Can Medicaid be a Solution to the Problem? Underinsurance in Medicaid Expansion Versus Non-Expansion States

Abstract

The positive effects of Medicaid expansions have been extensively documented in the literature. However, it is not clear whether the reform has had an equally meaningful effect with respect to underinsurance, which is the state of having health insurance yet lacking adequate coverage or facing substantial financial risks upon usage of services. Based on a quasi-experimental difference-in-differences approach, we analyzed the data from a nationally representative sample to estimate the effect of Medicaid expansion on the probability of underinsurance among the non-elderly low-income adult population of the U.S. We found no evidence of significant changes in the likelihood of underinsurance due to Medicaid expansion during the first 4 years after the ACA implementation. However, a supplementary analysis of the longer-term impact (2018-2019) suggests that there might be a time lag between Medicaid expansion and its effect on underinsurance. It is important to realize that expansion of coverage alone may not be sufficient to protect millions of Americans, particularly those with low incomes, from underinsurance. It is, therefore, crucial for policymakers to build legislative frameworks that protect individuals from excessive healthcare expenses and prevent treatment avoidance or delay.

Keywords

underinsurance out-of-pocket spending access to care Medicaid expansion Affordable Care Act

What do we already know about this topic?

Underinsurance rate in the U.S. has increased during the last decade and it is associated with detrimental effects on healthcare utilization and outcomes.

How does your research contribute to the field?

It is a severely understudied area of research. Our study is the first one, to our knowledge, to explore the link between a policy intervention and underinsurance.

What are your research’s implications toward theory, practice, or policy?

While Medicaid expansion might be a solution to the problem, expansion of coverage alone will not be sufficient to protect individuals from underinsurance risks. It is, therefore, critical for policymakers to design policies that ensure the adequacy of health insurance coverage. Additionally, states appear to have shortcomings in their ability to monitor Medicaid cost-sharing that exceeds 5% of a family’s income. This, coupled with various obstacles to accessing healthcare, could potentially compel beneficiaries to cover healthcare expenses directly out-of-pocket.

Introduction

The Affordable Care Act (ACA) marked a monumental transformation of the U.S. healthcare system, incorporating a multitude of pivotal policy changes aimed at enhancing access to healthcare and improving insurance coverage for millions of Americans, particularly those with low incomes. This comprehensive legislation introduced key provisions such as the establishment of Health Insurance Exchanges, introduction of protections for individuals with pre-existing conditions, provision of insurance subsidies and tax credits among many other important elements. One of the most significant policy changes within the ACA was the expansion of Medicaid eligibility to non-elderly adults with incomes below 138% of the Federal Poverty Level (FPL). Medicaid expansion was widely recognized as the primary catalyst for reducing the rate of uninsurance across the country, addressing a pressing concern and providing much-needed access to healthcare for those who were previously uninsured. While the original law intended to expand the program eligibility across the U.S., the Supreme court decision of unlawfulness of such a requirement hindered a nationwide implementation of Medicaid expansions.¹ By the end of our study period in 2019, thirty-five states including the District of Columbia expanded the program, while 16 had kept the income eligibility at prior levels.²

The positive effects of Medicaid expansions on a range of outcomes have been extensively documented in the literature³ and estimates suggest that up to 60% of insurance coverage gains could be attributed to Medicaid.⁴ However, it is not clear whether the law has had an equally meaningful effect with respect to underinsurance, which is defined as “the state in which persons who have medical coverage are still exposed because of medical care expenses to financial risk that causes them some form of harm.”⁵ While the primary goal of Medicaid expansion was to reduce uninsurance, it is possible that the expansion could have had an indirect impact on underinsurance in either positive or negative direction. Having access to insurance per se may not necessarily mean that a person has full financial protection from healthcare risks as is the case in many countries with universal healthcare systems. Research conducted between 1985 and 2005, which also employed a cost-based approach to conceptualize underinsurance, yielded varying rates of underinsurance in the U.S. ranging from 7% to 39%.⁶ More recent findings by Magge et al⁷ indicate that prior to the enactment of the ACA in 2010, up to 34.5% of low-income adults may have experienced underinsurance. The literature suggests that underinsurance is very similar to uninsurance in the way it leads to detrimental effects in healthcare utilization as well as outcomes.^5,8

The adequacy of health insurance can be assessed across several domains and scholars have previously used economic factors (eg, out-of-pocket expenditures exceeding a threshold of household income), health plan benefit design features (eg, lack of coverage of certain services) or access barriers (eg, forgoing care due to cost) to operationalize the concept.⁵ The challenge is that no data exists or is at least accessible to researchers to study underinsurance based on all possible domains. Threshold measures of underinsurance have been in wide use since its first introduction in the 1980s.^9
-11 Such measures are, however, based solely on actual utilization of healthcare services and can be seen as an “ex-post” definition of the notion. Hence, threshold measures are likely to underestimate the extent to which persons have less generous plans.¹² In order to consider “ex-ante” (ie, before utilization) factors, we include forgoing or delaying care due to cost as another domain of underinsurance in the study. This is consistent with prior research on the topic^7,13 and captures individuals who spend a small proportion of their income on health care costs but might still be burdened by financial risks associated with potential use of services.¹⁴ In the context of a significant rise in high-deductible plans available on the market in recent years,¹⁵ it becomes especially pertinent to note that individuals opting for these plans pay reduced premiums but might face the necessity of forgoing or postponing healthcare in order to evade the burden of high-deductible charges. This situation may be particularly applicable to low-income individuals who have to contend with competing financial obligations such as buying food or paying rent.¹⁶

In this study, our ultimate objective was to estimate the indirect effect of Medicaid expansion on underinsurance among non-elderly low-income (<138% FPL) adult population of the U.S, who were the main target of the intervention. To achieve this goal, we used a plausibly exogenous variation in the probability of insurance coverage due to Medicaid expansion decision between states that opted to expand and those that had not done so.

