Significance

Massachusetts health reform, the model for the ACA, reduced uninsurance and improved access to some types of care for children in Massachusetts. Repealing comparable provisions in the Affordable Care Act may produce modest detriments for children.

Introduction

In the United States, approximately 6 million (8%) of children are uninsured for all or part of a given year (Cohen and Martinez 2015). While fewer children are uninsured than adults, being uninsured in childhood has been associated with increased risk of hospitalization, childhood mortality, and worse adult health outcomes (Shakib et al. 2015; Tom et al. 2013; Hakim and Bye 2001; Johnson and Schoeni 2011; Rosen et al. 2009). Another 14 million (19%) children have insurance, but their insurance does not covered needed services or providers, such as specialty care (Kogan et al. 2010, 8). For children, inadequate insurance has been associated with lower immunization rates and increased risk of serious illness (Allred et al. 2007; Tom et al. 2010; Mosquera et al. 2014).

In the past 10 years, there have been two major adult-oriented health reform laws that could affect children—and their health outcomes. In 2006, Massachusetts passed major health reform legislation, including an individual mandate for adults (who were required to purchase insurance for themselves or face a penalty), Medicaid expansion (i.e., children’s eligibility for the state’s Medicaid-Children’s Health Insurance Program (CHIP) increased from 200 to 300% of the federal poverty level (FPL), and adult eligibility for Medicaid increased to 100% FPL), and minimum “essential benefits” for private insurance (i.e., coverage of basic specialty services, no co-pay or deductible for preventive care visits) (Guidance Regarding Minimum Creditable Coverage (MCC) Certification On and After January 1 et al. 2009). For adults in Massachusetts, these provisions led to significant reductions in uninsurance and improvements in access to care (Okoro et al. 2014; Dhingra et al. 2013; Sommers et al. 2014; Long and Dahlen 2014). Based on this success, the federal government passed the 2010 Patient Protection and Affordable Care Act (ACA) modeled after Massachusetts health reform. Key provisions of adult-oriented health reform—the individual mandate, Medicaid eligibility, and essential benefits—have the potential to affect child health insurance status and access to care, but little is known about the potential impact of those provisions on children.

One way to study the ACA’s potential impact on children is to examine the impact of the preceding Massachusetts health reform on insurance and access to care (Table 1). For example, while the individual mandate in Massachusetts did not require parents to enroll their children in insurance, studies suggest that, when adults acquire health insurance for themselves, they also enroll their children (DeVoe et al. 2015, 2008). Another example is the Medicaid expansion, which could decrease child uninsurance directly (i.e., the newly-eligible children between 200–299% FPL) or indirectly (i.e., with adult Medicaid expansion and the individual mandate, low-income parents may be more likely to enroll previously-eligible children) (Dubay and Kenney 2003; DeVoe et al. 2015; Kenney et al. 2010; Sonier et al. 2013; Sommers and Epstein 2011). Health reform provisions could also increase the likelihood of previously-eligible children receiving care: studies suggest that when low-income adults have access to care, their children are more likely to receive medical care (Gifford et al. 2005; Guendelman et al. 2006; Davidoff et al. 2003). For privately-insured children, the “essential benefits” provisions of both reforms requires insurers cover preventive care without cost-sharing and basic specialty services, such as office visits and hospitalizations (Guidance Regarding Minimum Creditable Coverage (MCC) Certification On and After January 1 et al. 2009). Like prior requirements that insurers cover maternity care, essential benefits were designed to provide a minimum standard of services for all privately-insured children and adults (Grace et al. 2014).

Table 1 Adult-oriented health reform and potential effects on child health insurance and access to care
Full size table

Alternatively, adult-oriented health reform may have a limited—or even detrimental—impact on children, given provisions were designed for adults. For example, efforts to standardize adult Medicaid services could inadvertently restrict children’s access to needed care by omitting child-centered services. Children have a different range of conditions and a greater need for preventive services than adults, so previous federal health insurance reforms, most notably the Children’s Health Insurance Program (CHIP), carved out children’s coverage and benefits from those of adults (McDonough 2011; Goldstein and Rosenbaum 2013).

