Trends in Insulin Out-of-Pocket Costs and Use Disparities, 2008-2021

ABSTRACT

Objective: To assess trends in insulin out-of-pocket (OOP) costs, use, and disparities among commercially insured patients from 2008 to 2021.

Study Design: Retrospective time series from a national insurance database, with members in all US states, including data from 2008 to 2021.

Methods: Insulin OOP costs and 30-day equivalent fills per year were quantified among insulin users aged 12 to 64 years, stratified by income (low- vs high-poverty zip code) and health plan type (high-deductible health plans with savings options [HDHP/SO] vs not). Participants were commercially insured insulin users aged 12 to 64 years with at least 1 full enrollment year. Characteristics of interest for disparities analysis included income level (low- vs high-poverty zip code) and health plan type (HDHP/SO vs non-HDHP/SO plan).

Results: After increases in adjusted mean annual insulin OOP costs from 2008 ($221 per non-HDHP/SO member and $313 per HDHP/SO member) to 2014 ($280 and $496, respectively), HDHP/SO members had persistent relative reductions in insulin use. In 2014, HDHP/SO members had 0.17 fewer annual fills, a disparity that increased until 2019 (0.79) before decreasing slightly by 2021 (–0.55). Lower-income members consistently had fewer insulin fills.

Conclusions: Insulin OOP cost reduction policies would be more efficient if they targeted HDHP/SO plan members and low-income patients.

Am J Manag Care. 2025;31(8):In Press

Takeaway Points

This retrospective, claims-based time series demonstrated that insulin out-of-pocket (OOP) costs peaked in 2014-2017, then declined by 2021, with members of plans with savings options (SO) having consistently higher OOP costs than those in non-SO plans. Insulin use declined more rapidly among SO plan members than among non-SO members in 2014 and afterward, whereas lower-income members consistently had lower insulin use than higher-income members. Lower rates of insulin use among lower-income and SO plan members compared with their counterparts suggest that insulin OOP cost reduction policies should target these subgroups.

• Plan members with health savings accounts and/or health reimbursement arrangements experience different costs and use different amounts of insulin than their counterparts without savings options.
• Plan members from lower-income neighborhoods consistently use less insulin than those from higher-income neighborhoods.

For individuals with diabetes, insulin is often critically important to prevent short- and long-term complications. However, multiple factors limit optimal use of insulin, including high out-of-pocket (OOP) costs. In recent years, employers have increasingly adopted high-deductible health plans (HDHPs) that require patients to pay $1000 or more per year before additional coverage begins.1 HDHPs often include health savings options (SOs)—health savings accounts (HSAs) or health reimbursement arrangements (HRAs)—that allow use of tax-advantaged funds for health care payments.2

Previous research suggests that lower-income enrollees and those who have SO plans are more susceptible to the unintended effects of HDHPs.3-5 These effects include increased high-severity emergency department and inpatient stays due to acute complications among low-income patients with diabetes and HDHPs with or without SO plans. Some of these complications may be caused by cost-related barriers to insulin use.

Insulin OOP costs increased from 2006 to 2014 among commercially insured patients with HDHP/SO plans, then declined somewhat in 2015-2016.6 Recent national and state health policies have sought to reduce insulin cost barriers through OOP caps. Updated trends in insulin OOP costs and use, accounting for enrollment in SO plans and income, could help policy makers target cost-sharing reductions to populations most burdened by insulin payments.

In this study, we used US national claims data to assess trends in insulin OOP costs, use, and related disparities by health plan type and income. We hypothesized that insulin OOP costs continued to decline through 2021 and that patients with HDHP/SO plans and lower incomes would demonstrate relatively lower insulin use over time.

RESEARCH DESIGN AND METHODS

Data Source and Study Population

We used a health insurance claims database with subscribers from a large national commercial (and Medicare Advantage) health plan. Data included member enrollment, demographics, and medical and pharmacy claims. We restricted the study to people with commercial insurance enrolled from January 2007 to September 2021. Using a standard algorithm,7 we identified patients aged 12 to 64 years with diabetes, then restricted the sample to enrolled months during and after a member’s first insulin fill. See the eAppendix Figure (eAppendix available at ajmc.com) for more details regarding sample creation.

