The Association between Medicaid Adult Dental Coverage and Children’s Oral Health

Abstract

While all state Medicaid programs cover children’s dental care, Medicaid-eligible children are more likely to experience tooth decay than children in higher income families. We examined the association between Medicaid adult dental coverage, an optional benefit, and children’s oral health. Adult dental coverage was associated with a significant 5 percentage point reduction in the prevalence of untreated caries among children after Medicaid-enrolled adults had access to coverage for at least one year. Lagged policies were also associated with a reduction in parent-reported fair/poor child oral health. Effects were concentrated among children under age 12. We estimated declines in poor oral health among all racial and ethnic subgroups, though there was some evidence that non-Hispanic black children experienced larger and more persistent effects than non-Hispanic white children. Future assessments of the costs and benefits of offering adult dental coverage may consider potential effects on the children of adult Medicaid enrollees.

Despite considerable progress, tooth decay remains the most common childhood chronic disease.¹ Medicaid-eligible children are more likely to experience tooth decay compared to children in higher-income families, but are less likely to visit the dentist annually (29% vs. 55%).² All state Medicaid programs cover a comprehensive set of preventive and restorative dental services for children under the Early and Periodic Screening, Diagnostic, and Treatment (EPSDT) benefit. While financial barriers are frequently reported as the reason for not receiving needed dental care among both adults and children,³ non-cost barriers may also play an important role in explaining income-based disparities in children’s dental care use.

Children are more likely to have regular dental visits when their parents have dental coverage or a recent dental visit.^4–6 Parental dental coverage may facilitate children’s dental care utilization in several ways. For example, providers may relay information about recommended dental care or dental benefits available to publicly insured children when a parent has a dental visit. Since many general dentists treat both adults and children,⁷ families may cluster their appointments when both parents and children have dental coverage, reducing transportation barriers and requiring less time off from work. Parent dental coverage may also reduce out-of-pocket health care spending,⁸ which could increase available resources for children’s health care needs.

In contrast to children, states are not required to provide any level of Medicaid adult dental benefits. Most states provide emergency services (e.g., tooth extractions for relief of pain and infection), but coverage of preventive and restorative services varies for adults across states, with many states expanding and contracting benefits in recent years.⁹ Studies that leverage state-level policy changes to examine the effects of providing adult dental coverage suggest increased dental visits among adults^5,8,9 and children.⁵ Most estimates of the impact of Medicaid adult dental coverage on adult dental visits range from 9 to 14 percentage points.^5,8,9 One recent study indicates that between 20 and 37% of parents that gain adult dental coverage also take their children to the dentist.⁵

This study examined the association between Medicaid adult dental coverage and exam-based and parent-reported measures of oral health among children in low-income families using state-by-year changes to these policies between 1999 and 2016. While past research has found that Medicaid adult dental coverage increases children’s use of dental care, it is unknown whether these policies improve child oral health. More generally, there is little research at the national level that examines the link between health care policies and objective, exam-based oral health measures among children. Our analysis adds to this limited evidence base.

Study Data and Methods

Data Sources and Outcomes

This study used the 1999–2016 National Health and Nutrition Examination Survey (NHANES) and the 2003, 2007, and 2011/12 waves of the National Survey of Children’s Health (NSCH). The NHANES is a nationally representative repeated cross-sectional survey that combines interviews with standardized physical examinations in mobile examination centers. The NHANES oral health assessment is the only national source of exam-based data on dental conditions.

The NHANES identifies teeth with carious lesions (i.e., coronal caries) using either a surface based exam (SBE) or basic screening exam (BSE), depending on the survey year. The SBE counts all tooth surfaces with a carious lesion, whereas the BSE examines each tooth until at least one tooth with a carious lesion is found, at which point the exam ends. The 1999–2004 and 2011–2016 NHANES used protocols based on the SBE, and in 2005–2010 used protocols based on the BSE. In addition, the examiner type differed between these two periods (licensed dentists in 1999–2004 and 2011–2016, and trained health technologists in 2005–2010). Finally, the minimum age of children included in the assessment varied from one to five years, depending on the survey year. These changes are summarized in Appendix Exhibit S1.¹⁰ Reliability analyses suggest high inter-examiner agreement for each of these different protocols.^11,12

