Associations Between Medicaid Expansion and Mental Health Among U.S. Racial and Ethnic Groups
Abstract
Objective:
The authors examined associations between Medicaid expansion and self-reported mental health by race-ethnicity, focusing on lagged associations.
Methods:
This retrospective, cross-sectional study used 2011–2019 data from the Behavioral Risk Factor Surveillance System. The sample included low-income, childless adults ages 25–64 years. Difference-in-differences (DID) analysis was used to estimate associations between Medicaid expansion and self-reported mental health. Lagged associations were examined by separating the postexpansion period into proximal (2014–2016) and distal (2017–2019) periods.
Results:
In the overall sample (N=327,248), Medicaid expansion was associated with a reduction in the mean number of self-reported past-month poor mental health days (DID=−0.12, 95% CI=−0.21 to −0.03), after adjustment for covariates. The expansion was associated with significant reductions in past-month poor mental health days for the following groups: non-Hispanic White (DID=−0.18, 95% CI=−0.29 to −0.07), non-Hispanic Asian (DID=−1.15, 95% CI=−1.37 to −0.93), non-Hispanic other (DID=−0.62, 95% CI=−1.03 to −0.21), and Hispanic (DID=−0.48, 95% CI=−0.73 to −0.23). The non-Hispanic Black group had a significant increase in past-month poor mental health days (DID=0.27, 95% CI=0.06 to 0.49), and no significant change was noted for the American Indian or Alaska Native (AIAN) group. Improvements in mental health observed at the beginning of the policy implementation (proximal period) were not sustained over time for some racial-ethnic minority groups.
Conclusions:
Although Medicaid expansion improved mental health for the overall sample, some racial-ethnic disparities were detected. The negative and insignificant associations for the non-Hispanic Black and AIAN groups, respectively, highlight the need to better understand why the Medicaid expansion affected racial-ethnic groups differently.
HIGHLIGHTS
•
Previous studies of the impact of Medicaid expansion under the Affordable Care Act have examined associations of the expansion with outcomes only in the short term and reported limited information about how associations differed by racial or ethnic groups.
•
In a sample of 327,248 low-income, childless adults ages 25–64 years, Medicaid expansion was associated with a reduction in the mean number of poor mental health days, but associations varied by racial-ethnic group.
•
Improvements in mental health observed in the first 3 years after Medicaid expansion were not sustained over time for some racial-ethnic groups, and further study of potential reasons is needed.
Medicaid remains the largest health insurance plan provided by the U.S. government, covering nearly 20% of the U.S. population (approximately 64 million individuals) (1). Medicaid expansion was initially proposed as a mandate through the Affordable Care Act (ACA) to provide health insurance coverage for individuals with income up to 138% of the federal poverty level (FPL) (2). After the Medicaid expansion mandate was revoked by the U.S. Supreme Court in 2012, states were allowed to freely opt out of the mandate. Policy advocates and critics have ongoing debates about the financial burden on state governments from Medicaid expansion (3). Although many states have chosen to implement Medicaid expansion over time since January 1, 2014, the earliest date on which a state could implement the ACA’s Medicaid expansion policy, 12 states have not implemented Medicaid expansion as of 2022 (4).
Research suggests that Medicaid expansion improved health care access and quality while not causing drastic budget increases for state governments (5, 6). Nevertheless, research findings of the policy’s effects on health outcomes were inconsistent. Although many studies reported that the policy may be associated with improvements in self-reported health outcomes (7–11), others documented no significant associations (12–18). On the basis of existing evidence, Medicaid expansion has been found to be associated with improvements in a population’s mental health, especially among low-income, childless adults (11, 19). These findings could be explained by the fact that the expansion helped newly eligible beneficiaries in expansion states to overcome financial barriers to treatment, which have long been perceived as a significant obstacle to accessing mental health care for the low-income population (20). Additionally, a study that used a randomized controlled trial design reported that under health care reform, improvements in mental health may be a more likely outcome than improvements in general medical health (21), which further supported the direction of our investigation into mental health.
We examined longer-term associations (i.e., lagged associations) between Medicaid expansion and mental health outcomes, focusing on racial-ethnic differences and current evidence gaps. Many previous studies have been limited because they analyzed only a few years of postexpansion data, which may explain findings of a lack of evidence of significant associations between the expansion policy and investigated outcomes (19). Some barriers, such as structural (e.g., accessibility of providers) and attitudinal (e.g., the perceived need for care) barriers, which also affect mental health, might become more apparent over time after beneficiaries gain access to care (22). Investigation of lagged associations of outcomes with the expansion policy could improve understanding about whether such barriers influence mental health outcomes in the long run, which is especially important for persons from racial-ethnic minority groups, because they are more susceptible to the aforementioned barriers (23, 24). To the best of our knowledge, only one study—Lee and Porell (9)—has examined mental health of different racial and ethnic groups during the first 2 years after Medicaid expansion and found no differential impact among the groups. Because additional data for a longer follow-up period are now available, we investigated lagged associations with the expansion policy, stratified by various racial and ethnic groups.
Our study objectives were to evaluate the association between Medicaid expansion and self-reported mental health, investigate whether associations between the policy and mental health differ by self-reported race-ethnicity, and examine lagged associations of the policy with mental health.
Methods
Study Design
We conducted a retrospective, cross-sectional study. Because each state's adoption of the Medicaid expansion policy was self-selected, this self-selection mechanism provided a quasi-experimental environment for our data analysis (18). This study was approved by the Auburn University Institutional Review Board. We followed the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for reporting the study (25).