Conceptual Framework

We conceptualize the effect of Medicaid expansion on underinsurance based on fundamental aspects of economic theory as it relates to individual’s demand for healthcare services. As millions of newly and previously eligible working-age adults gained insurance,⁴ one may have expected to observe increased demand for healthcare services. Indeed, prior studies demonstrated that Medicaid expansion was correlated with enhanced use of primary as well as preventative care,^17,18 outpatient care,¹⁹ and prescription medications.²⁰ Two related economic concepts could explain such trends in utilization: moral hazard and price elasticity of demand. The former suggests that having access to insurance results in increased demand for healthcare services because it essentially reduces the price for a consumer.²¹ In his seminal work, Pauly²² emphasizes that “the response of seeking more medical care with insurance than in its absence is a result not of moral perfidy, but of rational economic behavior.” The fact that the ACA was signed into law in 2010 and the presence of early expansion states might have also contributed to “pent-up demand” for care upon wide-scale implementation of the provision in 2014.²³

While the concept of moral hazard refers to the direction of the effect of coverage on the quantity of care demanded, price elasticity helps us quantify that effect in the form of a single unitless number.²⁴ Health services generally have a price elasticity of about −0.2, meaning that an increase of 10% in the out-of-pocket price reduces the use of services by 2%.²⁴ Low-income households who were the target of Medicaid expansion are particularly sensitive to price changes and it is well-established that even small out-of-pocket payments can sometimes present barriers to care for highly vulnerable populations.^25,26 The findings from the influential RAND Health Insurance Experiment provided evidence that low-income groups had twice the price responsiveness compared to high-income counterparts.²⁷ For this very reason, enhanced demand is expected among Medicaid beneficiaries in expansion regions. In the absence of the intervention, the majority of low-income households in non-expansion states were left in a “coverage gap” without access to Medicaid or any other subsidies introduced as part of the reform. For them, the price of healthcare services would have increased due to inflationary trends alone and would have made them bear the full cost of insurance. This, in turn, would have led to a lower demand among low-income population. However, price elasticity estimates differ considerably across types of healthcare services.²⁴ For instance, hospital care tends to be less sensitive to price changes and it is, therefore, not surprising to see mixed results with regards to the association between Medicaid expansion and inpatient hospital care.²⁸

As it relates to this study, the question is whether policy induced changes in price of and demand for care affect the probability of underinsurance? The first criterion used to define underinsurance in the study (ie, percent of income spent on out-of-pocket payments) is a function of unit price and the quantity of care encountered by a patient. Medicaid offers comprehensive coverage and substantially limits out-of-pocket expenditures.²⁹ While many states have cost-sharing for parents or other adults covered by Medicaid (eg, $4 for outpatient services and $8 for non-emergency use of the emergency department),²⁹ the amounts are nominal relative to private insurance coverage or out-of-pocket expenditures incurred without insurance. Even though increased demand is expected, most of the bill is to be covered by the government in this case. Other insurance dynamics such as the “crowd-out” of private insurance with Medicaid^30
-32 or individuals switching from employer-provided to subsidized insurance³³ may have led to improved coverage with potentially lower deductibles and copayments. Hence, it is plausible to foresee reduced out-of-pocket expenditures among Medicaid enrollees and decreased probability of criterion-specific underinsurance among low-income individuals in expansion states. Because this threshold definition is a function of actual utilization of care, non-expansion regions may also observe declines in criterion-specific underinsurance. In other words, people with insurance policies less generous than Medicaid would be subject to increased prices and underutilize healthcare services as a result of high price elasticity and “reverse” moral hazard.¹⁰ Thus, individuals in non-expansion states would become less likely to be underinsured under out-of-pocket threshold definition alone.

Underutilization of care is, however, to be picked up by the second criterion of underinsurance used in this study: whether a person failed to receive or delayed medical treatment or prescription medication due to cost. Whereas those in expansion states would have a financial incentive to use more care, insured individuals in non-expansion regions may be more likely to delay and forgo care compared to pre-ACA period, especially as healthcare cost growth tends to outpace income growth.^34,35 This might be particularly evident in the longer-term period.

Therefore, we hypothesize that the overall probability of underinsurance would move in opposite directions in Medicaid expansion versus non-expansion states, decreasing in the former and increasing in the latter regions. The empirical component of the study intends to test this hypothesis and is presented in the following sections.

Methods

Data and Study Population

The data for this study comes from the Medical Expenditure Panel Survey (MEPS), which is a nationally representative survey of healthcare utilization, insurance coverage and cost of care among the civilian non-institutionalized population of the United States.³⁰ The interviews are conducted in 5 rounds during a 2-year period, and new respondents are added on yearly basis. The survey instrument as well as detailed methodology reports are publicly available on the official website of the U.S. Department of Health and Human Services.³⁰ We created the analytic file by obtaining access to restricted state identifiers and matching them with the out-of-pocket spending, ability to receive care, insurance coverage, income, and other demographic variables publicly available in the Household Component (MEPS-HC) of the survey data from 2010 through 2017. In our supplementary analysis, we included 2 additional years of data (2018-2019).

For the main analysis, the sample was restricted to non-elderly adults (aged 19-64) who resided in one of the 44 U.S. states (see Online Appendix I) and were continuously insured within a year. Those who may have switched between insurance types were included as long as they did not have a spell of uninsurance. In addition, we only included individuals with incomes below 138% FPL to focus the analysis on the population potentially eligible for the ACA’s Medicaid expansion. We also excluded observations with zero-sampling weights and missing values for covariates if the overall rate of missingness was less than 1% for each variable. The flow diagram of the study population highlights the importance of each inclusion and exclusion step (Figure 1A in Online Appendix I). The resulting analytic sample contained 17 677 observations, which represents over 15.7 million U.S. adults after survey weighting. The study protocol was reviewed by the Office of Human Research Ethics which determined that the study does not require IRB approval.

Expansion Status

Individuals living in states that expanded Medicaid through the ACA or by introducing an equivalent program on January 1, 2014 (n = 26) formed the intervention group for this study, whereas the control group consisted of adults residing in states that had not expanded Medicaid eligibility during the study period (n = 18). Following prior research evaluating the effects of Medicaid expansions,^36,37 we excluded 7 states that implemented the expansion between January 1, 2014 and December 31, 2017. Even though Wisconsin has not formally adopted the ACA’s Medicaid expansion, we consider it an expansion state because on January 1, 2014, the state extended Medicaid coverage to adults with incomes up to 100% FPL under its Section 1115 waiver program. In addition to 7 states excluded from the main analyses, we dropped observations from Maine and Virginia in our supplementary analysis because those states expanded Medicaid on Jan 10, 2019, and Jan 1, 2019, respectively. Hence, the control group in supplementary analysis includes 16 states. Full classification of states can be found in the Online Appendix I.

