Racial Trends in Clinical Preventive Services Use, Chronic Disease Prevalence, and Lack of Insurance Before and After the Affordable Care Act – AJMC.com Managed Markets Network

ABSTRACT

Objectives: To compare the relative change in the use of clinical preventive services, prevalence of chronic disease, and share uninsured among White, Black, and Hispanic adults before and after the introduction of the Affordable Care Act (ACA).

Study Design: Retrospective analysis using the Medical Expenditure Panel Survey of adults aged 18 to 64 years. The regression relies on a fully interacted set of indicator variables of each racial group by 3 time periods: 2005-2009, 2010-2013, and 2014-2018.

Methods: Outcomes included indicators of mammography, colonoscopy, and lipid panel use. Several chronic conditions were examined, including diabetes, hyperlipidemia, hypertension, coronary heart disease, and mental health status. The final outcome variables examined health insurance (uninsured or not) and out-of-pocket spending as a share of family income. Regression models were used controlling for patient characteristics (age, income, education) and for a set of fully interacted indicator variables of race and time period. We tested for relative changes in White adults vs minority adults for each outcome variable. We used the Wald test (test command in Stata) to test for relative changes over time.

Results: We found reductions in baseline (pre-ACA) disparities over time in several of the measures between minority adults and White adults. This included greater growth in the use of mammograms and colonoscopies among minority populations. The results also saw relative reductions in hypertension, coronary heart disease, and fair or poor mental health. Finally, the share uninsured among Hispanic adults decreased at a faster rate than among White adults pre-ACA compared with the ACA period examined.

Conclusions: The ACA is associated with a reduction in baseline differences in the use of some clinical preventive services, chronic disease prevalence, health insurance coverage, and out-of-pocket spending. Continued efforts to promote prevention and further expansions of coverage would appear to pay dividends.

Am J Manag Care. 2022;28(4):In Press

_____

Takeaway Points

The Affordable Care Act (ACA) was associated with reductions in pre-ACA disparities in the use of clinical preventive services, chronic disease prevalence, and the number of uninsured.

Use of mammography screens increased faster over time among Hispanic women than among White women, and colonoscopy exams increased faster for both Hispanic and Black women.Hypertension and coronary heart disease prevalence in Hispanic and Black women decreased at faster rates over time in comparison with White women. The prevalence of Hispanic adults with fair or poor mental health decreased faster over time compared with White adults.The share of Hispanic adults uninsured decreased by more than 4% more after the ACA compared with White adults.

_____

The Affordable Care Act (ACA) (Pub L No. 111-148) was designed to reduce the number of uninsured as well as accelerate the transition toward value-based care. In 2010, 18.2% of the nonelderly population were uninsured compared with 11.1% in 2020.1 Although these trends are well documented through national surveys and Census data, less well known is whether the ACA reduced inequities in the use of care and in the number of uninsured.

Large differences in the use of clinical preventive services and percentage uninsured existed prior to the ACA. In 2010, the percentage uninsured differed widely by race. Among Black adults, 20.8% were uninsured, as were 30.7% of Hispanic adults, compared with 15.4% of White adults. Similar differences were observed across race in the use of clinical preventive services such as colorectal cancer screening. In 2010, nearly 60% of White adults aged 50 to 75 years had a screen, compared with 55.2% of Black adults and only 46.5% of Hispanic adults.2

Several studies have examined the impact of the ACA on changes in racial disparities in health insurance coverage and the use of services.3 The studies were recently summarized by the Kaiser Family Foundation.4 These studies found a larger reduction in the percentage uninsured among racial minorities and a reduction in delaying necessary care. However, some of these studies included data from only 2011 on and did not include trend data prior to the enactment of the ACA. Other studies tracked trends in the share uninsured starting in 2008 through 2017.5

We expand this previous work in 2 ways. First, compared with earlier work, we use a longer pre-ACA baseline time period to identify underlying secular changes apart from changes that may be associated with the ACA. Second, in addition to examining changes in insurance coverage, we also examine relative changes in the use of clinical preventive services, chronic disease prevalence, and out-of-pocket spending. Specifically, we examine trends by race in health insurance coverage, out-of-pocket health care spending, trends in chronic disease treatment patterns, and the use of clinical preventive services before and after the introduction of the ACA. The analysis included data from 2005 through 2018 from the Medical Expenditure Panel Survey–Household Component (MEPS-HC).6

ACA Provisions

The ACA passed in March 2010 contained several provisions that would expand coverage and reduce what individuals pay out of pocket for their care. Clinical preventive services were provided in the ACA benefit package with no cost sharing. A brief summary of these provisions is listed in Table 1.

