergency health care utilization is essential to assess the effectiveness of multiple recent policies that cap out-of-pocket spending on insulin.

We used data from the 2021 National Health Interview Survey (NHIS), a nationally representative, cross-sectional household survey of noninstitutionalized U.S. adults. The Yale University Institutional Review Board determined that this study qualified under the exempt category.

We included all respondents who reported a diabetes diagnosis and current insulin use. We defined insulin rationing as any self-report of skipping insulin doses, delaying buying insulin, or taking less insulin than needed to save money in the past year. We defined emergency health care utilization as self-report of one or more ED visits or overnight hospitalizations in the past year. We also examined the number of urgent care visits in the past year.

Additional variables included respondents’ age, sex, race, ethnicity, diabetes type, insurance type, family income as a percentage of the federal poverty level (FPL), whether they saw a health professional for a wellness visit in the past year, and whether they needed medical care but did not get it due to cost in the past year. Based on prior studies that quantified comorbidities using the NHIS and other self-reported data (11–13), we generated a comorbidity score that assigns 1 point for each of 17 self-reported conditions: angina, anxiety, arthritis, asthma, cancer, cirrhosis, congestive heart disease, chronic obstructive pulmonary disease, dementia, depression, diabetes, hepatitis, high cholesterol, hypertension, weak or failing kidneys, myocardial infarction, and stroke.

We classified survey respondents as younger (ages 18–64 years) or older adults (ages ≥65 years), because older adults are covered by Medicare, which likely moderates the relationship between insurance coverage, income, and rationing. For all models, we applied NHIS weights to generate nationally representative data.

To compare utilization between people who reported insulin rationing versus those who did not, we used logistic regressions for emergency health care utilization and zero-inflated negative binomial regressions for the number of urgent care visits. For both regressions, we ran unadjusted models using insulin rationing as the independent variable and fully adjusted models that included age, sex, race, ethnicity, diabetes type, comorbidities, insurance, and income as additional covariates.

In an exploratory analysis, we investigated the relationship between emergency health care utilization and each of the three rationing behaviors. As a sensitivity analysis, we excluded all respondents who reported a coronavirus disease 2019 (COVID-19) diagnosis. All analyses were conducted in R 9.0 software.

The data set analyzed in this study is available on the NHIS website (https://www.cdc.gov/nchs/nhis/2021nhis.htm).

Of the 29,482 respondents to the 2021 NHIS, 982 (3.3%) met our inclusion criteria, representing ∼7.6 million adults with diabetes who use insulin (Table 1). In this study population, median age was 61 (interquartile range 51–71), median family income was between 200% and 250% of the FPL, median number of comorbidities was four. Most (77%) had type 2 diabetes, 44% were privately insured, 5% were uninsured, and 95% had a wellness visit within the past year (Table 1).

The overall prevalence of insulin rationing was 16.5%. Adults who reported insulin rationing were younger (P < 0.001) and were more likely to report not getting needed medical care due to cost (P < 0.001) compared with those not reporting rationing. Among adults who rationed insulin, 50% had private insurance, and 42% had income between 200% and 399% of the FPL. However, differences in rates of insulin rationing by insurance or income did not reach statistical significance (Table 1). ED visit or hospitalization was reported by 38% of younger and 37% of older adults, and at least one urgent care visit was reported by 29% of younger and 23% of older adults.

Among younger adults who reported insulin rationing, 45% had an ED visit or hospitalization in the past year compared with 36% of younger adults who did not ration (P = 0.167) (Fig. 1). No statistically significant differences were found in fully adjusted models for either utilization end point among younger adults (Fig. 2).

Older adults who reported insulin rationing were 2.1 times more likely to have an additional urgent care visit than those not reporting rationing (P = 0.008) (Fig. 2). There was no significant association between rationing and the occurrence of at least one urgent care visit or with an ED visit or hospitalization (Fig. 2).

Among the three rationing behaviors, younger adults skipping insulin doses (vs. those who did not) were more likely to report an ED visit or hospitalization (56% vs. 35%) (Fig. 3). Skipping insulin doses was significantly associated with increased odds of an ED visit or hospitalization in fully adjusted models (odds ratio 2.4, 95% CI 1.2–4.7) (Supplementary Fig. 1). Sensitivity analyses revealed no significant relationship between rationing and use when the 141 respondents diagnosed with COVID-19 were excluded (Supplementary Fig. 2).