Outcomes

The primary outcome of interest—underinsurance—was defined using 2-part criteria: (1) out-of-pocket expenditures exceeding 5% of household income; or (2) whether a person failed to receive or delayed medical care or prescription medication due to cost. A respondent was considered underinsured if at least one criterion was met within a year.

Out-of-pocket spending includes deductible, coinsurance and copayment amounts as well as any direct payments for healthcare services encountered by a respondent outside of insurance coverage.³⁸ In the MEPS, annual premium contributions are only available for private insurance policy holders and, hence, were excluded from the definition of out-of-pocket spending in this study.

As for the second criterion of underinsurance, the following response options to a 2-part survey question were flagged to identify those who failed to receive or delayed care or medication: “could not afford care,” “insurance company would not approve/cover/pay,” “doctor refused family plan” and “was refused services.” Those who chose other response options such as “problems getting to doctor’s office,” “could not get time off work,” “don’t know where to go to get care”, “did not have time or took too long” and “other” were not counted as underinsured. In 2018, the MEPS replaced this sequence of cost-related access issue questions with a single question that inquires directly if a respondent forwent or delayed care or prescription drugs due to cost.³⁹ It is important to mention that the “due to cost” component of the question may have been perceived more broadly by respondents. While we attempted to be broadly inclusive in choosing response options from earlier years of the survey, we cannot determine if respondents’ interpretation of the question perfectly aligns across years. In fact, the question change resulted in a noticeable jump in the count of underinsured individuals in 2018 and 2019 (see Online Appendix I).

Statistical Analyses

We used a linear probability model within the framework of a quasi-experimental difference-in-differences (DiD) approach to explore the impact of the ACA’s Medicaid expansion on the probability of underinsurance among low-income individuals. The changes relative to pre-ACA years (2010-2013) were analyzed at short- (2014-2015) and mid-term (2016-2017) time periods. The regression models included separate interaction terms between the expansion status indicator and each post-ACA implementation time period indicator. The coefficients on these interaction terms indicate the changes in the outcome due to Medicaid expansion, that is, respective DiD estimates. This is reflective of the approach used in previous studies and allows for examination of dynamic trends.³⁶ To better understand the main results and disentangle the impact of Medicaid expansion on underinsurance, we conducted criterion-specific analyses by replacing overall underinsurance outcome with each of the 2 criteria separately. Our supplementary analysis of long-term (2018-2019) effects of the intervention followed the identical approach described above with inclusion of an additional interaction term. Recognizing the importance of exploring treatment dynamics and heterogeneity further, we incorporated the latest DiD approach proposed by Callaway and Sant’Anna.⁴⁰ Using this method, we were able to examine the effects of Medicaid expansion on underinsurance while considering the temporal dynamics of the intervention and accounting for potential heterogeneity across different groups. The inclusion of the not yet treated cohort allowed us to gain valuable insights into the evolving effects of the policy. For more details on the model specification and the results from the alternative DiD implementation, please refer to the Online Appendices I and II.

DiD rests on the assumptions that trends are parallel, and no other interventions were happening at the same time.⁴¹ While it is impossible to rule out the latter in the case of a widescale reform such as the ACA, we performed visual and formal tests of the parallel trend assumption using the data from pre-intervention periods. The differences in pre-trends between expansion and non-expansion states were statistically insignificant in both adjusted and unadjusted models. Although covariate balancing is not a prerequisite for DiD studies, the fact that intervention and control groups were similar across most of sociodemographic characteristics (Table 1) adds further validity to the identification strategy.⁴¹

Table 1.

Characteristics of the Study Sample by Medicaid Expansion Status.

	Pre-ACA		ACA
	Expansion	Non-expansion^a	Expansion	Non-expansion^a
	N = 5291	N = 3037	N = 6170	N = 3179
U.S. weighted population	9 247 986	5 530 817	11 022 397	5 675 093
Insurance type (%)
Medicaid	51.2	29.3 *	62.3	28.9 *
Medicare	6.7	10.7 *	5.9	11.0 *
Private	30.6	45.0 *	21.3	42.9 *
Dual Medicaid/Medicare	9.2	10.6 *	8.2	12.9 *
Other (e.g., VA or TRICARE)	2.4	4.5 *	2.3	4.2 *
Age (years)	40.0 (15.9)	40.5 (15.8)	40.0 (16.3)	41.9 (16.9) *
Sex (%)
Female	60.7	59.9	58.8	62.4 *
Race/ethnicity (%)
White	69.8	68.5 *	70.3	63.4 *
Black	18.3	27.0 *	17.5	28.4 *
American Indian/Alaska Native	0.9	1.2 *	1.1	1.3 *
Asian/Hawaiian/Pacific Islander	7.4	2.1 *	7.2	3.1 *
Multiple Races	3.5	1.2 *	3.9	3.8 *
Marital status (%)
Married	31.5	32.5	27.6	30.9
Family size	2.9 (1.9)	2.7* (1.8)	2.9 (1.9)	2.7* (1.8)
Education (%)
No degree	27.6	25.5	27.5	23.0
High school	37.7	39.3	37.6	38.0
College degree	8.3	6.4	7.7	8.5
Graduate degree	2.2	1.8	2.3	2.2
Other	11.1	13.5	12.3	15.9
Inapplicable	13.8	13.5	12.5	12.5
Employment status (%)
Employed	40.7	42.7	44.4	45.6
Self-reported health status (%)
Excellent	17.3	17.5	18.2	18.1
Very good	25.0	24.4	24.0	25.0
Good	31.0	28.7	32.5	27.8
Fair	17.5	20.4	17.9	20.4
Poor	9.2	9.0	7.4	8.7
Self-reported mental health status (%)
Excellent	27.2	28.3	28.2	27.6
Very good	24.4	25.0	22.7	22.5
Good	29.2	27.2	30.9	29.9
Fair	14.3	14.4	14.1	15.6
Poor	4.9	5.1	4.1	4.5
Family income	16 125.8 (14 072.7)	16 614.3 (13 123.5)	15 828.0 (14 678.3)	16 052.8 (14 191.2)
Out-of-pocket spending on healthcare	374.3 (1360.6)	504.3 (2066.4)	267.0 (1086.8)	432.2 * (1355.8)

Source. Author’s analysis of data from the Medical Expenditure Panel Survey–Household Component, 2010 to 17.