Data Source
We used data from the Behavioral Risk Factor Surveillance System (BRFSS) from 2011 to 2019. The BRFSS is a nationally representative cross-sectional telephone survey administered and supported by the Population Health Surveillance Branch of the Centers for Disease Control and Prevention (CDC). The BRFSS annually collects data at the state level via landline or cell phone interviews with U.S. residents ages ≥18 years regarding their health-related risk behaviors and chronic health conditions (26). We used data starting from 2011, as recommended by BRFSS because of changes in its survey methodology as of 2011 (27). We used BRFSS survey years through 2019 and excluded 2020 data to avoid the potential confounding effects of the COVID-19 pandemic on mental health (28).
Study Population
We included data from low-income, childless U.S. adults ages 25–64 years. Low income was defined as an income level ≤138% FPL (29, 30). Because income data from the BRFSS are reported categorically, we calculated respondents’ income in the form of a percentage of FPL by dividing the midpoint of each respondent's income category by the FPL value of the corresponding state and interview year (18). We used multivariable regression based on respondents’ sex, age, race-ethnicity, education level, marital status, employment, and state of residence to impute missing income data. Respondents with missing self-reported number of past-month poor mental health days (i.e., the primary outcome) were excluded.
Outcome Measures
The outcome was self-reported number of past-month poor mental health days, which was obtained by the following question: “Including stress, depression, and problems with emotions, for how many days during the past 30 days was your mental health not good?” Respondents could report a continuous value (0–30 days), or they could explicitly refuse to answer or simply not respond (missing). Respondents with missing data were excluded from the data analysis. This measure of poor mental health days is one of the four quality-of-life measures that have been applied to several population-based surveys (31). Frequent mental distress (defined as ≥14 poor mental health days) has been associated with many chronic diseases and mental disorders (32, 33). Previous research has also shown that ≥6 poor mental health days could be a reliable indicator of generalized mental distress (34).
Covariates
Covariates included individual-level variables (sex, age, race-ethnicity, education level, marital status, state of residence, and interview year [35]) and a state-level variable (based on the annual state unemployment rate from the Bureau of Labor Statistics [36]). All definitions and categorizations of variables followed the CDC definition for collecting the BRFSS data. For individual-level covariates, any missing data or no response was categorized into a single category, and the respondent’s data were included in our statistical analysis.
Statistical Analysis
We examined the characteristics of the study population with descriptive analysis. The baseline demographic comparisons between expansion and nonexpansion states were conducted with t tests and chi-square tests. Percentages were weighted to the U.S. population by applying the appropriate weight, strata, and cluster variables as recommended by BRFSS (27). Our statistical analyses considered the complex sampling design of BRFSS (27) and controlled for relevant covariates (i.e., individual-level and state-level variables mentioned above). An a priori level of statistical significance was set at 0.05, and hypothesis tests were two-sided. Analyses were conducted with SAS, version 9.4.
The mean number of past-month poor mental health days was analyzed by pre- and postexpansion periods and state expansion status (i.e., expansion or nonexpansion). The 2011–2013 period was defined as the preexpansion period, and 2014–2019 was defined as the postexpansion period. (The classification of state expansion status is further described in the online supplement to this article.) Difference-in-differences (DID) analysis was used to estimate the association between Medicaid expansion and mental health status. The DID model allowed us to examine the outcome changes after policy implementation while comparing persons affected by the policy with those not affected (37) (see the online supplement ). Difference-in-difference-in-differences (TD) analysis was used to compare how the associations with mental health status differed between various racial-ethnic groups (38) (see the online supplement ). The categorization of racial-ethnic groups was based on respondents’ self-reported race and ethnicity.
Finally, we examined lagged associations with expansion policy. The examination was motivated by a review by Soni et al. (19) that highlights the inability of short-term follow-up studies to measure lagged associations with policies. We examined lagged associations between Medicaid expansion and mental health by separating the follow-up period (i.e., 2014–2019) into proximal and distal postexpansion periods. We included 2014–2016 in the proximal postexpansion period to capture proximal associations with the policy, and we included 2017–2019 in the distal postexpansion period to identify distal policy associations. During the proximal postexpansion period, states with Medicaid expansion implemented for at least 1 year between 2014 and 2016 were categorized as expansion states. During the distal postexpansion period, states with Medicaid expansion implemented for at least 3 years between 2014 and 2018 were categorized as expansion states. When examining the lagged policy associations, we excluded the District of Columbia and four states—Delaware, Massachusetts, New York, and Vermont—that had implemented in-state health care plans before 2014 that were equivalent to the coverage under Medicaid expansion for low-income, childless adults.
Sensitivity Analysis
In the first sensitivity analysis, we ran the adjusted DID analysis and removed the four states and the District of Columbia. The second sensitivity analysis was the adjusted DID analysis excluding two states (i.e., Maine and Virginia) that implemented the Medicaid expansion policy in 2019. Maine and Virginia were categorized initially as nonexpansion states in the primary analysis. (For further descriptions of the special cases for excluding these states, see the online supplement .)