We constructed binary measures of any caries, any untreated caries, and any tooth restorations using the oral health assessment data. Considering the presence of at least one cavity or tooth restoration allowed us to use data from all survey years, including those that used the BSE. The any caries category included both untreated tooth decay and tooth restorations. The untreated caries and tooth restorations categories were mutually exclusive (i.e., if an individual had untreated caries and tooth restorations, they would be classified as having untreated caries). For comparison with existing work, we also examined children’s dental visits, which were available in survey years 1999–2004 and 2011–2016.

The NSCH is a nationally representative survey that includes detailed information on children’s health care access and health status. The survey was conducted by telephone in 2003, 2007, and 2011/12, and by mail and web beginning in 2016. Due to sampling and methodological changes, the earlier and later years of the NSCH are not comparable. We used data from the three earlier waves that spanned nine years and allowed us to leverage a substantial number of state-level changes in Medicaid adult dental coverage policies.

We constructed two binary variables based on parent-reported children’s oral health status (excellent, very good, good, fair, or poor). The first indicator was equal to one for responses of fair or poor and zero otherwise, and the second was equal to one for responses of excellent or very good and zero otherwise.

Approach and Statistical Analysis

This study used a difference-in-differences design that leveraged within-state changes in adult dental coverage over time. Dental coverage policies were obtained from the Kaiser Family Foundation and are described in previous research.^5,9 Following prior studies,^5,8,9 we defined a binary indicator equal to one in states that provided coverage of at least one preventive or restorative service beyond emergency care for adults, and zero otherwise. Nearly all states that were classified as providing dental coverage covered regular cleanings and dental exams. According to our definition, 19 states added, dropped, or both added and dropped adult dental coverage during 1999–2016 (AK, AR, CA, CO, DC, FL, HI, ID, IL, KS, MA, MI, MO, OK, SC, SD, UT, WA, WY) (Exhibit 1).

Exhibit 1. Medicaid adult dental coverage policies, 1999–2016.

Source: Authors’ analysis of Medicaid adult dental coverage policies from the Kaiser Family Foundation and other sources.

Since we anticipated that gaining adult dental coverage would first affect a parent’s use of dental care before changing children’s dental care use and oral health outcomes, we considered a policy indicator equal to one in the first and subsequent years after the addition of adult dental coverage (i.e., “concurrent coverage”) as well as lagged dental coverage policies. Existing evidence provides support for considering lagged coverage since adult dental coverage has been found to have larger effects on children’s dental visits after the policy has been in place for more than one year.⁵ We present results for both the one- and two-year lags of adult dental coverage (i.e., policy indicators equal to one after adult dental coverage was provided for at least one and at least two full years, respectively, and zero otherwise). The appendix provides more detail on the state-level policy changes included in each analysis (Appendix Exhibit S2).¹⁰

We estimated multivariable regression models that controlled for the adult dental coverage indicator, child and family characteristics, time-varying state-level variables, and state and year fixed effects. Child- and household-level controls included male sex, child age in years, race and ethnicity (Hispanic, non-Hispanic black, and non-Hispanic other race vs. non-Hispanic white), citizenship (NHANES only), household size, an indicator for highest adult educational attainment being a high school diploma or equivalent or less education, an indicator for family income under the federal poverty line (FPL), and an indicator for being in the cell phone sample (NSCH only). State indicators accounted for fixed characteristics that could be correlated with both dental coverage policies and outcomes, and year indicators accounted for secular trends in the outcomes over time. Time-varying state-level variables included the Medicaid income eligibility limit for working parents, the state earned income tax credit as a percentage of the federal benefit, the maximum temporary assistance for needy families benefit for a family of four, the unemployment rate, and dentists per capita. Inclusion of these controls accounted for changes in other policies, economic conditions, and dentist supply during our study period. We estimated linear probability models to enable interpretation of our estimates as percentage point effects. All models were weighted and errors were clustered at the state level.