Note. Data are presented as mean (SD) for continuous variables and as percentages for categorical variables. Values are weighted to be nationally representative. U.S. dollars are adjusted for inflation to 2021 using the GDP index.

Significance results in this column are shown for the difference between expansion and non-expansion groups, obtained using chi-square (categorical variables) and t-test (continuous variables).

P < .05.

We used survey weights, strata, and clusters provided by the MEPS in all regression analysis. Cluster-robust standard errors were computed to address correlation within each sampling unit and across years. Out-of-pocket expenditure and income variables were adjusted for inflation to 2021 dollars using the Gross Domestic Product (GDP) price index.⁴² All analyses were completed using Stata, version 16.1 (StataCorp, TX).

In the first set of sensitivity analyses, we modified the definition of underinsurance by extending the first criterion to those who spent over 10% of their income on healthcare services instead of 5%. Another modification of the outcome was to include dental care expenses as well as forgoing and delaying dental treatment in the definition of the concept. Private insurance plans usually separate dental expenses from health benefits, whereas Medicaid coverage of dental treatment differs from one state to another.⁴³ However, it may still be important to assess the underinsurance burden of dental care.

The second series of additional analyses was concerned with alternative sample definitions. In one, we removed individuals with zero or negative incomes because any out-of-pocket spending on healthcare would have identified them as underinsured. In the other, observations for Medicare and dual Medicare/Medicaid enrollees were dropped as their medical needs are substantially different from the rest of non-elderly adult population and are significantly associated with higher healthcare expenditures.⁴⁴ We also excluded individuals under the age of 26, who would have had different coverage options due to the implementation of the ACA’s young adult provision. Moreover, we ran the regression model including in the sample those who were uninsured at least for a month within a year, thus, relaxing the continuous coverage requirement.

We also re-analyzed the data by implementing alternative classifications of expansion status: excluding 5 states (Delaware, Hawaii, Massachusetts, New York and Vermont) that offered Medicaid coverage to persons with household incomes up to 100% FPL before the ACA; removing Wisconsin; and including 2 states (Michigan and New Hampshire) that introduced Medicaid expansion with a slight delay in 2014.

Finally, we conducted subpopulation analysis based on income status, separating those in 100% to 138% FPL bracket from the rest of the sample. As part of the ACA, this group in non-expansion states became eligible for substantial premium and cost-sharing subsidies. This suggests that trends might be different in this group compared to those below 100% FPL.⁴⁵

Limitations

The study results should be interpreted within the context of the following limitations. First, benefits design is another domain of underinsurance definition which we were not able to include in this study. In this regard, the current definition does not allow the effect estimation among those who are insured and healthy but may be subject to high healthcare costs in case medical need arises.

Second, the re-design of the MEPS survey with respect to the second underinsurance criterion noticeably affected the number of underinsured individuals we observed during the long-term follow-up period. While there is no reason to expect that the questionnaire re-design impacted respondents in intervention and control groups differently, the long-term results produced as part of our supplementary analysis should be interpreted with caution.

Third, pre-trend test results under the latest DiD approaches proposed by Callaway and Sant’Anna⁴⁰ showed that not all pre-treatment differences are statistically equal to zero. However, we should note that estimates were fairly noisy across the years preceding the intervention.

Finally, despite using DiD approach with a number of adjustment variables, one cannot guarantee that all unmeasured confounding factors were accounted for. There is still a possibility of residual confounding from unobserved factors that may have biased our findings. In addition, some of the intervention group states expanded Medicaid using the Section 1115 waivers.³⁷ Those programs were not exactly equivalent to the ACA’s Medicaid expansions, but those features are unlikely to affect the study findings in a meaningful way.

Results

The study population comprised of 11 461 survey respondents from Medicaid expansion states and 6 216 individuals from the states that had not expanded the program. In the pre-ACA cohort, low-income adults from the expansion regions were much more likely to be covered by Medicaid. With respect to other sociodemographic characteristics, we found no statistically significant differences between individuals in 2 groups across all measures except race and family size (Table 1). When compared to the pre-ACA period, the differences in sociodemographic composition of intervention and control groups remained consistent in the years after the ACA implementation. The only noticeable changes were with respect to age, sex and out-of-pocket healthcare expenditure variables, for which the differences between 2 groups became significant during the ACA period. The subjects in the control group were more likely to be older and identify as female. On average, adults in non-expansion versus expansion states would also spend $165 more on medical care and prescription medications after the intervention compared to the difference of $130 before the ACA.

The results of our primary analyses suggest that despite having full-year health insurance coverage, a quarter of intervention group adults experienced underinsurance between 2010 and 2013. The share of underinsured people in the control group was even higher, at 28.5%. While the probabilities declined in subsequent two periods (2014-2015 and 2016-2017), we did not observe statistically significant differential trends between the intervention and control groups [1.1 percentage points; CI: −3.4, 5.5; P = .64 and 1.7 percentage points; CI: −2.6, 6.0; P = .44]. As can be seen from the graph of annual trends (Figure 1), underinsurance rates had been consistently lower in expansion compared to non-expansion states both in pre- and post-ACA implementation periods. Criterion-specific analyses showed patterns similar to the ones observed in overall underinsurance trends with declining within-group differences and statistically indistinguishable DiD estimates between Medicaid expansion and non-expansion states. A full set of main results is presented in Table 2.

Figure 1.

Annual underinsurance trends in medicaid expansion versus non-expansion states.

Table 2.

Changes in Probability of Overall and Criterion-Specific Underinsurance Among Non-Elderly Adults in Medicaid Expansion Versus Non-Expansion States.