To date, there is very little evidence about the impact of ACA or Massachusetts health reform on children. As the debate over the future of the ACA continues, understanding the impact of its precursor in Massachusetts is essential to shape further health insurance legislation for children. Like the ACA, MA health reform was a two-step process with immediate implementation of the Medicaid expansion and later implementation of individual mandate and essential benefits. Available studies on children under Massachusetts health reform focus on 1–2 years post-reform, predating the implementation of the individual mandate (2007) and essential benefits (2008) (Kenney et al. 2010; Meara et al. 2014; Miller 2012; Smith and Chien 2014). Studies on the ACA are similarly limited to early years post-reform (Adams et al. 2018; Sommers et al. 2016; Courtemanche et al. 2017). These studies found minimal change in children’s uninsurance and access to care. However, they do not reflect the full potential of such health reform—and thus may underestimate the ACA’s impact. For adults under Massachusetts health reform, studies suggest that the implementation of the individual mandate resulted in further reductions in uninsurance (2–3%) and improved access to care (Long et al. 2012; Sabik and Bradley 2015). Use of services included in essential benefits also increased for adults (Long et al. 2012; Ellimoottil et al. 2014). In addition, with the adult and child Medicaid expansions, low-income children may particularly benefit from the ACA and Massachusetts health reforms, but the effect on them is unknown.

Thus, the aims of the present study were to (1) examine whether Massachusetts health reform was associated with reduced uninsurance and greater access to care for children at 1 and 5 years post-reform and (2) examine whether Massachusetts health reform was associated with changes in uninsurance and access to care for low income children at 1 and 5 years post-reform.

Methods

Study Design

We used a difference-in-differences (DD) study design, a quasi-experimental method that compares differences in trends over time between an intervention and comparison group to approximate the effect of an intervention. The DD approach allows for control of baseline differences between the groups and of secular trends; it is considered one of the best methods to approximate a randomized controlled trial (Dimick and Ryan 2014). Our intervention group comprised families with children in Massachusetts; our comparison group were families with children living in non-Massachusetts states. We considered three different non-Massachusetts choices as comparators and a priori specified the other New England states (Maine, Vermont, New Hampshire, Rhode Island, and Connecticut) as our main comparator because they had stable child Medicaid-CHIP eligibility levels during the study period and are socio-economically similar to those for Massachusetts (Appendix 1) (The Henry et al. 2016; Connaughton and Madsen 2012). Our study period includes one-time point before health reform occurred in Massachusetts (2003) and two points post-reform (2007 and 2011). The 2007 time point represents initial post-reform changes, such as the Medicaid expansion, and the 2011 time point represents Massachusetts health reform as fully-implemented, including the individual mandate (implemented in 2007) and essential benefits (implemented in 2008). This study was granted exemption by the Harvard Medical School and Boston Children’s Hospital Institutional Review Boards.

Data Source

We used parent-reported data on children aged 0–17 years from the 2003, 2007, and 2011–2012 waves of the National Survey of Children’s Health (NSCH). The NSCH is a telephone survey of families with one or more children designed to be representative at the state and national levels. The study population included a total of 5,760 children in the intervention group, Massachusetts, and 28,183 children in the comparison group, other New England states, across the three survey waves. We used pre-specified NSCH survey weights to adjust for possible non-response bias, multiple survey waves, and addition of a cellphone sample in the 2011–2012 wave (Centers for Disease Control and Prevention 2013).

Main Outcomes

Our first outcome was uninsurance levels. We defined a child as uninsured if their parents answered no to the question, “Does [your child] have any kind of health care coverage, including health insurance, prepaid plans such as HMOs, or government plans such as Medicaid?”. We excluded cases missing insurance status, which were less than 0.1% of cases in each survey year.

Our second outcome was access to primary and specialty care. We used two variables for access to primary care: (1) Having a personal doctor and (2) Receiving one or more preventive care visit. We characterized a child as having a personal doctor if their parent answered “Yes, one” or “Yes, more than one” to the question, “Do you have one or more persons you think of as [your child]’s personal doctor or nurses?” We defined a child as receiving one or more preventive care visits if their parent answered one or more to the question, “[During the past 12 months/since his/her birth], how many times did [your child] see a doctor, nurse, or other health care professional for preventive medical care such as a physical exam or well-child check-up?”.

We derived access to specialty care from two NSCH questions. The first question asked about need for specialty care: “[During the past 12 months/Since [your child’s] birth], did you or a doctor think that [he/she] needed to see a specialist?” The second asked about problems accessing specialty care: “How much of a problem, if any, was it to get care from the specialist doctor or doctors?” A child was characterized as having no problems accessing to specialty care if their parent answered “yes” to needing specialty care and “no” to having problems accessing specialty care. Children whose parents answered “no” to needing specialty care were excluded from this measure. As the 2003 wave included specialty care questions only when children had a personal doctor, we limited analyses of specialty care to children with a personal doctor to standardize across survey waves.