The analytic samples included children ages 1–17 in low-income families with complete outcome and demographic information. Income information was collected differently in the NHANES and NSCH, so we used slightly different thresholds to define the sample in each dataset. The NHANES sample included 17,588 children ages 1–17 with family incomes up to 250% FPL who participated in the oral health assessment. The NSCH analysis sample included 78,721 children with family incomes up to 200% FPL. Nearly all states had adult income eligibility limits below 250% FPL during our entire study period. Appendix Exhibit S3 summarizes sample characteristics for both datasets.¹⁰

We conducted subgroup analyses by child age (1–11 vs. 12–17 years old) and race and ethnicity. Child age groups were selected based on existing evidence that adult dental coverage policies have larger effects on dental visits among children under age 12.⁵ To estimate subgroup effects, we stratified our sample based on each characteristic and estimated models identical to those used in our analysis of all children on each subsample.

We conducted additional analyses to assess the robustness and validity of our results. First, we estimated the association between adult dental coverage and outcomes among children in higher-income families less likely to have a parent that meets Medicaid income-eligibility criteria. Second, our analytical strategy required that outcome trends in states that changed their dental coverage policies would have mirrored those in other states absent any policy changes (“parallel trends”). It was not possible to assess this assumption in the NSCH given there were only three survey waves conducted several years apart. Further, due to the sensitivity of the restricted NHANES data, we were not permitted to conduct a typical event history analysis that assessed single year state differences. Therefore, we used a modified analysis that assessed the difference in linear outcome trends over a number of years in a select set of states that changed their policies later in the study period, as described in more depth in the appendix.¹⁰ Third, we examined the sensitivity of our NHANES analysis to excluding survey years that used the BSE protocol (2005–2010), and also to restricting the sample to children ages five and older to ensure consistency in the sample child age range across years. Finally, we tested the sensitivity of our results to using alternative income thresholds to define the study samples, and estimated logit models instead of linear probability models for all NHANES and NSCH outcomes.

Limitations

Our analysis had important strengths and weaknesses. While both the NHANES and NSCH datasets are nationally representative, the NHANES does not include respondents from each state in every year, and the sample is smaller relative to the NSCH, which is state representative. The NSCH, while having a larger sample size, was only available for 2003, 2007, and 2011/12. However, the NHANES is the only nationally representative, exam-based source of data on oral health conditions for the US population, allowing us to address a gap in the literature on the potential for health policy to improve objective oral health measures. The parent-report of child oral health information available in the NSCH complemented our analysis of exam-based caries measures in the NHANES.

In addition, our analysis focused on children in lower-income families more likely to be affected by Medicaid dental coverage policies rather than children with a parent on Medicaid since parent insurance status was not available in either of our data sources. However, an advantage of considering all children in lower-income families is that estimates are not biased by any potential effects of adult dental coverage availability on a parent’s enrollment decisions. Further, the protocol, examiner type, and minimum age of children included in the NHANES oral health assessment varied across survey years. However, the 1999–2004 and 2011–2016 survey years used similar protocols and licensed dentists to conduct the exam, and results were similar when we restricted the analysis to these years.

Finally, similar to other observational studies, we cannot exclude the possibility that changes to other state policy or population variables contributed to our findings. However, our analysis did not suggest significant divergence in outcome trends in states that did and did not change their policies before these changes occurred. Relatedly, emerging research suggests that difference-in-differences results may be biased when there is substantial variation in the timing when states change their policies.¹³ While we were limited in our ability to assess this possibility in the present analysis, a previous study of adult dental coverage policies suggested that this was an unlikely source of bias for estimates of effects on children’s dental visits.⁵

Results

Main Regression Results

Exhibit 2 presents the main difference-in-differences estimates for the NHANES and NSCH outcomes. While concurrent adult dental coverage policies were not significantly associated with children’s outcomes, children’s oral health appeared to improve in the years following a state’s decision to provide adult dental coverage. We estimated that providing adult dental coverage was associated with percentage point declines of about 3.43 (p<0.05) and 2.98 (p<0.10) in the prevalence of any caries when considering the one- and two-year lagged policies, respectively. These estimates represent reductions of about 7% and 6%, respectively, relative to the average rate of any caries among the sample (52%). These declines were driven by a decrease of about 5 percentage points in any untreated caries that was similar when considering the one- and two-year lags of adult dental coverage. This estimate represents a more than 22% reduction relative to the average rate of any untreated caries (22%). The lagged associations between adult dental coverage and the likelihood that a child had any tooth restorations were positive, about 2 percentage points each, but not statistically significant.