		Expansion States			Non-Expansion States			Difference-in-Differences
		Pre-ACA	Post-ACA	Within-group Difference^a	Pre-ACA	Post-ACA	Within-group Difference^a	Unadjusted^a	Adjusted^a
Short-term Period (2014-2015)	Overall underinsurance	24.3%	20.0%	−4.3	28.5%	24.8%	−3.7	−0.6	1.1
	OOP spending over 5% of income	19.7%	15.4%	−4.3	22.9%	19.4%	−3.5	−0.8	0.7
	Forgoing/delaying care due to cost	7.2%	6.5%	−0.7	7.9%	7.9%	0.0	−0.7	−0.2
Mid-term Period (2016-2017)	Overall underinsurance	24.3%	18.9%	−5.4	28.5%	22.1%	−6.4	1.0	1.7
	OOP spending over 5% of income	19.7%	14.7%	−5.0	22.9%	18.8%	−4.1	−0.9	−0.3
	Forgoing/delaying care due to cost	7.2%	5.4%	−1.8	7.9%	6.9%	−1.0	−0.8	−0.4

Source. Author’s analysis of data from the Medical Expenditure Panel Survey–Household Component, 2010 to 17.

Note. Adjusted analyses control for age, sex, race, education, marital and employment status, family size, physical and mental health status as well as state and year fixed effects. Values are weighted to be nationally representative. The sample includes adults (ages 19-64) with incomes below 138% Federal Poverty Level who were continuously insured within a year (N = 17 677, weighted N = 15 738 146).

Percentage points.

P < .05.

When 2 more years of data were added in supplemental analysis, we found statistically significant evidence of lower probability of underinsurance in Medicaid expansion versus non-expansion states in the long-term (2018-2019) period. The adjusted change was equal to −5.4 percentage points [CI: −10.4, −0.4; P = .03]. The differences in the probability became more apparent as the rate of change was more pronounced for adults residing in non-expansion states (see Online Appendix I). Specifically, the probability of underinsurance defined solely by out-of-pocket spending threshold was 4.1 percentage points [CI: −8.5, .4; P = .07] greater in the control group. If we were to define the outcome by the second criterion (ie, forgoing/delaying care due to cost) only, individuals in non-expansion states would have had a 5.1-percentage-points [CI: −8.6, −1.6; P = .01] higher likelihood of being underinsured compared to counterparts in expansion states.

Study findings were qualitatively unaffected when we modified the outcome definition; excluded Medicare and dual eligibles, or adults with zero incomes; allowed for uninsurance spells; or applied alternative classifications of expansion status (see Online Appendix I). We still found no evidence of changes due to Medicaid expansion in short- and mid-term time periods. The only exception were the results of analysis based on income thresholds, where we noticed a statistically significant higher probability of underinsurance among expansion state subjects (vs non-expansion) in years 2014 to 2015.

Discussion

The implementation of the ACA has brought about significant changes in the insurance landscape, including expansions of Medicaid, the establishment of insurance marketplaces, and the availability of subsidies to help individuals afford coverage. These changes have led to a substantial increase in insurance coverage rates across the population, with millions of individuals gaining access to health insurance. As a result, the rate of uninsurance among the nonelderly population declined from 16.8% in 2013 to 10.0% in 2016.⁴⁶ In a comprehensive assessment of the ACA’s effects, Frean and colleagues⁴ estimated that 60% of the coverage gains could be attributed to Medicaid expansions, while 40% can be explained by premium subsidies and other provisions aimed at enhancing insurance value. It is not surprising that the increase in coverage was not evenly distributed between expansion and non-expansion states, with the proportion of insured individuals in expansion states increasing twice as much as in non-expansion states (5.9 vs 2.5 percentage points).³³

While the ACA has had a profound impact on expanding insurance coverage, it remains to be seen whether the reform has had an effect on protecting people from the threat of underinsurance. It was not an explicit aim of the reform but many of the law provisions should have reduced the burden associated with inadequate insurance protections. For example, Medicaid tends to provide the most comprehensive coverage with remarkably low cost-sharing and the extension of program benefits under the ACA to populations with incomes below 138% FPL should be considered a potential solution to the problem of underinsurance. However, a recent report from the Commonwealth Fund indicates that the share of inadequately covered individuals in the U.S. has increased from 16% in 2010 to 23% in 2018.⁴⁷ To our knowledge, this is the first study evaluating the impact of the ACA, and Medicaid expansion in particular, on underinsurance using a nationally representative sample. Based on the analysis of 8 years of data, we found no evidence of significant changes in the likelihood of underinsurance due to Medicaid expansion during the first 4 years after the ACA implementation. However, when we included 2 more years of data, there was a positive effect in the fifth and sixth years after the reform. To be precise, a point estimate of a 5.4-percentage-point decrease would mean that the intervention prevented 5.3 millions of ninety-nine million non-elderly adults in expansion states⁴⁸ from being exposed to underinsurance risks in 2019.

A potential explanation for the absence of Medicaid expansion effects during short- and mid-term periods might be related to a slow implementation of the program and general confusion among the public and officials as the expansions were rolled out. It is estimated that 4.2 million non-elderly adults were eligible for Medicaid but still uninsured in 2021 and two-thirds of them lived in expansion states.⁴⁹ Another reason could be that other elements of the ACA had had a disproportionately positive impact on non-expansion states. For instance, one of the important provisions under the ACA was the removal of patient cost-sharing for preventative care services. As a larger proportion of adults in non-expansion states are privately insured, this provision may have benefited them in a similar way Medicaid expansion was advantageous to new Medicaid enrollees in expansion states; thus, offsetting some of the losses due to non-expansion. Additionally, there is evidence suggesting that previously eligible populations gained Medicaid coverage post-ACA (“woodwork effect”) in non-expansion states as well,⁴ which may have further contributed to parallel trends we observed between 2014 and 2017. Previous research examining Medicaid expansions also documented gradual increases in the impact of the reform and commonly found no effect during the early years of the implementation.³⁶ While we suspect that our long-term findings are affected by the survey redesign, there may have been other factors that contributed to differential effects with regard to underinsurance. One significant factor to consider is the federal approval of extended-duration short-term insurance plans, which occurred in 2018.⁵⁰ These plans provide an alternative coverage option but often come with limited benefits and increased financial risks for individuals. It is worth noting that the majority of non-expansion states do not have bans or restrictions on the duration of these plans, unlike many expansion states.⁵¹ Evidence suggests that states permitting short-term insurance coverage for durations close to 1 year witnessed a 27% decrease in enrollment in ACA-compliant plans not offered through the exchanges.⁵⁰ The availability and uptake of these plans could have had an impact on underinsurance rates by offering lower levels of coverage compared to comprehensive insurance plans.