Results
Characteristics of the Study Population
A total of 327,248 respondents were retained in this study population of low-income, childless adults ages 25–64. Of this population, 112,912 respondents were retained in the preexpansion period (2011–2013) and 214,336 in the postexpansion period (2014–2019). Most of the respondents from pre- and postexpansion periods were ages ≥45, and fewer than half were female (Table 1). During the preexpansion period, non-Hispanic White respondents represented more than half the sample (N=74,433, 54% [weighted percentage]); during the postexpansion period, this group represented less than half the sample (N=134,288, 48% [weighted percentage]).
| 2011–2013 | 2014–2019b | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Nonexpansion states (N=45,781) | Expansion states (N=67,131) | Nonexpansion states (N=88,790) | Expansion states (N=125,546) | |||||||
| Characteristic | N | % | N | % | p | N | % | N | % | p |
| Female | 27,628 | 49.1 | 39,204 | 47.4 | 49,313 | 48.1 | 66,245 | 46.1 | ||
| Age in years | <.001 | <.001 | ||||||||
| 25–34 | 3,576 | 16.0 | 5,976 | 18.7 | 10,969 | 19.9 | 17,225 | 23.3 | ||
| 35–44 | 4,158 | 14.5 | 6,432 | 14.4 | 9,380 | 14.9 | 13,267 | 14.3 | ||
| 45–54 | 14,247 | 32.5 | 20,630 | 30.7 | 23,836 | 27.3 | 33,065 | 26.0 | ||
| 55–64 | 23,800 | 37.0 | 34,093 | 36.2 | 44,605 | 38.0 | 61,989 | 36.4 | ||
| Race-ethnicity | <.001 | <.001 | ||||||||
| Non-Hispanic White | 30,067 | 54.0 | 44,366 | 54.3 | 55,782 | 49.0 | 78,506 | 48.5 | ||
| Non-Hispanic Black | 9,475 | 24.4 | 8,875 | 16.0 | 16,941 | 24.2 | 14,466 | 15.1 | ||
| Non-Hispanic Asian | 239 | 1.0 | 1,222 | 4.3 | 714 | 1.7 | 3,412 | 5.4 | ||
| Non-Hispanic American Indian or Alaska Native | 1,498 | 2.1 | 2,067 | 2.2 | 3,768 | 1.8 | 5,205 | 1.9 | ||
| Non-Hispanic other | 1,445 | 2.5 | 3,152 | 2.8 | 2,803 | 2.2 | 6,403 | 2.7 | ||
| Hispanic | 3,057 | 16.1 | 7,449 | 20.4 | 8,782 | 20.6 | 17,554 | 25.7 | ||
| Education | <.001 | <.001 | ||||||||
| Less than high school | 2,721 | 9.9 | 3,831 | 9.4 | 5,337 | 10.1 | 7,375 | 11.2 | ||
| Some high school or graduate | 25,600 | 57.9 | 34,774 | 54.2 | 47,634 | 55.9 | 63,750 | 51.7 | ||
| Some college or graduate | 17,333 | 31.9 | 28,120 | 35.7 | 35,397 | 33.3 | 53,792 | 36.3 | ||
| Unknown | 127 | .3 | 406 | .6 | 422 | .7 | 629 | .8 | ||
| Marital status | <.001 | <.001 | ||||||||
| Married | 13,199 | 31.0 | 17,312 | 27.8 | 22,422 | 27.5 | 28,383 | 24.6 | ||
| Divorced | 13,540 | 22.6 | 19,106 | 20.5 | 25,054 | 22.4 | 33,986 | 20.2 | ||
| Widowed | 4,436 | 6.6 | 5,564 | 5.3 | 7,591 | 6.2 | 6,998 | 5.3 | ||
| Separated | 2,830 | 6.8 | 3,439 | 5.5 | 5,698 | 7.2 | 7,202 | 6.1 | ||
| Never married | 10,477 | 29.0 | 19,115 | 35.5 | 24,314 | 31.3 | 42,727 | 37.5 | ||
| One of an unmarried couple | 1,031 | 3.5 | 2,036 | 4.8 | 3,029 | 4.5 | 5,135 | 5.4 | ||
| Unknown | 268 | .6 | 559 | .6 | 682 | 1.0 | 1,115 | 1.0 | ||
a
Low income was defined as an income level ≤138% of the federal poverty level. The sample excluded respondents with no data on the primary outcome (i.e., poor mental health days): 2.8% of the total sample during 2011–2013 and 2.9% of the total sample during 2014–2019. All percentages were weighted to be representative of the population.
b
January 1, 2014, was the earliest date that a state could implement the Medicaid expansion policy under the Affordable Care Act. Because expansion states implemented the policy at various times, respondents were categorized as living in an expansion state if their state of residence had implemented Medicaid expansion on the first day of the survey interview month.
Poor Mental Health Days
Compared with the preexpansion period, the unadjusted mean number of past-month poor mental health days generally decreased during the postexpansion period, except for the non-Hispanic White group and the non-Hispanic Asian group living in nonexpansion states (Table 2). Medicaid expansion was associated with a significant reduction in the mean number of past-month poor mental health days for the overall sample (adjusted DID=−0.12, p=0.01).