This study was approved by the Harvard Pilgrim Health Care and Duke University institutional review boards.

Outcomes and Statistical Analyses

The primary outcomes were OOP costs (deductibles plus co-payments) and number of 30-day equivalent insulin fills per insulin user per benefit year. All estimates were adjusted for changes in age group, sex, and geographical area of residence over time (Table). Using a vendor-supplied variable, we categorized HDHP/SO members as those with an HSA/HRA. We categorized enrollees as lower income if they lived in zip codes where at least 15% of households lived below the poverty level, based on American Community Survey (ACS) data.8

We first plotted unadjusted mean insulin OOP costs and number of 30-day fills by year, both overall and stratified by HDHP/SO and income level. We then used generalized estimating equations9 to fit an individual-level negative binomial regression with the log link function to estimate all outcomes, the Huber-White estimator for unbiased standard errors, and the Stata margins10 command for adjusted annual rates. All analyses were conducted using SAS 9.4 (SAS Institute Inc) or Stata 14.2 (StataCorp LLC).

Data and Resource Availability

The data sets generated during and/or analyzed during the current study are not publicly available due to the requirements of our data use agreement, but analytic data are available from the corresponding author upon reasonable request.

RESULTS

The Table presents population characteristics in selected years. Group demographics remained comparable over time, with slight decreases in non-White and lower-income zip codes. The percentage of insulin users with HDHP/SO plans increased from 6% in 2008 to 27% in 2021; the percentage from low-income zip codes remained relatively stable.

Adjusted mean annual insulin OOP costs increased from 2008 (HDHP/SO: $313; 95% CI, $296-$329; non-HDHP/SO: $221; 95% CI, $219-$223) to 2014 (HDHP/SO: $496; 95% CI, $485-$508; non-HDHP/SO: $280; 95% CI, $277-$283) among insulin users, with the steepest rise from 2012-2014 among patients with HDHP/SO plans (Figure [A] and eAppendix Tables 1 and 2). Among non-HDHP/SO patients, OOP costs began to decrease, down to $257 in 2015 (95% CI, $254-$260) and $197 in 2021 (95% CI, $194-$200) (eAppendix Tables 1 and 2). HDHP/SO members experienced relatively stable insulin OOP costs between 2014 and 2019, followed by decreases in 2020 ($451; 95% CI, $439-$464) and 2021 ($424; 95% CI, $412-$436).

Mean adjusted annual insulin 30-day fills were similar in 2008 among HDHP/SO (7.43; 95% CI, 7.20-7.67) and non-HDHP/SO (7.37; 95% CI, 7.31-7.43) members (adjusted difference, 0.05; 95% CI, –0.11 to 0.20). Adjusted differences remained statistically indistinguishable until 2014 (Figure [B]), with HDHP/SO plan members using significantly less insulin per year after that. Both groups increased insulin use between 2019 (HDHP/SO: 7.47; 95% CI, 7.37-7.58; non-HDHP/SO: 8.28; 95% CI, 8.23-8.34) and 2021 (HDHP/SO: 8.00; 95% CI, 7.89-8.12; non-HDHP/SO: 8.55; 95% CI, 8.48-8.62). Adjusted differences in fills also declined somewhat over those years (eAppendix Tables 1 and 3).

Mean adjusted 30-day insulin fills were stable through 2018 among lower-income (range, 6.95 [95% CI, 6.80-7.10] to 7.21 [95% CI, 7.09-7.32]) and higher-income (range, 7.72 [95% CI, 7.63-7.81] to 7.92 [95% CI, 7.85-7.98]) patients, then increased among both subgroups (Figure [C]). However, lower-income members consistently used less insulin per year than their higher-income counterparts (eAppendix Table 4). By 2021, lower-income patients filled 8.01 (95% CI, 7.89-8.13) 30-day supplies per year, and higher-income patients filled 8.54 (95% CI, 8.48-8.61).