Consistent with findings from previous studies conducted after the implementation of the ACA,³ our descriptive analysis reveals a significant rise in the percentage of Medicaid beneficiaries in states that expanded the program. However, we should note that a substantial part of increase in Medicaid enrollment can be attributed to those who were previously on private coverage. This phenomenon is commonly referred to as “crowd-out".³⁰ A number of studies have now documented crowd-out of private insurance by Medicaid among low-income populations.^31,32,52 While we do not observe similar trends in non-expansion states, private insurance remains to be the largest coverage option for low-income people in those regions. In light of our null findings with regard to underinsurance, this may suggest that the expansion of Medicaid may not be the sole factor influencing underinsurance rates. The availability and accessibility of private insurance plans including subsidized options offered through insurance marketplace might have influenced individuals’ insurance choices and the overall dynamics of the insurance market. While our study primarily focuses on the effects of Medicaid expansion on underinsurance among the low-income population targeted by the expansion, it is crucial to acknowledge the parallel dynamics in private insurance market in non-expansion states and its potential impact on underinsurance rates.

Income-based subpopulation analysis in our study showed patterns similar to the main results when the sample was restricted to those with incomes below 100% FPL. Contrastingly, we noticed a statistically significant differential effect between expansion and non-expansion states when the analysis included adults within 100% to 138% FPL income bracket. Though the effect was present only in the short-term period. This may be explained by the fact that individuals with incomes between 100 and 138% FPL in non-expansion states became eligible for substantial cost-sharing subsidies under the ACA,⁵³ which may have provided underinsurance protections similar to or even more substantial than those in regions that had expanded Medicaid.

Prior studies demonstrated that Medicaid expansion is associated with declines in out-of-pocket spending.³ Similar to overall underinsurance results, our criterion-specific findings with regards to out-of-pocket healthcare expenses exceeding 5% of family income were not statistically significant either. Within the context of prior research, this reveals that lower out-of-pocket amounts had not materialized into meaningful financial protections and could still be high relative to income levels of many low-income Americans.

A number of studies have previously evaluated the impact of Medicaid expansions on affordability or cost-related unmet need measures, noting favorable effects of the intervention.⁵⁴ Nevertheless, it is important to mention that avoiding or delaying care (ie, unmet need) is statistically uncorrelated with out-of-pocket expenditures as the percent of income.¹⁴ Furthermore, a sizeable proportion of underinsured individuals tend to be classified as such due to a combination of these reasons.⁷ Therefore, it is important to explore the impact of the ACA beyond single measures and to consider multiple aspects (ie, ex-ante and ex-post) together to uncover underinsurance trends in the population.

As research on underinsurance continues to evolve, there are 2 potential avenues for future investigation. Firstly, researchers should examine the heterogeneity in the effects of Medicaid expansion on underinsurance. The administration of Medicaid at the state level introduces significant variation in program design and benefit structures. In addition to eligibility criteria, Medicaid programs differ in terms of covered services (eg, prescription drugs, mental health services, and long-term care options) and reimbursement rates for healthcare providers. Moreover, administrative structures and processes differ, affecting the efficiency and effectiveness of program implementation. These variations in design and benefit structures have the potential to influence underinsurance outcomes. Secondly, researchers should explore the spillover effects of Medicaid expansion to other income populations. While Medicaid expansion primarily targeted low-income populations below 138% FPL, it may have had implications for underinsurance rates among individuals with incomes above this threshold. By examining the impact on a broader range of income groups, researchers can gain insights into the overall effects of Medicaid expansion on underinsurance and the potential reach of its benefits.

Conclusion

Uninsurance has been the center of the U.S. healthcare policy discussions for a while, but underinsurance is certainly a growing problem and policy makers should consider potential ways to address the issue. In this regard, our study found no evidence of significant changes in underinsurance probability due to Medicaid expansions during the first 4 years of the ACA. However, supplementary analysis suggests that there might be a time lag between Medicaid expansion and its effect on underinsurance. If the longer-term results of this study are confirmed by more research on the subject, then further extension of Medicaid eligibility, especially in 12 non-expansion states, could be a plausible solution to the problem. Having said so, it is important to understand that expansion of coverage alone may not be sufficient to protect millions of Americans, particularly those with low incomes, from underinsurance. It is, therefore, crucial for policy makers to build legislative frameworks that protect individuals from excessive healthcare expenses and prevent treatment avoidance or delay.

Supplemental Material

sj-docx-1-inq-10.1177_00469580231202640 – Supplemental material for Can Medicaid be a Solution to the Problem? Underinsurance in Medicaid Expansion Versus Non-Expansion States

Supplemental material, sj-docx-1-inq-10.1177_00469580231202640 for Can Medicaid be a Solution to the Problem? Underinsurance in Medicaid Expansion Versus Non-Expansion States by Aniyar Izguttinov and Justin G. Trogdon in INQUIRY: The Journal of Health Care Organization, Provision, and Financing

Supplemental Material

sj-docx-2-inq-10.1177_00469580231202640 – Supplemental material for Can Medicaid be a Solution to the Problem? Underinsurance in Medicaid Expansion Versus Non-Expansion States

Supplemental material, sj-docx-2-inq-10.1177_00469580231202640 for Can Medicaid be a Solution to the Problem? Underinsurance in Medicaid Expansion Versus Non-Expansion States by Aniyar Izguttinov and Justin G. Trogdon in INQUIRY: The Journal of Health Care Organization, Provision, and Financing

Footnotes

Acknowledgements

We would like to thank Mark Holmes and Sean Sylvia at the University of North Carolina for providing feedback on an earlier version of the manuscript. We also express our gratitude to Ray Kuntz and his colleagues at AHRQ for their assistance in obtaining access to the restricted data files.

Disclaimer

The research in this paper was conducted at the CFACT Data Center. The results and conclusions in this paper are those of the authors and do not indicate concurrence by AHRQ or the Department of Health and Human Services.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Research Ethics and Patient Consent Statement

The study protocol (#22-0380) was reviewed by the Office of Human Research Ethics at the University of North Carolina at Chapel Hill, which determined that the study does not constitute human subjects research as defined under federal regulations [45 CFR 46.102 (e or l) and 21 CFR 56.102(c)(e)(l)] and does not require IRB approval.