| Sample and residencea | M days (unadjusted) | Unadjusted analysis | Adjusted analysis | |||||
|---|---|---|---|---|---|---|---|---|
| 2011–2013 | 2014–2019 | DIDb | 95% CI | p | DIDb | 95% CI | p | |
| Full sample | −.25 | −.35 to −.16 | <.001 | −.12 | −.21 to −.03 | .010 | ||
| Expansion state | 7.8 | 7.3 | ||||||
| Nonexpansion state | 8.0 | 7.7 | ||||||
| Non-Hispanic White | −.01 | −.13 to .10 | .84 | −.18 | −.29 to −.07 | <.001 | ||
| Expansion state | 8.6 | 8.9 | ||||||
| Nonexpansion state | 9.1 | 9.4 | ||||||
| Non-Hispanic Black | .04 | −.18 to .26 | .71 | .27 | .06 to .49 | .010 | ||
| Expansion state | 7.1 | 6.8 | ||||||
| Nonexpansion state | 6.8 | 6.5 | ||||||
| Non-Hispanic Asian | −1.50c | −1.99 to −1.00 | <.001 | −1.15c | −1.37 to −.93 | <.001 | ||
| Expansion state | 4.2 | 4.0 | ||||||
| Nonexpansion state | 2.7 | 3.4 | ||||||
| Non-Hispanic American Indian or Alaska Native | .11c | −.96 to .74 | .79 | .14c | −.52 to .79 | .680 | ||
| Expansion state | 9.6 | 8.7 | ||||||
| Nonexpansion state | 11.0 | 10.2 | ||||||
| Non-Hispanic other | −.37 | −.96 to .22 | .22 | −.62 | −1.03 to −.21 | .003 | ||
| Expansion state | 10.7 | 9.3 | ||||||
| Nonexpansion state | 10.8 | 9.9 | ||||||
| Hispanic | −.75 | −1.04 to −.47 | <.001 | −.48 | −.73 to −.23 | <.001 | ||
| Expansion state | 6.3 | 5.0 | ||||||
| Nonexpansion state | 5.7 | 5.0 | ||||||
a
January 1, 2014, was the earliest date that a state could implement the Medicaid expansion policy under the Affordable Care Act. Because expansion states implemented the policy at various times, respondents were categorized as living in an expansion state if their state of residence had implemented Medicaid expansion on the first day of the survey interview month.
b
DID, difference-in-differences analysis (reference group: nonexpansion states). A negative value indicates a decrease in past-month poor mental health days in states that expanded Medicaid, compared with states that did not expand Medicaid, whereas a positive value indicates an increase. The adjusted analyses accounted for covariates, including individual characteristics, such as sex, age, education, marital status, income, state of residence, and interview year, as well as the annual state-level unemployment rate.
c
The coefficient may be biased because of violation of the parallel-trend assumption. The assumption takes the trends in outcome (i.e., poor mental health days) between the expansion and nonexpansion states to be the same before the intervention (i.e., the Medicaid expansion policy) (for details about testing the parallel-trend assumption, see the online supplement ).
Poor Mental Health Days by Race-Ethnicity
The association between Medicaid expansion and poor mental health days differed by race-ethnicity. For the non-Hispanic White, non-Hispanic Asian, non-Hispanic other, and Hispanic groups, the expansion was associated with a significant reduction in the mean number of poor mental health days in the past month (adjusted DID for non-Hispanic White, −0.18; non-Hispanic Asian, −1.15; non-Hispanic other, −0.62; and Hispanic, −0.48; p<0.05 for all) (Table 2). No significant association was found for the American Indian or Alaska Native (AIAN) group. The expansion policy was associated with a significant increase in the mean number of past-month poor mental health days for non-Hispanic Black respondents (adjusted DID=0.27, p=0.01). In the sensitivity analyses, excluding the District of Columbia and the four states that had in-state health care plans similar in coverage to that of Medicaid expansion resulted in a nonsignificant association of the expansion with mental health status for the overall sample and for non-Hispanic Black, non-Hispanic AIAN, other non-Hispanic, and Hispanic groups; excluding the two states that implemented Medicaid expansion in 2019 resulted in no apparent difference from results when the two states were included (Table 3).
| Sample | Adjusted analysis | Sensitivity analysis 1b | Sensitivity analysis 2b | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Adjusted DID | 95% CI | p | Adjusted DID | 95% CI | p | Adjusted DID | 95% CI | p | |
| Full sample | −.12 | −.21 to −.03 | .010 | −.08 | −.18 to .02 | .110 | −.13 | −.22 to −.04 | .006 |
| Non-Hispanic White | −.18 | −.29 to −.07 | <.001 | −.14 | −.25 to −.02 | .030 | −.21 | −.33 to −.09 | <.001 |
| Non-Hispanic Black | .27 | .06 to .49 | .010 | .10 | −.15 to .34 | .430 | .30 | .08 to .52 | .010 |
| Non-Hispanic Asianc | −1.15 | −1.37 to −.93 | <.001 | −1.21 | −1.52 to −.89 | <.001 | −1.07 | −1.03 to −.84 | <.001 |
| Non-Hispanic American Indian or Alaska Nativec | .14 | −.52 to .79 | .680 | .40 | −.29 to 1.07 | .260 | .23 | −.44 to .89 | .500 |
| Non-Hispanic other | −.62 | −1.03 to −.21 | .003 | −.15 | −.62 to .31 | .510 | −.68 | −1.11 to −.25 | .002 |
| Hispanic | −.48 | −.73 to −.23 | <.001 | −.28 | −.57 to .02 | .070 | −.45 | −.70 to −.19 | <.001 |
a
DID, difference-in-differences analysis (reference group: nonexpansion states). A negative value indicates a decrease in past-month poor mental health days in states that expanded Medicaid, compared with states that did not expand Medicaid, whereas a positive value indicates an increase. The adjusted analyses accounted for covariates, including individual characteristics, such as sex, age, education, marital status, income, state of residence, and interview year, as well as the annual state-level unemployment rate.
b
In sensitivity analysis 1, the District of Columbia and four states—Delaware, Massachusetts, New York, and Vermont—were excluded because they had implemented in-state health care plans before 2014 that were equivalent to coverage under Medicaid expansion for low-income, childless adults. Sensitivity analysis 2 excluded two states—Maine and Virginia—that implemented the Medicaid expansion policy in 2019.
c
The coefficient for non-Hispanic Asian and for non-Hispanic American Indian or Alaska Native may be biased because of the violation of the parallel-trend assumption. The assumption takes the trends in outcome (i.e., poor mental health days) between the expansion and nonexpansion states to be the same before the intervention (i.e., the Medicaid expansion) (for details about testing the parallel-trend assumption, see the online supplement ).
