Private Insurance Coverage for Diabetes Before and After Enactment of the Preexisting Condition Mandate of the Affordable Care Act, 2005–2016

Abstract

Objectives. To examine private insurance coverage for persons with diabetes before and after enactment of the preexisting condition mandate of the Affordable Care Act (ACA) in the United States.

Methods. We conducted a nationwide study in adults aged 20 to 59 years with private health insurance with the Clinformatics Data Mart Database (2005–2016). We used fixed-effects negative binomial regression to evaluate differences in pre–post mandate trends.

Results. There was a 4% decline in prevalence rates of type 1 diabetes in adults with private health insurance before the mandate and an 11% increase afterward (P < .001). Coverage increased to the greatest extent (–6% before, +20% after) in those aged 50 to 59 years (P < .001). For type 2 diabetes, there was a significant decline in prevalence before the mandate, which increased afterward in those aged 40 to 49 years (–4% before, 3% after; P = .031) and 50 to 59 years (–6% before, 15% after; P < .001).

Conclusions. Adults with diabetes may have benefited in obtaining private health insurance after implementation of the preexisting condition mandate of the ACA.

Public Health Implications. Efforts to limit enforcement of these protections are likely to contribute to setbacks in access to care.

In June 2018, the Justice Department filed a brief stating that, within the Affordable Care Act (ACA), the “provisions containing the individual mandate as well as the guaranteed-issue and community-rating requirements will all be invalid beginning on January 1, 2019.”^1(p20) To date, this issue is still under litigation. These provisions include guarantees that insurers cannot deny coverage to people with preexisting conditions (“guaranteed issue”) and that insurers must provide affordable premiums, adjusted only for age, tobacco use, and geographic area (“community rating”).² Both provisions went into effect on January 1, 2014.²

Approximately 52 million adults in the United States (27% of those younger than 65 years) have preexisting conditions that would potentially make them uninsurable without these protections.³ Before the enactment of the ACA, diabetes mellitus was among the list of conditions for which insurers could deny health care coverage.³ A study from Johns Hopkins University indicated that, before the ACA, 25% of individuals with diabetes reported being denied health insurance or offered insurance at a prohibitive rate.⁴ Given recent federal interest in modifying the preexisting condition mandate, we designed a study to examine persons with diabetes with private health insurance before and after the mandate. We hypothesized that the prevalence of adults with diabetes with private health insurance declined before the mandate and increased after enactment of the preexisting condition mandate on January 1, 2014.

METHODS

We used a database from a nationwide private health insurer in the United States with approximately 75 million members (Clinformatics Data Mart Database) from January 1, 2005, to June 30, 2016. It contained integrated health information regarding outpatient and inpatient services, membership, diagnoses and procedures, medications, and laboratory data. Enrollees of working age were eligible for this study and we stratified data by decade (ages 20–29, 30–39, 40–49, and 50–59 years at the time of enrollment). We excluded persons with Medicare coverage.

We used a validated algorithm to determine individuals with type 1 diabetes.⁵ We determined individuals with type 2 diabetes by using standard International Classification of Diseases, Ninth Revision, and International Classification of Diseases, Tenth Revision, codes^6,7 from inpatient and outpatient files. We calculated prevalence rates of diabetes by year, with the number of individuals with diabetes as the numerator (existing and incident cases) and the person-years of enrolled beneficiaries as the denominator. To evaluate trends, we used fixed-effects (conditional) negative binomial regression with robust variance estimates, offset by the natural logarithm of person-years.⁸ We calculated changes in the prevalence of diabetes, comparing the time trend before the mandate (2005–2013) to the time trend after the mandate (2014–2016). We evaluated differences in pre–post mandate slopes with an interaction term. We also evaluated the possibility of age discrimination, with the hypothesis that older adults would be more likely to be denied coverage than would younger adults. To test this, we used a 3-way interaction term (years∗pre–post∗age group). We set α at 0.05, 2-tailed. We conducted analyses in Stata/MP version 15.1 (StataCorp LP, College Station, TX).