Some of the more notable changes were the elimination of cost sharing for clinical preventive services and the establishment of insurance exchanges with subsidized premiums and cost sharing. Both provisions increased coverage and the use of clinical preventive services. We explore whether these provisions reduced disparities in coverage and in the use of clinical preventive services starting in 2010.

METHODS

The MEPS-HC is the main data source for the analysis. The MEPS tracks individual and household demographic, socioeconomic, and health-related characteristics, providing a nationally representative sample of the US civilian noninstitutionalized population (the study population of inference). The sampling frame is drawn from respondents to the National Health Interview Survey, which is conducted by the National Center for Health Statistics. The MEPS-HC collects data from a nationally representative sample of households through an overlapping panel design. A new panel of sample households is selected each year, and data for each panel are collected for 2 calendar years. The 2 years of data for each panel are collected in 5 rounds of interviews that take place over a 2.5-year period. This provides continuous and current estimates of health care expenditures at both the individual and household levels for 2 panels for each calendar year. To provide estimates that are representative of a national US population, the MEPS-HC panels have oversampled subgroups such as Hispanic, Black, and Asian individuals; low-income households; and those likely to incur high medical expenditures.

We examine the level and changes, by race, in the percentage uninsured, out-of-pocket spending, chronic disease prevalence, and the use of key preventive services. Specifically, our outcome variables include:

Trends in the uninsured among those aged 18 to 64 yearsOut-of-pocket spending as a percentage of total incomeTrends in prevalence of diabetes, hyperlipidemia, hypertension, heart disease, and mental disordersTrends in key preventive health screenings (mammography, colon cancer, lipid panels). We use the recommendations from the US Preventive Services Task Force for periodicity. They are mammography for women aged 45 years and older, colorectal cancer screening for adults aged 50 to 75 years, and lipid testing every 2 to 3 years for men 35 years and older and women 45 years and older.7

The analysis will compare changes in each of the listed outcomes (clinical preventive service use, chronic disease prevalence, out-of-pocket spending, changes in the number of uninsured) comparing trends in White adults relative to changes among Black adults and Hispanic adults. To do this we estimate a fully interacted model that interacts each of the 3 time periods by White, Hispanic, Black, and other adults. This fully interacted model takes the following form:

Yt = B1 + B2(2010-2013*White) + B3(2014-2018*White) + B4(2005-2009*Black) + B5(2010-2013*Black) + B6(2014-2018*Black) + B7(2005-2009*Hispanic) + B8(2010-2013*Hispanic) + B9(2014-2018*Hispanic) + B10(2005-2009*Other) + B11(2010-2013*Other) + B12(2014-2018*Other) + et

For example, to compare changes in the use of colonoscopies across the groups in the first years of the ACA (when the refundable tax credits started) compared with the pre-ACA period, we test the change in use among White adults compared with the change in use among Black adults. We use the same approach for Hispanic adults. Because 2005-2009*White is the reference category, it has a value of 0. Thus, to compare the relative change in our outcome variables for Black adults and Hispanic adults over time compared with White adults (reported in Table 2), we compare these coefficients:

B3(2014-2018*White) = ([B6(2014-2018*Black)] –
[B4(2005-2009*Black)])

We use the test command in Stata version 16 (StataCorp LLC) to determine whether the relative changes across racial groups are significantly different. The Wald tests in Stata are 2-sided tests of equality. We set up the test statement for comparison of the change in White adults vs the change in Black adults and Hispanic adults (2005-2009 to 2014-2018).

We also present results for each of the outcome variables for Black adults and Hispanic adults in the 2014-2018 time period relative to White adults in 2005-2009. Here we are simply reporting the coefficients B6 and B9 from the model outlined above. These data were available only through 2016.

The regression analysis also controls for various patient characteristics, including age, income, region, education, and insured vs uninsured. The analysis focuses on adults aged 18 to 64 years because many of the major changes in the ACA affected this population.

Our dependent variables are both continuous as well as dichotomous. Dichotomous dependent variables are most appropriately estimated using a logistic or probit regression. However, introducing interactions into these models complicates the interpretation of marginal effects. As a result, for the models with dichotomous dependent variables we use a linear probability model. These models have some statistical issues (eg, heteroscedasticity; predictions may be more or less than 0 and 1). However, they do provide a reasonable estimate of how our race/ethnicity variables of interest are associated with changes in the probability of coverage and use of services over time.