The statistical significance of the differential associations stratified by race-ethnicity were examined with the TD model. Compared with the non-Hispanic White group, the non-Hispanic Black and non-Hispanic Asian groups had a relative increase in the mean number of past-month poor mental health days (adjusted TD coefficient=0.33 for both, p<0.05) and the non-Hispanic AIAN, non-Hispanic other, and Hispanic groups had a relative decrease (adjusted TD coefficient=−0.96, −0.60, and −0.82, respectively, p<0.001) (Table 4).
| Comparisona | Coefficientb | 95% CI | p |
|---|---|---|---|
| Non-Hispanic Black | .33 | .26 to .40 | <.001 |
| Non-Hispanic Asian | .33c | .02 to .64 | .036 |
| Non-Hispanic American Indian or Alaska Native | −.96c | −1.24 to −.67 | <.001 |
| Non-Hispanic other | −.60 | −.70 to −.50 | <.001 |
| Hispanic | −.82 | −1.04 to −.60 | <.001 |
a
Reference was non-Hispanic White for all comparisons (for details about this difference-in-difference-in-differences analysis, see the online supplement ).
b
Estimated with difference-in-difference-in-differences analysis. A negative coefficient indicates a relative decrease in past-month poor mental health days associated with the expansion policy, compared with the non-Hispanic White group, whereas a positive coefficient indicates a relative increase. Estimates were adjusted for covariates, including individual characteristics, such as sex, age, education, marital status, income, state of residence, and interview year, as well as the annual state-level unemployment rate.
c
The coefficient may be biased because of the violation of the parallel-trend assumption. The assumption takes the trends in outcome (i.e., poor mental health days) between the expansion and nonexpansion states to be the same before the intervention (i.e., the Medicaid expansion policy) (for details about testing the parallel-trend assumption, see the online supplement ).
Lagged Associations With Medicaid Expansion
During the first 3 years after the Medicaid expansion (2014–2016, the proximal postexpansion period), the number of past-month poor mental health days was reduced for the sample overall, but the reduction was not statistically significant for the non-Hispanic White, non-Hispanic AIAN, and Hispanic groups (Table 5). For >3 years after the Medicaid expansion (2017–2019, the distal postexpansion period), a significant reduction in the number of poor mental health days was observed for the non-Hispanic White, non-Hispanic Asian, and Hispanic groups (adjusted DID=−0.22, −0.44, and −0.35, respectively, p<0.05 for all), whereas an increase was observed for the non-Hispanic AIAN and non-Hispanic other groups (adjusted DID=0.96 and 0.84, respectively; p<0.05 for both). The distal association was insignificant for the overall sample and the non-Hispanic Black group.
| Sample | Postexpansion period (2014–2019) | Proximal postexpansion period(2014–2016) | Distal postexpansion period (2017–2019) | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Adjusted DIDa | 95% CI | p | Adjusted DIDa | 95% CI | p | Adjusted DIDa | 95% CI | p | |
| Full sample | −.08 | −.18 to .02 | .110 | −.16 | −.26 to −.06 | .002 | −.06 | −.18 to .06 | .350 |
| Non-Hispanic White | −.14 | −.25 to −.02 | .030 | −.01 | −.13 to .11 | .860 | −.22 | −.38 to −.07 | <.005 |
| Non-Hispanic Black | .10 | −.15 to .34 | .430 | −.54 | −.81 to −.28 | <.001 | .21 | −.11 to .53 | .200 |
| Non-Hispanic Asianb | −1.21 | −1.52 to −.89 | <.001 | −2.31 | −2.63 to −1.99 | <.001 | −.44 | −.81 to −.07 | .020 |
| Non-Hispanic American Indian or Alaska Nativeb | .40 | −.29 to 1.07 | .260 | −.62 | −1.34 to .11 | .090 | .96 | .13 to 1.79 | .020 |
| Non-Hispanic other | −.15 | −.62 to .31 | .510 | −.67 | −1.14 to −.20 | <.010 | .84 | .28 to 1.39 | .003 |
| Hispanic | −.28 | −.57 to .02 | .070 | −.25 | −.54 to .05 | .100 | −.35 | −.68 to −.02 | .040 |
a
DID, difference-in-differences analysis (reference: nonexpansion states). The adjusted analyses accounted for covariates, including individual characteristics, such as sex, age, education, marital status, income, state of residence, and interview year, as well as the annual state-level unemployment rate. January 1, 2014, was the earliest date that a state could implement the Medicaid expansion policy under the Affordable Care Act. A negative coefficient indicates a relative decrease in past-month poor mental health days associated with the expansion policy, compared with nonexpansion states, whereas a positive coefficient indicates a relative increase.
b
The coefficient for non-Hispanic Asian and for non-Hispanic American Indian or Alaska Native may be biased because of the violation of the parallel-trend assumption. The assumption takes the trends in outcome (i.e., poor mental health days) between the expansion and nonexpansion states to be the same prior to the intervention (i.e., the Medicaid expansion policy) (for details about testing the parallel-trend assumption, see the online supplement ).