To understand how Massachusetts health reform might differentially impact low-income children, we stratified for children previously and newly-eligible for Medicaid-CHIP. For children in Massachusetts, we defined previously-eligible children as those living in families between 0-199% FPL (i.e., pre-reform child Medicaid-CHIP eligibility) and newly-eligible children as those in families between 200–299% FPL (i.e., post-reform child Medicaid-CHIP eligibility). We compared previously and newly-eligible children in Massachusetts to children in the same income brackets in other New England states.

Covariates

We selected covariates that may mediate access to care among insured children and are known to differ between Massachusetts and other New England states (Smith and Chien 2014). In our regression model, we adjusted for the following covariates: child age (0–5, 6–9, 10–13, 14–17 years); gender; race/ethnicity (white, non-white); non-English language at home; and having special health care needs (yes, no). Children with special health care needs were defined by the NSCH screener as a child with “an ongoing health condition for which he/she experiences one or more of the following: (1) need or use of prescription medications; (2) an above routine use of services; (3) need or use of specialized therapies or services; (4) need or use of mental health counseling (5) a functional limitation.”

Statistical Analysis

We assessed for baseline differences in demographic characteristics between children in Massachusetts and the other New England states using the 2003 survey wave and Chi square tests.

Consistent with a DD approach, we first assessed unadjusted pre/post trends within the intervention (Massachusetts) and comparison (other New England states) groups for outcomes of interest. We subtracted the rates of our outcomes of interest in the baseline year 2003 from 2007 and 2011–2012 to calculate pre-post trends for 1 and 5 years post-reform respectively. The base equation was DD = (Massachusettspost-reform−Massachusettspre-reform)−(Comparisonpost-reform−Comparisonpre-reform). For each outcome of interest, we used a multivariate logistic regression model to evaluate the significance of pre-post trends adjusted for covariates. We used a multivariate logistic regression model to implement our difference-in-differences model adjusted for covariates.

For sensitivity analysis, we compared families and children in Massachusetts to two additional non-Massachusetts groups: (1) states with a similar child Medicaid-CHIP expansion from 200 to 300% FPL and no adult expansion (3 states) and (2) all states without a child Medicaid-CHIP expansion during the study time period (27 states, including the other New England states). These additional comparators provided further insight for the impact of Medicaid expansion. We also analyzed access to care for insured children only (partial or full-year insured, full-year insured).

We considered an adjusted p-value of less than 0.05 to be significant. For uninsurance and access to primary care, we had 80% power to detect pre-post differences of ≥ 3% at a p-value of 0.05. For access to specialty care and analyses of low-income children, we had 80% power to detect pre-post differences of ≥ 10% at a p-value of 0.05. Analyses were conducted with Stata-11 (StataCorp, College Station, TX).

Results

Pre-reform, children in Massachusetts and other New England states did not differ in terms of age, sex, uninsurance, or levels of private and public insurance coverage. Children in Massachusetts were significantly more likely to have a special health care need, be non-white, speak a non-English language at home, and have higher family incomes than children in other New England states (Table 2).

Table 2 Demographic characteristics of children in Massachusetts and other New England states before Massachusetts health reform, 2003
Full size table

Uninsurance Levels

In Massachusetts, there was no significant difference in uninsurance at 1 year post-reform (p-for-trend = 0.49), but uninsurance was significantly lower at 5 years post-reform (p-for-trend < 0.001) (Table 3). In other New England states, the proportion of uninsured children was not different at 1 year post-reform (p-for-trend = 0.22), but decreased significantly at 5 years post-reform (p-for-trend < 0.001). Thus, at 1 year post-reform, trends in Massachusetts uninsurance rates were not significantly different from those in other New England states (p-for-DD = 0.27). At 5 years post-reform, health reform was associated with a non-significant trend toward decreased uninsurance in Massachusetts (DD = − 1.1, p-for-DD = 0.05).

Table 3 Uninsurance and access to care for children before and after Massachusetts health reform
Full size table

Access to Care

Having a Personal Doctor

At 1 and 5 years post-reform, having a personal doctor increased significantly for children in Massachusetts (p-for-trend < 0.01) and in other New England states (p-for-trend < 0.001). In the DD model, there was no significant difference in having a personal doctor between Massachusetts and other New England states at 1 or 5 years post-reform (Table 3).