CONCLUSIONS

Among commercially insured insulin users, insulin OOP costs were substantially higher for HDHP/SO plan members from 2008 to 2021. Insulin OOP costs among HDHP/SO members increased rapidly from 2008 to 2014; thereafter, HDHP/SO patients used less insulin than their counterparts. This disparity decreased slightly in 2020-2021 when insulin OOP costs declined moderately. Importantly, enrollment in HDHP/SO plans increased rapidly (from 6% to 27%) during the study period, implying that observed disparities are becoming more impactful on a population level. Lower insulin use among HDHP/SO plan and lower-income enrollees may be due to higher insulin OOP costs in those subgroups.

Insulin OOP cost trends are consistent with those seen in our previous study, which demonstrated declines across health plan types from 2013 to 2017.6 Findings through 2021 confirm that OOP costs have continued to decrease.

Previous research about cost-related underuse of insulin has been limited. A survey-based study suggested that up to 25% of patients, especially those with low income, reported not using insulin as prescribed due to its cost.11 National Health Interview Survey respondents with diabetes enrolled in HDHPs reported more frequent cost-related nonadherence than those in traditional plans, particularly insulin users.12 We found more modest income-related disparities in insulin use than reported in survey-based studies. Differences might be related to our use of insulin fills rather than self-reports, our large sample size (> 1 million patient-years), and/or demographic differences in populations studied.4,10

Our study adds to previous research by identifying SO plans as key drivers of insulin use disparities, quantifying disparities by income more precisely, and demonstrating recent continuation of concerning disparities. Rising insulin costs have generated significant policy attention in recent years. More than half of US states have enacted caps on insulin OOP costs for commercially insured residents whose employers do not self-insure. The 2022 Inflation Reduction Act codified a $35 cap for Medicare recipients.13 Our data suggest that policies could be more efficient by targeting insulin users with lower incomes or higher insulin OOP costs. Policies that reduce insulin OOP costs for higher-income patients who have less difficulty affording insulin contribute to increased health insurance premiums for all. Such blanket policies could also harm efforts to reduce insulin prices.13

Our study has a few limitations. Health insurance claims document prescription fills but not use. This observational study cannot establish a causal link between costs and use. The sample we included is unlikely to represent the overall US population by race/ethnicity,14 and our use of ACS zip code race/ethnicity, education, and income data is subject to misclassification at the person level.

Following pronounced increases in insulin OOP costs from 2008 to 2014, HDHP/SO members experienced persistent reductions in insulin use compared with non-HDHP/SO members. Income-related disparities in insulin use were observed in all years. These findings suggest the importance of policies to enhance insulin affordability among low-income individuals with high cost-sharing health plans.

Author Affiliations: Division of Endocrinology, Department of Pediatrics, Boston Children’s Hospital (EST), Boston, MA; Department of Pediatrics, Harvard Medical School (EST), Boston, MA; Department of Population Medicine, Harvard Pilgrim Health Care Institute (SA, FZ, DR-D, JFW), Boston, MA; Harvard Medical School (FZ, DR-D, JFW), Boston, MA; Duke University Division of General Internal Medicine and the Duke-Margolis Institute for Health Policy (JFW), Durham, NC.

Source of Funding: This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) of the National Institutes of Health (NIH) under award numbers K12DK094721 and K23DK133685. This study was additionally funded by grants 1U18DP006122 and 1U18DP006527 from the CDC and NIDDK. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Author Disclosures: Dr Tremblay reports consulting for a company developing a new lab assay for ketones. Dr Argetsinger reports receiving grants from the CDC/NIDDK (5U18DP006122) and the NIDDK Health Delivery Systems Center for Diabetes Translational Research (1P30-DK092924). Dr Zhang receives income from Pfizer and GSK for serving as the statistician on safety studies to satisfy FDA requirements. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (DR-D, JFW); acquisition of data (SA, JFW); analysis and interpretation of data (EST, FZ, DR-D, JFW); drafting of the manuscript (EST); critical revision of the manuscript for important intellectual content (FZ, DR-D, JFW); statistical analysis (EST, FZ); obtaining funding (JFW); administrative, technical, or logistic support (SA); and supervision (SA, DR-D, JFW).