Discussion
Our findings indicate that Medicaid expansion was associated with significantly improved self-reported mental health among low-income, childless adults ages 25–64. However, the association of Medicaid expansion with mental health differed by race-ethnicity. The expansion was associated with significantly improved mental health for the non-Hispanic White, non-Hispanic Asian, non-Hispanic other, and Hispanic groups living in expansion states. In contrast, Medicaid expansion was significantly associated with worsened mental health for the non-Hispanic Black group, and the association was nonsignificant for the non-Hispanic AIAN group. Examining the lagged associations of Medicaid expansion with mental health status revealed improvements in mental health at the beginning of the expansion (i.e., in the proximal postexpansion period) for most racial-ethnic groups. In contrast, in the distal postexpansion period, the expansion was not associated with changes in mental health status, and some racial-ethnic groups reported worse mental health status.
Our finding of the association of Medicaid expansion with improved mental health is consistent with results of several previous studies (7, 9–11, 39). In terms of mental health outcomes among various racial-ethnic groups, most of our results were consistent with the previous work by Lee and Porell (9). A key difference in our findings was the insignificant proximal association observed for non-Hispanic White respondents, whereas Lee and Porell reported a significant association of Medicaid expansion and improved mental health among non-Hispanic Whites during their study period (2011–2016).
Race-ethnicity should be considered a social construct rather than a biological trait. We acknowledge that the inclusion of several confounders in our adjusted models might under- or overestimate the associations between Medicaid expansion and mental health status for various racial-ethnic groups, because race-ethnicity could be a proxy for many omitted variables. In considering this possibility, we analyzed both unadjusted and adjusted models (Table 2), with the unadjusted model showing that the expansion was significantly associated with mental health status only in the overall sample and in the non-Hispanic Asian and Hispanic groups. However, the adjusted analyses were intended to adjust for confounders of the associations between Medicaid expansion and mental health status, because the confounders included have been widely recognized in the literature. Therefore, we focus on the adjusted model results in the following discussion.
Although Medicaid expansion might be expected to increase access to care by reducing financial barriers, the Medicaid expansion policy was not necessarily associated with improved mental health outcomes, especially for the non-Hispanic Black group. Our result is supported by findings from a study by Breslau et al. (40), which reported that non-Hispanic Black individuals residing in Medicaid expansion states had fewer outpatient mental health visits, compared with non-Hispanic Black individuals residing in nonexpansion states; however, this finding was not statistically significant. For the non-Hispanic Black population, historical events that have caused mistrust in health care, fear of discrimination, and social stigma can contribute to significant barriers that influence their willingness to access mental health care (41). Earlier work has shown that non-Hispanic Black individuals who needed mental health care were 10% less likely than their non-Hispanic White counterparts to receive care, even if they had insurance coverage (42).
Even after health care reform, persons from racial-ethnic minority groups continued to encounter more provider-related or logistic access barriers, compared with the White population. Examples from previous research of provider-related obstacles might include “providers don’t speak their language,” “providers don’t respect their religious belief,” or “providers don’t understand their culture” (43). Logistic access barriers might include difficulty obtaining transportation to receive mental health care or long wait times for health care services (43, 44). After implementation of the Medicaid expansion policy, the provider-related or logistic access barriers could have played a role as determinants of racial-ethnic disparities in mental health status over time.
The gradual influence of provider-related and logistic access barriers over time after the policy implementation might be observable in our analyses of proximal and distal associations. Analysis of the proximal period indicated that Medicaid expansion was associated with improvements in mental health for most racial-ethnic groups, which likely resulted from the policy’s reduction of financial barriers to health care access. However, racial-ethnic disparities in mental health became more evident in the analysis of distal associations of the policy with mental health status, indicating that barriers other than cost could be crucial long-term determinants of racial-ethnic mental health disparities. Accordingly, further efforts are necessary to investigate potential ways to address such barriers to maximize the benefit of Medicaid coverage. A potential solution is to emphasize linkages to mental care, which has been widely discussed as an effective approach to addressing logistic access barriers (45, 46).
Our discussion of Medicaid expansion and mental health status has primarily focused on access to health services because the policy was initially aimed at addressing financial barriers to health care access. As we have noted, Medicaid expansion could be associated with changes in mental health via other potential mechanisms, such as the quality of mental health care or levels of worry and distress about health-related expenses. It is important to consider additional mechanisms that may influence changes in mental health status, given that mental health was measured by self-reported poor mental health days rather than by use of a clinical measure.
Several limitations should be acknowledged. First, although the adjusted DID method can provide a relatively robust causal inference compared with traditional cross-sectional analysis controlling for underlying time-dependent trends, we might not have been able to rule out all potential confounding factors because of the observational study design. Second, the DID model can control only for unobservable factors that are constant over time. Factors that varied over time and were not included as covariates would bias the estimations. However, to the best of our knowledge, no major time-varying events influenced policy effects during our study period, and thus our study estimates should not have been affected by significant bias. Third, a valid interpretation of the DID and TD models relied on the parallel-trend assumption (see the online supplement ). The analyses for non-Hispanic Asian and AIAN groups, where the assumption was violated, may have biased interpretation of the results. Fourth, mental health outcomes in our study relied on self-reported health status instead of clinical diagnoses or claims databases involving professional medical judgment. However, self-reported subjective measures have been shown to correlate highly with objective measures of health outcomes (47, 48).