Having One or More Preventive Visits in a Year

For children in Massachusetts, the proportion of families reporting their child received one or more preventive care visits increased at 1 year (p-for-trend < 0.001) and returned to pre-reform levels at 5 years post-reform (p-for-trend = 0.35). For children in other New England states, the proportion reporting one or more preventive care visits increased significantly at both 1 and 5 years post-reform (p-for-trend < 0.001). Given the relative increase in other New England states, at 5 years post-reform, Massachusetts health reform was associated with a statistically significant decrease in access to preventive care (DD = − 3.4, p-for-DD = 0.004).

No Problems Accessing Specialty Care

For children in Massachusetts, the percentage of children having no problems accessing specialty care was not significantly different at 1 year post-reform, but increased significantly at 5 years post-reform (p-for-trend = 0.006). For children in other New England states, the proportion of families reporting no problems accessing specialty care was stable at 1 and 5 years post-reform. Massachusetts health reform was associated with a non-significant trend toward increased access to specialty care at 5 years post-reform (DD = 7.7, p-for-DD = 0.06).

Children Previously Eligible for Medicaid-CHIP (0-199% FPL)

For children eligible for Medicaid-CHIP prior to health reform, uninsurance decreased in both Massachusetts and other New England states at 5 years post-reform (p-for-trend = 0.02 and < 0.001), so Massachusetts health reform was not associated with significant reductions in uninsurance (Table 4a). Massachusetts health reform was associated with a significant improvement in access to a personal doctor (DD = 6.6% (0.4–12.8), p-for-DD = 0.04) at 5 years post-reform. There were no significant changes in access to preventive care or specialty care for previously-eligible children in Massachusetts compared to children in the same income bracket in other New England states.

Table 4 Uninsurance and access to care for children previously and newly-eligible for Medicaid-CHIP before and after Massachusetts health reform
Full size table

Children Newly Eligible for CHIP (200–299% FPL)

For children newly Medicaid-eligible, uninsurance decreased non-significantly in Massachusetts and other New England states, so Massachusetts health reform was not associated with a significant reduction in uninsurance at 1 or 5 years post-reform (Table 4b). Massachusetts health reform was associated with a significant improvement in access to specialty care for newly-eligible children at 1 year post-reform (DD = 18.3 (95% CI 2.5–34.0), p-for-DD = 0.03). There were no significant changes in access to preventive care. At 1 and 5 years post-reform, having a personal doctor was stable for this subpopulation of children in Massachusetts, while it increased significantly in other New England states (p-for-trend < 0.001). Given the relative increase in other New England states, Massachusetts health reform was associated with a statistically significant decrease in having a personal doctor at 5 years post-reform for newly-eligible children (DD = − 7.4% (95% CI − 14.3, − 0.5), p-for-DD = 0.02).

Sensitivity Analysis

Our sensitivity analyses with 2 alternate comparison groups (states with similar child Medicaid-CHIP expansions, all states with no child Medicaid-CHIP expansion) confirmed our main findings (Appendix Tables 7, 8). In all states with no child Medicaid-CHIP expansion, children’s uninsurance decreased significantly, and in the DD model, Massachusetts was associated with a statistically significant increase in uninsurance for children overall at 5 years post-reform (DD = 1.0%, p-for-DD = 0.04).

Our sensitivity analyses of insured children (partial or full-year insurance, full-year insured) found no major differences in pre-post trends or DDs for access to primary care (Appendix Table 9). In addition, for full-year insured children, Massachusetts was associated with increased access to specialty care at 5 years post-reform (DD = 9.5%, p-for-DD = 0.01).

Discussion

This study found that adult-oriented health reform in Massachusetts was associated with a trend toward reduced uninsurance and improved access to specialty care for children overall at 5 years post-reform. For low-income children, health reform was associated with increased access to a personal doctor for children previously Medicaid-eligible and increased access to specialty care for children newly Medicaid-eligible.