We report 2 sets of results. First, we estimate the relative change (trends) in these measures before and after the introduction of the ACA among White, Black, and Hispanic adults. Using Wald tests, we examine whether the trends in these measures were statistically different from the pre-ACA trends. Second, we compare the probability of the use of clinical preventive services, chronic disease prevalence, out-of-pocket spending as a percentage of total income, and percentage uninsured among Black and Hispanic adults in 2014-2018 compared with White adults in 2005-2009.

RESULTS

Table 3 shows the means and CIs for each of the outcome variable overall and by race from 2005 to 2018.The data from 2005 through 2018 included an average of 63.4% White adults, 12.6% Black adults, and nearly 16% Hispanic adults. The shares of these minority populations nationally increased in total from 26% (both Black and Hispanic) in 2005 to 28.5% by 2018. Table 3 shows that the share of the population without insurance declined from 23.6% in the year before the ACA was passed to 13.2% by 2018. The prevalence of key chronic diseases such as hyperlipidemia, hypertension, and coronary heart disease also decreased over this same time period. However, the prevalence of diagnosed diabetes increased from 6.0% to 7.3% between 2009 and 2018.

Trends in the use of clinical preventive services were mixed. The share of women aged 30 to 64 years with a mammography (mean number per year) showed a slight decrease from 76.0% to 74.4% by 2018. The share of adults aged 50 to 75 years with a colonoscopy remained the same between 2009 and 2018, at 52.0%. Finally, the share of adults with an annual lipid panel increased over time, from 42.9% in 2009 to 50.3% by 2018.

Table 4 provides insight into how much of these changes were associated with the ACA compared with pre-ACA secular trends. We start by estimating the difference in each outcome variable between White adults and Black adults and between White adults and Hispanic adults for each year. We then calculate the mean change in this difference for each variable for 2 time periods, pre-ACA (2005-2009) and the ACA period of 2010-2018 (2016 for the clinical preventive services). There were no discernable trends in these differences in the pre-ACA years. The results are presented below.

We start with the clinical preventive services. Prior to the ACA, the mean annual percentage-point difference in mammography use was 1.4 percentage points higher for Black women and 4.3 percentage points lower for Hispanic women compared with White women. These mean annual differences changed substantially after the ACA as Black women had rates 3.4 percentage points higher than those of White women and Hispanic women had the same mammography rates as White women (P < .05).

We found similar results for colonoscopy screening. Prior to the ACA, screening rates for Black adults and Hispanic adults were a mean of 4.4 and 16.8 percentage points lower, respectively, compared with White adults. With the ACA, screening among Black adults increased sharply, rising to a mean of 2.7 percentage points higher than among White adults (P < .05). Similarly, screening for Hispanic adults increased during the ACA period so that the rates were only 12.1 percentage points lower than for White adults (P < .05).

Finally, we saw smaller changes in lipid panels. Prior to the ACA, lipid panels were ordered a mean of 1.7 percentage points more often for Black adults and 1.4 percentage points more often for Hispanic adults compared with White adults. After the ACA, lipid panels were ordered 2.4 percentage points more often for Black adults (P < .05) and 1.7 percentage points less often for Hispanic adults compared with White adults (the latter not statistically significant; P > .10).

We also find a larger reduction in the share uninsured over time among Hispanic adults relative to White adults. Prior to the ACA, the share of uninsured Hispanic adults was 26.6 percentage points higher. After the ACA, the share of uninsured Hispanic adults was 22.8 percentage points higher (P < .05). The relative reduction in the share uninsured over time among Black and White adults was similar.

We also conducted another test to distinguish underlying pre-ACA secular trends from trends associated with the ACA. We compared changes in each of the clinical preventive measures and share uninsured over time to see if there were statistically significant increases in the measures and reductions in the share uninsured relative to the pre-ACA trends. Our measure for this was:

([White adults 2010-2013] – [White adults 2005-2009]) –
([Black adults 2010-2013] – [Black adults 2005-2009])

The same tabulation was completed for 2014-2016 (2018 for uninsured) compared with the pre-ACA trends in 2005 to 2009. The comparisons were made for both Black and Hispanic adults relative to the baseline pre-ACA trends for White adults. Using the Wald test, we found that Hispanic women had statistically significant increases in mammography tests in the 2014-2016 period relative to the pre-ACA trend among White women. Colonoscopy use also increased at a faster rate for both Black and Hispanic adults during both of the ACA time periods examined relative to trends pre-ACA among White adults (all P < .05).