Conclusions
Medicaid expansion was significantly associated with improved mental health for the overall sample, although racial-ethnic disparities were observed. Because several barriers to accessing health care exist, Medicaid expansion alone may not be sufficient to improve mental health for all racial-ethnic groups in the long run. Further research is needed to understand why certain racial-ethnic groups benefited from Medicaid expansion whereas others did not. It is also necessary to investigate solutions to address potential barriers that contribute to racial-ethnic disparities in mental health.
Supplementary Material
File (appi.ps.20220394.ds001.docx)
- Download
- 40.19 KB
References
1.
Rosso RJ: US Health Care Coverage and Spending. Washington, DC, Congressional Research Service, 2021. https://crsreports.congress.gov/product/pdf/IF/IF10830
2.
Jost T: The Affordable Care Act and the Supreme Court: American health care reform inches forward despite dysfunctional political institutions and politics. Health Econ Policy Law 2013; 8:113–118
3.
Sommers BD, Epstein AM: US governors and the Medicaid expansion—no quick resolution in sight. N Engl J Med 2013; 368:496–499
4.
Status of State Medicaid Expansion Decisions: Interactive Map. San Francisco, Kaiser Family Foundation, 2023. https://www.kff.org/medicaid/issue-brief/status-of-state-medicaid-expansion-decisions-interactive-map. Accessed March 7, 2023
5.
Mazurenko O, Balio CP, Agarwal R, et al: The effects of Medicaid expansion under the ACA: a systematic review. Health Aff 2018; 37:944–950
6.
Ward B: The Impact of Medicaid Expansion on States’ Budgets. New York, Commonwealth Fund, 2020. https://doi.org/10.26099/5q66-1k77. Accessed March 7, 2023
7.
Cawley J, Soni A, Simon K: Third year of survey data shows continuing benefits of Medicaid expansions for low-income childless adults in the US. J Gen Intern Med 2018; 33:1495–1497
8.
Graves JA, Hatfield LA, Blot W, et al: Medicaid expansion slowed rates of health decline for low-income adults in southern states. Health Aff 2020; 39:67–76
9.
Lee H, Porell FW: The effect of the Affordable Care Act Medicaid expansion on disparities in access to care and health status. Med Care Res Rev 2020; 77:461–473
10.
Simon K, Soni A, Cawley J: The impact of health insurance on preventive care and health behaviors: evidence from the first two years of the ACA Medicaid expansions. J Policy Anal Manage 2017; 36:390–417
11.
Winkelman TNA, Chang VW: Medicaid expansion, mental health, and access to care among childless adults with and without chronic conditions. J Gen Intern Med 2018; 33:376–383
12.
McMorrow S, Gates JA, Long SK, et al: Medicaid expansion increased coverage, improved affordability, and reduced psychological distress for low-income parents. Health Aff 2017; 36:808–818
13.
Miller S, Wherry LR: Health and access to care during the first 2 years of the ACA Medicaid expansions. N Engl J Med 2017; 376:947–956
14.
Courtemanche C, Marton J, Ukert B, et al: Effects of the Affordable Care Act on health care access and self-assessed health after 3 years. Inquiry 2018; 55:46958018796361
15.
Courtemanche C, Marton J, Ukert B, et al: Early effects of the Affordable Care Act on health care access, risky health behaviors, and self-assessed health. South Econ J 2018; 84:660–691
16.
Wherry LR, Miller S: Early coverage, access, utilization, and health effects associated with the Affordable Care Act Medicaid expansions: a quasi-experimental study. Ann Intern Med 2016; 164:795–803
17.
Sommers BD, Blendon RJ, Orav EJ, et al: Changes in utilization and health among low-income adults after Medicaid expansion or expanded private insurance. JAMA Intern Med 2016; 176:1501–1509
18.
Sommers BD, Gunja MZ, Finegold K, et al: Changes in self-reported insurance coverage, access to care, and health under the Affordable Care Act. JAMA 2015; 314:366–374
19.
Soni A, Wherry LR, Simon KI: How have ACA insurance expansions affected health outcomes? Findings from the literature. Health Aff 2020; 39:371–378
20.
Walker ER, Cummings JR, Hockenberry JM, et al: Insurance status, use of mental health services, and unmet need for mental health care in the United States. Psychiatr Serv 2015; 66:578–584
21.
Baicker K, Taubman SL, Allen HL, et al: The Oregon experiment—effects of Medicaid on clinical outcomes. N Engl J Med 2013; 368:1713–1722
22.
Mojtabai R, Olfson M, Sampson NA, et al: Barriers to mental health treatment: results from the National Comorbidity Survey Replication. Psychol Med 2011; 41:1751–1761
23.
Courtemanche C, Marton J, Ukert B, et al: The impact of the Affordable Care Act on health care access and self-assessed health in the Trump era (2017–2018). Health Serv Res 2020; 55:841–850
24.
Manuel JI: Racial/ethnic and gender disparities in health care use and access. Health Serv Res 2018; 53:1407–1429
25.
Vandenbroucke JP, von Elm E, Altman DG, et al: Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. Ann Intern Med 2007; 147:W163–W194
26.
Behavioral Risk Factor Surveillance System (BRFSS). Atlanta, Centers for Disease Control and Prevention. https://www.cdc.gov/brfss/about/index.htm. Accessed March 7, 2023
27.