ORCID iD

Aniyar Izguttinov

Supplemental Material

Supplemental material for this article is available online.

References

Rosenbaum

Westmoreland

TM.

The Supreme Court’s surprising decision on the Medicaid expansion: how will the federal government and states proceed?

Health Aff. 2012;31(8):1663-1672. doi:10.1377/hlthaff.2012.0766

Kaiser Family Foundation. Status of State Action on the Medicaid Expansion Decision. November 9, 2022. Accessed December 14, 2022. https://www.kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act

Guth

Garfield

Rudowitz

The Effects of Medicaid Expansion Under the ACA: Updated Findings From a Literature Review. KFF; 2020.

Frean

Gruber

Sommers

BD.

Premium subsidies, the mandate, and Medicaid expansion: coverage effects of the Affordable Care Act. J Health Econ. 2017;53:72-86. doi:10.1016/j.jhealeco.2017.02.004

Lavarreda

Brown

Bolduc

CD.

Underinsurance in the United States: an interaction of costs to consumers, benefit design, and access to care. Annu Rev Public Health. 2011;32:471-482. doi:10.1146/annurev.publhealth.012809.103655

Blewett

Ward

Beebe

TJ.

How much health insurance is enough? Revisiting the concept of underinsurance. Med Care Res Rev. 2006;63(6):663-700. doi:10.1177/1077558706293634

Magge

Cabral

Kazis

Sommers

BD.

Prevalence and predictors of underinsurance among low-income adults. J Gen Intern Med. 2013;28(9):1136-1142. doi:10.1007/s11606-013-2354-z

Schoen

Doty

Robertson

Collins

SR.

Affordable Care Act reforms could reduce the number of underinsured US adults by 70 percent. Health Aff. 2011;30(9):1762-1771. doi:10.1377/hlthaff.2011.0335

Schoen

Doty

Collins

Holmgren

AL.

Insured but not protected: how many adults are underinsured?

Health Aff. 2005;24:W5-289. doi:10.1377/hlthaff.w5.289.

10.

Collins

Bhupal

Doty

MM.

Health insurance coverage eight years after the ACA: fewer uninsured Americans and shorter coverage gaps, but more underinsured. Commonw Fund. 2019. doi:10.26099/penv-q932

11.

Farley

PJ.

Who are the underinsured?

Milbank Mem Fund Q Health Soc. 1985;63(3):476-503. doi:10.2307/3349844

12.

Abraham

DeLeire

Royalty

AB.

Moral Hazard Matters: measuring relative rates of underinsurance using threshold measures. Health Serv Res. 2010;45(3):806-824.

13.

Voorhees

Fernald

Emsermann

, et al. Underinsurance in primary care: a report from the State Networks of Colorado Ambulatory Practices and Partners (SNOCAP). J Am Board Fam Med. 2008;21(4):309-316. doi:10.3122/jabfm.2008.04.080001

14.

Kielb

Rhyan

Lee

JA.

Comparing health care financial burden with an alternative measure of unaffordability. Inquiry. 2017;54:46958017732960. doi:10.1177/0046958017732960

15.

Kaiser Family Foundation. 2019 Employer Health Benefits Survey. September 25, 2019. Accessed January 24, 2023. https://www.kff.org/health-costs/report/2019-employer-health-benefits-survey/

16.

Johnson

Palakshappa

Basu

Seligman

Berkowitz

SA.

Examining the bidirectional relationship between food insecurity and healthcare spending. Health Serv Res. 2021;56(5):864-873. doi:10.1111/1475-6773.13641

17.

Miller

Wherry

LR.

Health and access to care during the first 2 years of the ACA Medicaid expansions. New Engl J Med. 2017;376(10):947-956. doi:10.1056/NEJMsa1612890

18.

Sommers

Blendon

Orav

Epstein

AM.

Changes in utilization and health among low-income adults after Medicaid expansion or expanded private insurance. JAMA Intern Med. 2016;176(10):1501-1509. doi:10.1001/jamainternmed.2016.4419

19.

DeLeire

Dague

Leininger

Voskuil

Friedsam

Wisconsin experience indicates that expanding public insurance to low-income childless adults has health care impacts. Health Aff. 2013;32(6):1037-1045. doi:10.1377/hlthaff.2012.1026

20.

Ghosh

Simon

Sommers

BD.

The effect of health insurance on prescription drug use among Low-Income adults:evidence from recent Medicaid expansions. J Health Econ. 2019;63:64-80. doi:10.1016/j.jhealeco.2018.11.002

21.

Lieberthal

RD.

What Is Health Insurance (Good) For? An Examination of Who Gets It, Who Pays for It, and How to Improve It; 2016.

22.

Pauly

MV.

The Economics of Moral Hazard: Comment. Am Econ Rev. 1968;58(3):531-537.

23.

Fertig

Carlin

Ode

Long

SK.

Evidence of pent-up demand for care after Medicaid expansion. Med Care Res Rev. 2018;75(4):516-524. doi:10.1177/1077558717697014

24.

Morrisey

MA.

Health Insurance. Health Administration Press; 2020.

25.

Domino

Martin

Wiley-Exley

, et al. Increasing time costs and copayments for prescription drugs: an analysis of policy changes in a complex environment. Health Serv Res. 2011;46(3):900-919. doi:10.1111/j.1475-6773.2010.01237.x

26.

Snyder

Rudowitz

. Premiums and Cost-Sharing in Medicaid: A Review of Research Findings. KFF; 2013. Accessed January 24, 2022. https://www.kff.org/wp-content/uploads/2013/02/8417-premiums-and-cost-sharing-in-medicaid.pdf

27.

Manning

Newhouse

Duan

Keeler

Leibowitz

Marquis

MS.

Health insurance and the demand for medical care: evidence from a randomized experiment. Am Econ Rev. 1987;77(3):251-277.

28.

Admon

Valley

Ayanian

Iwashyna

Cooke

Tipirneni

Trends in hospital utilization after medicaid expansion. Med Care. 2019;57(4):312-317. doi:10.1097/MLR.0000000000001082

29.