In this nationally representative sample of adults with diabetes using insulin, cost-related insulin rationing was not associated with an ED visit or hospitalization or with having an urgent care visit. However, older adults who rationed insulin were more than twice as likely to have an additional urgent care visit than those not reporting rationing.

In contrast to our results, existing studies have demonstrated a positive relationship between nonadherence to insulin and emergency health care utilization (7–10). One potential explanation is that the same financial forces leading people to ration insulin may also deter them from using emergency health care, thereby masking any effect of rationing on adverse health events. Our finding that adults who report cost-related insulin rationing were nearly nine times more likely to report cost-related deferral of necessary medical care supports this explanation. In addition, prior studies (7–10) included only patients who had insurance coverage throughout the study period. Respondents with unknown or absent health insurance, who comprised 16% of our study population, may be more likely to concurrently ration emergency health care services than people with insurance.

In addition, the insulin rationing questions in the NHIS may not capture the full spectrum of circumstances that disrupt insulin access. For example, respondents may have faced barriers to obtaining insulin not related to cost, such as prior authorizations or disruptions in pharmacy supply (14). These participants may have used emergency health care at higher rates than those without barriers to obtaining insulin, but the NHIS would not classify them as rationing insulin due to the narrow scope of the survey questions. On the other hand, skipping, delaying buying, or taking less insulin may each carry different risks. It is possible that skipping doses may precipitate more acute health complications than delaying or taking less insulin, which could explain our finding that skipping insulin doses was associated with significantly higher rates of emergency health care utilization among younger adults.

Older adults who rationed insulin in our study had more urgent care visits (but not ED visits or hospitalizations). This suggests that poor health outcomes secondary to rationing, such as higher A1C levels, may be insidious and present in lower-acuity settings, particularly in older adults. Financial forces may also favor urgent care utilization in this population, as Medicare patients face higher copays for ED than urgent care visits (15). Therefore, emergency health care utilization may not be the best metric with which to capture adverse effects of insulin rationing. Our study found high rates of wellness visits among adults with diabetes using insulin, which represent an opportunity to identify people at risk for insulin rationing and intervene before engagement with emergency health care services.

Study limitations include the cross-sectional data precluding causal conclusions, the small sample constraining the power of these analyses, the self-reported data subject to misclassification and social desirability bias, and the lower NHIS response rate during the COVID-19 pandemic. Additionally, the NHIS health care utilization questions do not include associated diagnoses, so we are unable to discern which encounters were related to complications of diabetes. It is also possible that the COVID-19 pandemic and consequent changes in medication rationing behaviors (16), reductions in emergency health care utilization (17), and disruptions in insulin supplies (18) may have confounded or obscured the relationship between rationing and health care utilization. However, whether pandemic-related deferral of emergency health care may have disproportionately affected people who ration insulin is unknown, and our sensitivity analyses did not show an impact of COVID-19 diagnosis.

High insulin prices have led to the passage of federal policies to improve insulin affordability, principally among Medicare beneficiaries. However, insulin rationing is likely to remain prevalent among those who are uninsured or underinsured, which includes those with private insurance. The 2021 NHIS offered the first opportunity to characterize cost-related insulin rationing on a national scale, but only for a single survey year. More data are needed to explore the impact of national policies on the relationship between cost-related insulin rationing and health care utilization.

Funding. This work was funded by the National Institute of Diabetes and Digestive and Kidney Diseases R01 grant (R01DK129616) and the Yale School of Medicine Office of Student Research. K.J.L. receives research support from Patient-Centered Outcomes Research Institute and other support from Centers for Medicaid & Medicare Services (CMS) to develop and evaluate publicly reported quality measures. L.M.N. receives grant funding for Yale University from the NIH.

Duality of Interest. L.M.N. receives grant funding for Yale University from Medtronic and is a consultant for WebMD, Calm, and Medtronic. K.J.L. receives royalties from UpToDate to write and edit content. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. C.G.B. conducted the analyses and wrote the first draft of the manuscript. C.G.B., B.F.B., L.M.N., and K.J.L. were involved in the conception and design of the study and the analysis of results. B.F.B., L.M.N., and K.J.L. edited, revised, and gave final approval of the manuscript. K.J.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented as an abstract at the AcademyHealth 2024 Annual Research Meeting, Baltimore, MD, 29 June–2 July 2024.

Handling Editors. The journal editors responsible for overseeing the review of the manuscript were Steven E. Kahn and Matthew Crowley.