Address Correspondence to: Elise S. Tremblay, MD, MPH, Division of Endocrinology, Boston Children’s Hospital, 333 Longwood Ave, 6th Floor, Boston, MA 02115. Email: elise.tremblay@childrens.harvard.edu.

REFERENCES

1. Kullgren JT, Galbraith AA, Hinrichsen VL, et al. Health care use and decision making among lower-income families in high-deductible health plans. Arch Intern Med. 2010;170(21):1918-1925. doi:10.1001/archinternmed.2010.428

2. Cohen RA, Zammitti EP. High-deductible health plan enrollment among adults 18–64 with employment-based insurance coverage. National Center for Health Statistics. August 2018. Accessed October 24, 2023. https://www.cdc.gov/nchs/data/databriefs/db317.pdf

3. Wharam JF, Zhang F, Eggleston EM, Lu CY, Soumerai S, Ross-Degnan D. Diabetes outpatient care and acute complications before and after high-deductible insurance enrollment: a Natural Experiment for Translation in Diabetes (NEXT-D) study. JAMA Intern Med. 2017;177(3):358-368. doi:10.1001/jamainternmed.2016.8411

4. Wharam JF, Zhang F, Eggleston EM, Lu CY, Soumerai SB, Ross-Degnan D. Effect of high-deductible insurance on high-acuity outcomes in diabetes: a Natural Experiment for Translation in Diabetes (NEXT-D) study. Diabetes Care. 2018;41(5):940-948. doi:10.2337/dc17-1183

5. Garabedian LF, Zhang F, LeCates R, Wallace J, Ross-Degnan D, Wharam JF. Trends in high deductible health plan enrolment and spending among commercially insured members with and without chronic conditions: a Natural Experiment for Translation in Diabetes (NEXT-D2) study. BMJ Open. 2021;11(9):e044198. doi:10.1136/bmjopen-2020-044198

6. Meiri A, Zhang F, Ross-Degnan D, Wharam JF. Trends in insulin out-of-pocket costs and reimbursement price among US patients with private health insurance, 2006-2017. JAMA Intern Med. 2020;180(7):1010-1012. doi:10.1001/jamainternmed.2020.1302

7. Ross-Degnan D, Wallace J, Zhang F, Soumerai SB, Garabedian L, Wharam JF. Reduced cost-sharing for preventive drugs preferentially benefits low-income patients with diabetes in high deductible health plans with health savings accounts. Med Care. 2020;58(suppl 6, suppl 1):S4-S13. doi:10.1097/MLR.0000000000001295

8. American Community Survey: Public Use Microdata Sample (PUMS). US Census Bureau. Accessed October 24, 2023. https://www.census.gov/programs-surveys/acs/microdata.html

9. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73:13-22. doi:10.1093/biomet/73.1.13

10. StataCorp. Stata 14 Base Reference Manual. Stata Press; 2016.

11. Herkert D, Vijayakumar P, Luo J, et al. Cost-related insulin underuse among patients with diabetes. JAMA Intern Med. 2019;179(1):112-114. doi:10.1001/jamainternmed.2018.5008

12. Rastas C, Bunker D, Gampa V, et al. Association between high deductible health plans and cost-related non-adherence to medications among Americans with diabetes: an observational study. J Gen Intern Med. 2022;37(8):1910-1916. doi:10.1007/s11606-021-06937-9

13. Wharam JF, Rosenthal MB. The increasing adoption of out-of-pocket cost caps: benefits, unintended consequences, and policy opportunities. JAMA. 2023;330(7):591-592. doi:10.1001/jama.2023.9455

14. Dahlen A, Charu V. Analysis of sampling bias in large health care claims databases. JAMA Netw Open. 2023;6(1):e2249804. doi:10.1001/jamanetworkopen.2022.49804