Brooks

Roygardner

Artiga

Pham

Dolan

Medicaidand CHIP Eligibility, Enrollment, and Cost Sharing Policies as of January 2020: Findings From a 50-State Survey. KFF; 2020.

30.

Cutler

Gruber

Does Public Insurance Crowd out Private Insurance?

Q J Econ. 1996;111(2):391-430. doi:10.2307/2946683

31.

Sharma

Effect of Medicaid expansion on health insurance for Low-Income Nursing Home Aides. J Appl Gerontol. 2023;42(2):231-240. doi:10.1177/07334648221132121

32.

Woronkowicz

Soni

Freedman

Simon

How have recent health insurance expansions affected coverage among artist occupations in the usa?

J Cult Econ. 2020;44:117-154. doi:10.1007/s10824-019-09352-5

33.

Courtemanche

Marton

Ukert

Yelowitz

Zapata

. Early impacts of the affordable care act on health insurance coverage in Medicaid expansion and non-expansion states. J Policy Anal Manag. 2017;36(1):178-210. doi:10.1002/pam.21961

34.

Rae

Copeland

Cox

. Tracking the rise in premium contributions and cost-sharing for families with large employer coverage: Peterson-KFF Health System Tracker. August 14, 2019. Accessed January 24, 2022. https://www.healthsystemtracker.org/brief/tracking-the-rise-in-premium-contributions-and-cost-sharing-for-families-with-large-employer-coverage/

35.

Collins

Radley

Baumgartner

JC.

Higher Costs for Workers and Their Families. Commonwealth Fund; 2019.

36.

Gotanda

Jha

Kominski

Tsugawa

Out-of-pocket spending and financial burden among low income adults after Medicaid expansions in the United States: quasi-experimental difference-in-difference study. BMJ. 2020;368:m40. doi:10.1136/bmj.m40

37.

Goldman

Sommers

BD.

Among Low-Income adults enrolled in Medicaid, churning decreased after the Affordable Care Act. Health Aff. 2020;39(1):85-93. doi:10.1377/hlthaff.2019.00378

38.

Gwet

Machlin

SR.

STATISTICAL BRIEF #507: Out-of-Pocket Health Care Expenses for Non-Elderly Families by Income and Family Structure, 2015. January 2018. Accessed January 23, 2022. https://meps-ahrq-gov.libproxy.lib.unc.edu/data_files/publications/st507/stat507.shtml

39.

Agency for Healthcare Research and Quality. MEPS HC-209: 2018 Full Year Consolidated Data File. August 2020. Accessed January 26, 2023. https://meps.ahrq.gov/data_stats/download_data/pufs/h209/h209doc.shtml

40.

Callaway

Sant’Anna

PHC

. Difference-in-Differences with multiple time periods. J Econom. 2021;225:200-230. doi:10.1016/j.jeconom.2020.12.001

41.

Wing

Simon

Bello-Gomez

RA.

Designing difference in difference studies: best practices for public health policy research. Annu Rev Public Health. 2018;39:453-469. doi:10.1146/annurev-publhealth-040617-013507

42.

Agency for Healthcare Research and Quality. Using Appropriate Price Indices for Analysis of Healthcare Expenditures or Income Across Multiple Years. June 2021. Accessed January 8, 2022. https://meps.ahrq.gov/about_meps/Price_Index.shtml

43.

Lyu

Shane

Wehby

GL.

Effects of the recent medicaid expansions on dental preventive services and treatments. Med Care. 2020;58(8):749-755. doi:10.1097/MLR.0000000000001344

44.

Cubanski

Swoope

Boccuti

Jacobs

Casillas

Griffin

. A Primer on Medicare: Key Facts About the Medicare Program and the People it Covers. 2015. Accessed January 24, 2022. https://www.kff.org/medicare/report/a-primer-on-medicare-key-facts-about-the-medicare-program-and-the-people-it-covers/

45.

Blavin

Karpman

Kenney

Sommers

BD.

Medicaid versus marketplace coverage for near-poor adults: effects on out-of-pocket spending and coverage. Health Aff. 2018;37(2):299-307. doi:10.1377/hlthaff.2017.1166

46.

Tolbert

Drake

Damico

. Key Facts about the Uninsured Population. KFF; 2022. Accessed June 22, 2023. https://www.kff.org/uninsured/issue-brief/key-facts-about-the-uninsured-population/

47.

Collins

Gunja

Aboulafia

. U.S. Health insurance coverage in 2020: a looming crisis in affordability — findings from the Commonwealth Fund Biennial Health Insurance Survey, 2020. Commonw Fund. 2020. doi:10.26099/6aj3-n655

48.

Kaiser Family Foundation. Population Distribution by Age. 2021. Accessed July 28, 2022. https://www.kff.org/other/state-indicator/distribution-by-age/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D

49.

Orgera

Rudowitz

Damico

. A Closer Look at the Remaining Uninsured Population Eligible for Medicaid and CHIP. November 18, 2021. Accessed January 23, 2023. https://www.kff.org/uninsured/issue-brief/a-closer-look-at-the-remaining-uninsured-population-eligible-for-medicaid-and-chip/

50.

Hall

McCue

MJ.

Short-term health insurance and the ACA market. Commonw Fund. 2022. doi:10.26099/x1my-ag25

51.

Kaiser Family Foundation. ACA Open Enrollment: For Consumers Considering Short-Term Policies. October 25, 2019. Accessed June 22, 2023. https://www.kff.org/health-reform/fact-sheet/aca-open-enrollment-for-consumers-considering-short-term-policies/

52.

Kaestner

Garrett

Chen

Gangopadhyaya

Fleming

Effects of ACA medicaid expansions on health insurance coverage and labor supply. J Policy Anal Manag. 2017;36(3):608–42-42. doi:10.1002/pam.21993

53.

Kaiser Family Foundation. Explaining Health Care Reform: Questions About Health Insurance Subsidies. October 29, 2021. Accessed November 30, 2021. https://www.kff.org/health-reform/issue-brief/explaining-health-care-reform-questions-about-health-insurance-subsidies/

54.

Kominski

Nonzee

Sorensen

The Affordable Care Act’s impacts on access to insurance and health care for Low-Income populations. Annu Rev Public Health. 2017;38:489-505. doi:10.1146/annurev-publhealth-031816-044555

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.28 MB

0.06 MB