Relative to the pre-ACA trends, lipid tests also increased (P < .05) among Black adults in the 2010-2013 period at a faster rate than observed among trends in White adults. There were no significant differences in trends in lipid tests for Hispanic adults under the ACA compared with the pre-ACA period.

Finally, there was a larger reduction in uninsured Hispanic adults starting in 2014-2018 compared with trends among White adults. The changes in the number of uninsured among Black and Hispanic adults during the ACA were similar to those measured among White adults.

Table 4 reports trends in overall averages that may mask changes over time in these variables by race or other demographics. To account for this potential, we estimated regression analyses using the formulation outlined earlier in the article. We also expanded our analysis to include out-of-pocket spending, chronic disease prevalence, and health status measures.

The regression results are presented in Table 2. The first column compares the means (levels) for each variable among Black and Hispanic adults in 2014-2018 compared with White adults in 2005-2009. The second column compares the relative change in each outcome among White, Black, and Hispanic adults between 2014-2018 (the ACA period) compared with 2005-2009 (before the ACA).

The use of clinical preventive services in the 2014-2018 period was higher for minority adults than the levels for White adults in 2005-2009 (all P < .05). The share of Hispanic women receiving a mammography was 7.1 percentage points higher and Black women 5.2 percentage higher than that of White women in 2005-2009. Colonoscopy exams were 1.8 percentage points higher for Hispanic adults in 2014-2018 than for White adults in 2005-2009. The most dramatic differences were for lipid panels: The share of Hispanic adults receiving them was 19.3 percentage points higher and of Black adults nearly 23 percentage points higher than the share of White adults in 2005-2009.

The health measures showed mixed results. Diagnosed diabetes prevalence was 2.3 percentage points higher among Hispanic adults in 2014-2018 compared with White adults in 2005-2009 (P < .05). However, the shares of minority adults with fair or poor reported mental health outcomes were 2.0 percentage points lower for Hispanic adults and 1.2 percentage points lower for Black adults in the later period compared with White adults in 2005-2009 (the latter P < .05).

Out-of-pocket spending as a share of household income was also lower for Hispanic families (1.5 percentage points) and Black families (1.7 percentage points) in 2014-2018 compared with White families in 2005-2009 (P < .05).

We now turn to changes in the relative growth of these outcome variables, comparing 2014-2018, the years the exchanges started, with the years prior to the enactment of the ACA. Growth in the use of 2 clinical preventive services, mammography and colonoscopy, increased faster over time for Hispanic adults compared with White adults (P < .05). The rate of colonoscopy also increased faster among Black adults relative to White adults before the ACA. There were no differences over time in the use of lipid panels for the 3 groups.

There were also significant differences in the growth in prevalence of key chronic conditions. There were no differences in the change in diabetes prevalence over time among White, Hispanic, and Black adults. The growth in hyperlipidemia increased 0.2 percentage points higher for Hispanic adults compared with White adults (P < .05). There were no differences in growth rates in hyperlipidemia for Black and White adults over time. Changes in the prevalence of hypertension decreased at a faster rate over time for Black (P < .10) and Hispanic (P < .05) adults compared with White adults. For Hispanic adults, the change in prevalence relative to White adults was 1.4 percentage points lower and for Black adults, it was 0.8 percentage points lower.

The prevalence of coronary heart disease also grew at a slower rate over time among Black and Hispanic adults compared with White adults. The prevalence of coronary heart disease among minority adults decreased by 0.1 percentage points faster than among White adults between 2014-2018 compared with 2005-2009 (P < .05). There were also reductions in racial disparities over time among adults with fair to poor mental health. Relative to the growth among White adults, the share of Hispanic adults with fair to poor mental health decreased by 2.4 percentage points. Similarly, the change in prevalence among Black adults with fair to poor mental health was 0.8 percentage points lower over time compared with White adults (both P < .05).

Finally, 2 of the major targets of the ACA were to reduce the number of uninsured and the amount that families pay out of pocket for health care. Overall, the share of adults without health insurance decreased from 23.6% in the year before the ACA (2009) to 13.2% by 2018. The percentage-point reduction in the number of uninsured was faster for Hispanic adults (P < .05), but among Black adults it decreased at the same rate as among White adults. Among Hispanic adults, the change in the uninsured rate decreased by 4.8 percentage points more over this period compared with White adults.

Finally, out-of-pocket spending as a share of family income also decreased at a faster rate among Black adults compared with White adults. There were no significant changes in out-of-pocket spending among Hispanic adults compared with White adults over time.