Exhibit 2.

Regression estimates of the association between Medicaid adult dental coverage and oral health among children in low-income families

NHANES 1999–2016
Outcome/policy variable	Concurrent coverage	One-year lag	Two-year lag
Any caries	0.0001	−0.0343**	−0.0298*
Rate = 52%
Any untreated caries	−0.0041	−0.0538***	−0.0477**
Rate = 22%
Any restored teeth	0.0042	0.0195	0.0178
Rate = 30%
NSCH 2003, 2007, 2011/2012
Outcome/policy variable	Concurrent coverage	One-year lag	Two-year lag
Excellent or very good condition of teeth	−0.0162	−0.0149	0.0065
Rate = 57%
Fair or poor condition of teeth	0.0242	0.0189	−0.0118**
Rate = 14%

Source: The 1999–2016 National Health and Nutrition Examination Survey and the 2003, 2007, and 2011/12 National Survey of Children’s Health.

Notes: Abbreviations: NHANES = National Health and Nutrition Examination Survey. NSCH = National Survey of Children’s Health. Estimates are the coefficients on the Medicaid adult dental coverage indicator (concurrent, one-year lag, and two-year lag, as labelled). All regressions controlled for male sex, child age in years, race and ethnicity, household size, an indicator for the highest adult education level being high school/GED or less, an indicator for household income under the federal poverty level (FPL), state and year indicators and time-varying state variables including the temporary assistance for needy families benefit for a family of four, the state earned income tax credit as a percentage of federal, dentists per capita, the unemployment rate, and the Medicaid eligibility limit for working parents.

The NHANES analysis also controlled for child citizenship status, which was not available in the NSCH, and the NSCH analysis included an indicator for being in the cell phone sample. The NHANES sample included all children with incomes at or below 250% FPL who participated in the oral health examination during 1999–2016. The NSCH sample included all children with incomes at or below 200% FPL and with responses to the question about the condition of their teeth. All regressions were weighted and errors were clustered at the state level.

^*p<0.10,

^**p<0.05,

^***p<0.01.

Similar to the NHANES exam-based measures, changes in fair/poor oral health status in the NSCH appeared to occur with a lag after changes to adult dental coverage. However, we only observed a significant effect two or more years after a change to adult dental coverage policies. We estimated that providing adult dental coverage was associated with a significant 1.18 percentage point decline in the likelihood that the condition of a child’s teeth was reported as fair or poor (8% reduction relative to the average rate), and a non-significant increase of 0.65 percentage points in reports of excellent or very good condition when considering the two-year lag of adult dental coverage policies.

Consistent with previous research,⁵ we estimated a positive association between adult dental coverage and children’s past six month dental visits (Appendix Exhibit S4).¹⁰ The estimate for the concurrent policy indicator (3.91 percentage points) was in line with existing estimates, though not statistically significant at conventional levels. We estimated significant increases of about 7.41 (p<0.01) and 6.57 (p<0.05) percentage points when considering the one- and two-year lagged policies, respectively.

Exhibits 3 and 4 present the association between adult dental coverage and untreated caries by age and race/ethnicity, respectively. Similar to the results for the full sample, concurrent coverage was not associated with significant declines in untreated caries among any of the subsamples we considered. Lagged policies were associated with significant decreases among children under age 12, while the corresponding estimates for children ages 12–17 were much smaller in magnitude and not statistically significant (Exhibit 3).

Exhibit 3. Regression estimates of the association between Medicaid adult dental coverage and untreated caries among children in low-income families by child age, NHANES 1999–2016.