The fact that uninsurance rates did not significantly decline after health reform in Massachusetts is likely multifactorial. Massachusetts had a low rate of child uninsurance at a baseline, making it challenging to detect any further decrease related to the Medicaid expansion and the individual mandate of health reform (e.g., “floor” effect). The remaining uninsured may be difficult to reach. The study period also included the common shock of the Great Recession (2008–2009) during which falling family incomes led to more children and adults becoming eligible for, and enrolling in, Medicaid-CHIP nationwide (Rowland 2009). Recession-related increases in Medicaid-CHIP enrollment nationwide may bias later gains under Massachusetts health reform toward the null. Nonetheless, there was a trend toward reduced insurance at 5 years post-reform, which suggests a possible benefit from the adult-only individual mandate. Since the ACA’s individual mandate includes children as well as adults, long-term reductions in uninsurance for children—and thus potential detriments of repeal—are likely greater under the ACA than under Massachusetts health reform.

Similarly, the benefits of adult-oriented health reform for low-income children under the ACA may be greater than that seen in Massachusetts. The increased access to a personal doctor for children previously-Medicaid eligible may reflect indirect benefits of parental coverage on children’s access to care, although Massachusetts health reform did not require parents to insure their children. The ACA requires adults to enroll their children in insurance, so gains may be greater under the ACA. Other states also have lower rates of child Medicaid-CHIP enrollment than Massachusetts did pre-reform, so improvements in access to care and any potential effect of adult insurance on child enrollment may be greater for low-income children nationwide under the ACA (Kenney et al. 2015).

The lack of effect of adult-oriented health reform on access to care for children generally contrasts with significant gains seen in adults’ access to care and the anticipated effects of essential benefits, including removal of cost-sharing for well-child visits, in Massachusetts (Okoro et al. 2014; Dhingra et al. 2013; Sommers et al. 2014; Long and Dahlen 2014). In fact, although access to preventive care was stable in Massachusetts, the comparison states made more significant improvements in preventive care during the study period for children generally and for children newly Medicaid-eligible. Several New England states implemented large Medicaid medical home projects during the study period, which may have contributed to their relative gains in access to care (Appendix Table 5). Other studies using the National Health Interview Survey and Medical Expenditures Panel Survey collaborate this trend, finding that children’s access to preventive care increased in other states from 2000 to 2008 (Miller 2012; Abdus and Selden 2013). Child-specific reforms may be necessary to ensure receipt of recommended preventive care.

Our study had strengths and limitations. It is the first study to use the NSCH to examine Massachusetts health reform, and we examined effects two years further post-reform than prior studies on children. The sample size of low-income children was relatively small, so we had limited power to detect significance of changes in uninsurance and access to care. Our data was potentially limited by declining response rates to the NSCH from 55.0% in 2003 to 46.7% in 2007 to 23.0% in 2011–2012, although we adjusted for possible non-response bias with validated weighting tools (Skalland and Blumberg 2012). While state variability in Medicaid-CHIP eligibility and initiatives as well as demographic differences limits any comparison group, our primary comparison group—other New England states—implemented no changes in Medicaid-CHIP eligibility and experienced similar economic trends from the Great Recession during the study period. These similar economic trends precluded the need to control for economy-level variables, such as parental employment (The Henry et al. 2016; Connaughton and Madsen 2012). Although CHIP was reauthorized in 2009, the main components of the reauthorization, such as outreach grants to increase enrollment, were implemented in late 2010 and 2011. None of the earliest components of the 2010 Affordable Care Act (i.e., dependent coverage expansion for children over age 26) were unlikely to affect children ages 0–17 years old. Our findings are further corroborated by the similar trends using alternative comparison groups in the sensitivity analysis.

Our findings at 1-year post-reform are also corroborated by the earlier post-reform studies. Our 1.1% decrease in uninsurance at 1 year post-reform is similar to the 2-2.8% decrease reported in the two prior difference-in-differences studies on children (1 using the Current Population Survey, 1 National Health Interview Survey), both of which used New England states as the baseline comparison group (Kenney et al. 2010; Miller 2012). For low-income children, the one previous study reported a decrease in uninsurance of 5.2% at 2-years post-reform; our finding of a 1.3% decrease in uninsurance at 1-year post reform fits with this trend (Kenney et al. 2010). The slight differences may be due to samples size, survey administration, and/or data processing (Adams et al. 2015). To our knowledge, our findings represent the first estimates of Massachusetts health reform at 5 years post-reform.

Conclusions

Adult-oriented health reforms may have reduced uninsurance and improved access to some types of care for children in Massachusetts. Repealing comparable provisions in the Affordable Care Act may produce modest detriments in insurance and access to care for children.