RESULTS

From 2005 to 2016, there were 1 366 309 individuals with type 2 diabetes and 104 991 individuals with type 1 diabetes. Age at first enrollment was evenly distributed for patients with type 1 diabetes (26.9% in those aged 20–29 years, 25.3% in those aged 30–39 years, 25.1% in those aged 40–49 years, and 22.7% in those aged 50–59 years). However, those with type 2 diabetes tended to be older, with 43.7% in those aged 50 to 59 years, 31.9% in those aged 40 to 49 years, 17.3% in those aged 30 to 39 years, and 7.1% in those aged 20 to 29 years. There was no gender difference between those with type 1 versus type 2 diabetes (47.3% were women in both groups).

We plotted prevalence rates for type 1 diabetes (Figure 1). Before 2014, there was a 3% annual decline in the prevalence for those younger than 50 years and a 6% decline in those aged 50 to 59 years (Table A, available as a supplement to the online version of this article at http://www.ajph.org). After 2014, the prevalence increased by 6% to 20%. The pre–post differences in slopes were significant for each age group. For all age groups together, there was a 4% decline before the mandate and an 11% increase afterward (P < .001). When we added age into a 3-way interaction, it was significant (P < .001) indicating that the gap among age groups narrowed after enactment of the mandate (Figure 1).

FIGURE 1—

Changes in the Annual Prevalence Rate for Type 1 Diabetes Mellitus Before and After the Preexisting Condition Mandate: United States, 2005–2016

For persons with type 2 diabetes, the greatest pre–post difference was evident in those aged 50 to 59 years; there was a 6% decline in prevalence of type 2 diabetes before the mandate and a 15% increase afterward (P < .001, difference in slopes; Table A and Figure A, available as supplements to the online version of this article at http://www.ajph.org). There was also a significant change in those who were aged 40 to 49 years (4% decline before, 3% increase after; P = .031). For the younger adults (aged 20–39 years), however, there was no significant increase in prevalence after the mandate. In fact, for those in the youngest age category (20–29 years), there was a greater decline after the mandate. Unlike patients with type 1 diabetes, patients with type 2 diabetes did not experience a narrowing of age disparities after the mandate was enacted (P = .171 for interaction term).

In secondary analyses, we found that the percentage of patients who had a diabetic complication when their first diabetes diagnosis was recorded under private health insurance was 24.4% (95% confidence interval [CI] = 24.3%. 24.4%) before the mandate and 33.4% (95% CI = 30.0%, 36.9%) after the mandate was enacted.

DISCUSSION

We found that the prevalence of diabetes in middle-aged adults in a privately insured population increased after enactment of the ACA’s preexisting condition protections in 2014. The prevalence increased after enactment across all age groups for those with type 1 diabetes and increased in those aged 40 to 59 years with type 2 diabetes.

These findings suggest that the ACA preexisting condition protections may have been effective in increasing coverage among those who may have been previously denied care or found it unaffordable. Because the prevalence of diabetes mellitus was increasing nationwide from 2005 to 2016,⁹ our findings of decreasing prevalence before the mandate are unlikely to be attributable to national trends. In 2005, 5.61% of the US population had diabetes, which increased to 7.18% in 2013.⁹ It may be argued that the increase after the mandate was attributable to additional diabetes screening in newly insured patients (and not in those with preexisting disease). However, we found that a diagnosis of diabetes with complications at the time of first enrollment into health insurance was more prevalent after the mandate (compared with before). This suggests that, after the mandate, a greater proportion of patients with preexisting diabetes enrolled.

In addition to preexisting conditions, age itself has been used as a reason for denial of health coverage or increased premiums.^10,11 The ACA protections limited the previously wide variation in premium rates for older adults to no more than 3 times the rates for younger adults.² Our results suggest that the preexisting condition mandate decreased age disparities in health insurance for patients with type 1 diabetes, as the gap in prevalence rates narrowed after 2014.

Study limitations include the fact that data were only available until 2016 for this study and that diagnoses were based on claims encounters. However, claims and billing data have been found to be superior for detecting diabetes when compared with other electronic databases.⁵

PUBLIC HEALTH IMPLICATIONS

Adults with diabetes may have benefited in obtaining private health insurance after implementation of the preexisting condition mandate of the ACA. Efforts to limit enforcement of these protections are likely to contribute to setbacks in access to care.