DISCUSSION

The ACA reduced the share of adults without health insurance by nearly half, from 23.6% to 13.2% between 2009 and 2018. The reduction in the percentage uninsured was faster among Hispanic adults compared with White adults over this period. Prior to the introduction of the ACA, the share of Hispanic adults who were uninsured was 5.2 percentage points higher than that of White adults. The 4.8-percentage-point larger reduction in the uninsured among Hispanic adults compared with White adults has dramatically reduced the pre-ACA inequities in uninsured rates. The ACA was also associated with a faster reduction in out-of-pocket spending as a share of family income for Black families relative to White families.

The ACA’s focus on eliminating cost sharing associated with clinical preventive services was also associated with a reduction in baseline racial differences in the use of these key services. The growth in the use of mammography (Hispanic women) and colonoscopy screening also increased at a higher percentage point rate among both Hispanic and Black adults compared with White adults with the implementation of the ACA.

Although approximately 13% of adults overall are still without health insurance, the share of adults without health insurance decreased substantially among Hispanic adults. Although the rates of adults without insurance remain high, the ACA has had important impacts on reducing racial disparities in health insurance coverage, the use of clinical preventive services, and out-of-pocket payments. The changes in key chronic conditions over time were more mixed. The percentage-point increases in hypertension and hyperlipidemia were higher among some minority adults compared with White adults.

Limitations

One limitation is that our data on diabetes are of diagnoses and do not include undiagnosed diabetes, so it is not clear whether these changes reflect overall increases. A second limitation relates to causation. Although the results show an association of the ACA with the outcome measures we examined, we cannot prove causation. However, there were no other exogenous events as large as the ACA during this time period.

CONCLUSIONS

Given the large differences in the share of uninsured and the use of clinical preventive services among Black and Hispanic adults relative to White adults pre-ACA, the ACA does appear to have reduced the differences between minority adults and White adults. This suggests that additional efforts to reduce the cost of health insurance even further could result in continued reductions in racial differences in the number of uninsured. Continued expansion of the ACA could prove an important vehicle going forward to accomplish this goal.

Author Affiliation: Department of Health Policy and Management, Rollins School of Public Health, Emory University, Atlanta, GA.

Source of Funding: Blue Cross and Blue Shield Association.

Author Disclosures: Dr Thorpe has received grants from Blue Cross and Blue Shield Association.

Authorship Information: Concept and design; analysis and interpretation of data; drafting of the manuscript; and statistical analysis.

Address Correspondence to: Kenneth E. Thorpe, PhD, Emory University, 1518 Clifton Rd NE, Ste 620, Atlanta, GA 30322. Email: kthorpe@emory.edu.

REFERENCES

1. Trends in the U.S. uninsured population, 2010-2020. Assistant Secretary for Planning and Evaluation Office of Health Policy. February 11, 2021. Accessed February 28, 2021. https://aspe.hhs.gov/system/files/pdf/265041/trends-in-the-us-uninsured.pdf

2. Health, United States, 2016: With Chartbook on Long-term Trends in Health. National Center for Health Statistics; 2017. Accessed February 20, 2021. https://www.ncbi.nlm.nih.gov/books/NBK453380/

3. Chen J, Vargas-Bustamante A, Mortensen K, Ortega AN. Racial and ethnic disparities in health care access and utilization under the Affordable Care Act. Med Care. 2016;54(2):140-146. doi:10.1097/MLR.0000000000000467

4. Guth M, Garfield R, Rudoitz R. The effects of Medicaid expansion under the ACA: studies from January 2014 to January 2020. Kaiser Family Foundation. March 17, 2020. Accessed February 21, 2021. https://www.kff.org/medicaid/report/the-effects-of-medicaid-expansion-under-the-aca-updated-findings-from-a-literature-review/

5. Buchmueller T, Levy H. The ACA’s impact on racial and ethnic disparities in health insurance coverage and access to care. Health Aff (Millwood). 2020;39(3):395-402. doi:10.1377/hlthaff.2019.01394

6. Medical Expenditure Panel Survey: restricted data files available at the data centers. Agency for Healthcare Research and Quality. Updated April 22, 2019. Accessed February 21, 2021. https://meps.ahrq.gov/mepsweb/data_stats/onsite_datacenter.jsp

7. Breast cancer: screening. US Preventive Services Task Force. January 11, 2016. Accessed February 21, 2021. https://www.uspreventiveservicestaskforce.org/uspstf/recommendation/breast-cancer-screening