Statistics from Altmetric.com
The adoption of the Affordable Care Act (ACA)1 in the USA expanded health insurance for low-income Americans and took two main forms: Medicaid expansion in some states and subsidized private health insurance through insurance exchanges available in all states, with deep subsidies for persons with incomes from 138% to 250% of the federal poverty limit (FPL) in Medicaid expansion states and from 100% to 250% of the FPL in non-expansion states. Prior studies found a statistically significant slightly negative2 effects of the ACA on diabetes diagnoses and controversial (from insignificantly slightly positive3 to significantly positive4) effects on diabetes therapies at county and state levels. We examined the effect of both forms of ACA reform on the improvement of diabetes diagnostics and management in low-income patients who had access to healthcare before the ACA expansion (2011–2013).
We used electronic health records (EHR) from 11 major academic health systems in 8 states in the USA (Illinois, Iowa, Wisconsin, Kansas, Nebraska, Missouri, Texas, Indiana). The sample (see table 1 for demographics) was limited to patients aged 55–74 over 2011–2018 who used care (any encounter type) at the study facilities at least once in the pre-expansion period. Due to inconsistent depiction of insurance status in EHR, patient residence in a socially deprived5 census tract (see online supplemental appendix for details) was used as proxy for persons who were more likely to gain insurance under the ACA. Therefore persons aged 55–64 from the socially deprived census tracts were the treatment group. Persons aged 65–74 from socially deprived census tracts were the control group because they had Medicare insurance. For each age group, we studied the per cent of patients of interest with newly detected diabetes6 and the per cent of patients with prevalent diabetes receiving diabetes-related medications before (2011–2013) and during (2014–2018) the ACA expansion. Combined age discontinuity and difference-in-difference research design was employed.
Different from individuals who had no access to healthcare2 before the ACA, our sample of patients from socially deprived tracts shows no increase in rates of newly diagnosed diabetes (figure 1). An insignificant drop of −0.72 (95% CI −3.22 to 1.77) in newly diagnosed diabetes for the treated group was detected. We have to note the identification of diabetes in the sample was not limited to ambulatory settings. This makes us conclude that the study centers may have already been using all available resources to accurately diagnose diabetes before 2014, including for low-income patients. Therefore, the ACA did not lead to an improvement in diagnostics for our sample. The decline in new diabetes cases may be a positive effect of the improved access to other preventive care7 services and medications during the ACA.
We also assessed whether the ACA led to low-income persons with prevalent diabetes having better access to diabetes medications. We detected an insignificant increase of 0.21 (95% CI −2.10 to 2.52) in the prescription for diabetes medications in the treatment group. The observed trend for the prescribed diabetes medications matched the 2010–2016 dispensed medication trend detected with the Medicaid State Drug Utilization Data.3 Overall, the reported increase in diabetes medication due to the ACA tended to be modest if a ‘per enrollee’-like measure was selected as opposed to an ‘all prescriptions’4 one.
In summary, we would like to stress that selected health outcomes are not doing the ACA justice and, as a result, underestimating the presumed improvement in the health services for low-income patients-clients of the academic centers before the ACA implementation. Such patients would face a different level of improvement in access to care comparing with ones who were completely isolated from the healthcare system before the policy took place.
Data availability statement
All data relevant to the study are included in the article or uploaded as online supplemental information.
The CAPRICORN and GPC institutional review boards approved this study (IRB#16030901 and IRB#STUDY00004015, respectively).
The Rush team would like to acknowledge the contribution of Ms Ekta Kishen and Mr J Alan Simmons of the Bioinformatics/Biostatistics Core for fulfilling the data queries used in these analyses. The Loyola team would like to acknowledge the contribution of Ms Susan Zelisko of Information Technologies for conducting the data queries used in these analyses. The Northwestern team would like to thank Ms Anna Nasiri for providing valuable insight about the Northwestern Medicine EHR. The UTSW team would like to acknowledge the contribution of Mr Phillip Reeder for conducting the data queries used in this project. Data from Indiana were produced by the Regenstrief Data Services team. Daniel Hood, the Regenstrief Project Manager for GPC, led the data extraction processes for this project.
Contributors BSB, ANK and AF conceived and designed the study. All authors interpreted the data, critically reviewed the manuscript for important intellectual content and approved the final manuscript. AF and BSB developed the analytical plan. AF performed the statistical analyses. AF and BSB drafted the initial manuscript. ANK supervised the study. AF, ML, XS, LRW, JRM, KO, AS, EC, LGC, RCS, JCM, UT, VM, ASMM, DG, FA, AP, WET, TWM and ANK contributed to data acquisition. AF and BSB contributed to analysis of data. RCS, ML, XS, JRM, KO, AS, EC, JCM, UT, ASMM, WET, TWM, LR-T, PJE and ANK contributed to interpretation of data and manuscript writing.
Funding This work is supported by the Centers for Disease Control and Prevention (CDC) and the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases (grant no 5U18DP006120-05-00) as part of the Natural Experiments for Translation in Diabetes 2.0 (NEXT-D2) collaborative. The study funders had no role in the study design, data collection, data analysis, data interpretation or the writing of this article. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.