Behavioral Risk Factor Surveillance System Survey Data. Atlanta, Centers for Disease Control and Prevention. https://www.cdc.gov/brfss/data_documentation/index.htm. Accessed March 7, 2023
28.
Ettman CK, Abdalla SM, Cohen GH, et al: Prevalence of depression symptoms in US adults before and during the COVID-19 pandemic. JAMA Netw Open 2020; 3:e2019686
29.
Gai Y, Marthinsen J: Medicaid expansion, HIV testing, and HIV-related risk behaviors in the United States, 2010–2017. Am J Public Health 2019; 109:1404–1412
30.
Medicaid Eligibility for Adults as of January 1, 2014. San Francisco, Kaiser Family Foundation, 2013. https://www.kff.org/wp-content/uploads/sites/2/2013/12/8497-medicaid-eligibility-for-adults-as-of-january-1-2014.pdf
31.
Measuring Healthy Days: Population Assessment of Health-Related Quality of Life. Atlanta, Centers for Disease Control and Prevention, 2020. https://www.cdc.gov/hrqol/pdfs/mhd.pdf
32.
Cree RA, Okoro CA, Zack MM, et al: Frequent mental distress among adults, by disability status, disability type, and selected characteristics—United States, 2018. MMWR Morb Mortal Wkly Rep 2020; 69:1238–1243
33.
Slabaugh SL, Shah M, Zack M, et al: Leveraging health-related quality of life in population health management: the case for healthy days. Popul Health Manag 2017; 20:13–22
34.
Bossarte RM, He H, Claassen CA, et al: Development and validation of a 6-day standard for the identification of frequent mental distress. Soc Psychiatry Psychiatr Epidemiol 2011; 46:403–411
35.
Bertrand M, Duflo E, Mullainathan S: How much should we trust differences-in-differences estimates? Q J Econ 2004; 119:249–275
36.
Local Area Unemployment Statistics. Washington, DC, US Bureau of Labor Statistics. https://www.bls.gov/lau/#tables. Accessed March 7, 2023
37.
Dimick JB, Ryan AM: Methods for evaluating changes in health care policy: the difference-in-differences approach. JAMA 2014; 312:2401–2402
38.
Olden A, Møen J: The Triple Difference Estimator. Norwegian School of Economics, Department of Business and Management Science, discussion paper 2020/1. Rochester, NY, Elsevier SSRN, 2020. https://doi.org/10.2139/ssrn.3582447
39.
Griffith KN, Bor JH: Changes in health care access, behaviors, and self-reported health among low-income US adults through the fourth year of the Affordable Care Act. Med Care 2020; 58:574–578
40.
Breslau J, Han B, Lai J, et al: Impact of the Affordable Care Act Medicaid expansion on utilization of mental health care. Med Care 2020; 58:757–762
41.
Alang SM: Mental health care among Blacks in America: confronting racism and constructing solutions. Health Serv Res 2019; 54:346–355
42.
Alegría M, Lin J, Chen CN, et al: The impact of insurance coverage in diminishing racial and ethnic disparities in behavioral health services. Health Serv Res 2012; 47:1322–1344
43.
Call KT, McAlpine DD, Garcia CM, et al: Barriers to care in an ethnically diverse publicly insured population: is health care reform enough? Med Care 2014; 52:720–727
44.
Weech-Maldonado R, Morales LS, Elliott M, et al: Race/ethnicity, language, and patients' assessments of care in Medicaid managed care. Health Serv Res 2003; 38:789–808
45.
Fuller JD, Perkins D, Parker S, et al: Effectiveness of service linkages in primary mental health care: a narrative review part 1. BMC Health Serv Res 2011; 11:72
46.
Fuller JD, Perkins D, Parker S, et al: Building effective service linkages in primary mental health care: a narrative review part 2. BMC Health Serv Res 2011; 11:66
47.
DeSalvo KB, Bloser N, Reynolds K, et al: Mortality prediction with a single general self-rated health question. A meta-analysis. J Gen Intern Med 2006; 21:267–275
48.
Miilunpalo S, Vuori I, Oja P, et al: Self-rated health status as a health measure: the predictive value of self-reported health status on the use of physician services and on mortality in the working-age population. J Clin Epidemiol 1997; 50:517–528
Information & Authors
Information
Published In
History
Received: 30 July 2022
Revision received: 8 November 2022
Revision received: 10 January 2023
Accepted: 16 February 2023
Published online: 5 April 2023
Published in print: November 01, 2023
Keywords
Authors
Author Contributions
Mr. Lai and Ms. McDaniel contributed equally to this work.
Competing Interests
The Behavioral Risk Factor Surveillance System data set is an open resource available from the Population Health Surveillance Branch of the Centers for Disease Control and Prevention website.
The authors report no financial relationships with commercial interests.
Metrics & Citations
Metrics
Citations
Export Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
For more information or tips please see 'Downloading to a citation manager' in the Help menu.
View Options
View options
PDF/EPUB
View PDF/EPUBLogin options
Already a subscriber? Access your subscription through your login credentials or your institution for full access to this article.
Personal login Institutional Login Open Athens loginNot a subscriber?
PsychiatryOnline subscription options offer access to the DSM-5-TR® library, books, journals, CME, and patient resources. This all-in-one virtual library provides psychiatrists and mental health professionals with key resources for diagnosis, treatment, research, and professional development.
Need more help? PsychiatryOnline Customer Service may be reached by emailing [email protected] or by calling 800-368-5777 (in the U.S.) or 703-907-7322 (outside the U.S.).