Source: The 1999–2016 National Health and Nutrition Examination Survey. Notes: Abbreviations: NHANES = National Health and Nutrition Examination Survey. Estimates are the coefficients on the Medicaid adult dental coverage indicator (concurrent, one-year lag, and two-year lag, as labelled) from regressions stratified by child age group. All regressions controlled for male sex, child age in years, race and ethnicity, household size, citizenship status, an indicator for the highest adult education level being high school/GED or less, an indicator for household income under the federal poverty level (FPL), state and year indicators and time-varying state variables including the temporary assistance for needy families benefit for a family of four, the state earned income tax credit as a percentage of federal, dentists per capita, the unemployment rate, and the Medicaid eligibility limit for working parents. The sample included all children with family incomes at or below 250% FPL who participated in the NHANES oral health examination during 1999–2016. All regressions were weighted and errors were clustered at the state level. * p<0.10, ** p<0.05, *** p<0.01.

Exhibit 4. Regression estimates of the association between Medicaid adult dental coverage and untreated caries among children in low-income families by child race and ethnicity, NHANES 1999–2016.

Source: The 1999–2016 National Health and Nutrition Examination Survey. Notes: Abbreviations: NHANES = National Health and Nutrition Examination Survey. Estimates are the coefficients on the Medicaid adult dental coverage indicator (concurrent, one-year lag, and two-year lag, as labelled) from regressions stratified by child race and ethnicity. Estimates for children from other racial/ethnic groups are not shown given small sample sizes. All regressions controlled for male sex, child age in years, race and ethnicity, household size, citizenship status, an indicator for the highest adult education level being high school/GED or less, an indicator for household income under the federal poverty level (FPL), state and year indicators and time-varying state variables including the temporary assistance for needy families benefit for a family of four, the state earned income tax credit as a percentage of federal, dentists per capita, the unemployment rate, and the Medicaid eligibility limit for working parents. The sample included all children with family incomes at or below 250% FPL who participated in the NHANES oral health examination during 1999–2016. All regressions were weighted and errors were clustered at the state level. * p<0.10, ** p<0.05, *** p<0.01.

Lagged policies were associated with significant declines in untreated caries among all racial and ethnic groups, though the magnitude and patterns of these effects differed (Exhibit 4). Estimates for non-Hispanic black children were largest in magnitude and statistically significant when considering both the one- and two-year lagged policy variables. Estimates for Hispanic children were statistically significant when considering the two-year lag, but not the one-year lag of adult dental coverage, whereas non-Hispanic white children exhibited the opposite pattern. Differences between the estimates for each racial and ethnic group were generally not statistically significant, except for the difference between non-Hispanic black and non-Hispanic white children when considering the two-year lagged policy variable. However, it is important to note that the non-Hispanic white subgroup was the smallest sample we analyzed and included only 3,651 children that completed the NHANES oral health assessment.

Sensitivity and Placebo Analyses

Appendix Exhibit S5 presents placebo estimates of the association between concurrent and lagged adult dental coverage policies and the NHANES and NSCH outcomes among children in higher-income families.¹⁰ The estimates were generally small in magnitude relative to analogous estimates for children in lower-income families shown in Exhibit 2 and also were generally not statistically significant at conventional levels. Our NHANES analysis of differences in outcome trends before adult dental coverage policy changes did not suggest significant differences for any of the outcomes (Appendix Exhibit S6).¹⁰ Results for the NHANES outcomes were generally similar when we excluded the 2005–2010 survey years, and restricted the sample to children ages five and older (Appendix Exhibit S7).¹⁰ Estimates were qualitatively similar but larger in many instances when considering alternative income thresholds of 100% FPL and 150% FPL (Appendix Exhibit S8).¹⁰ Estimates from logit models were also qualitatively similar to our main estimates (Appendix Exhibit S9).¹⁰

Discussion

This study found an association between providing Medicaid adult dental coverage and improvements in children’s oral health after parents had access to coverage for at least one year. These results contribute to the limited evidence base on the impacts of oral health policy at the family level, a gap in the oral health services literature recently identified by AcademyHealth’s Oral Health Research Interest Group.¹⁴ Our findings were consistent with a recent study of children’s dental visits that examined a similar period using comparable methods and National Health Interview Survey (NHIS) data.⁵ The magnitude of the association between concurrent adult dental coverage and children’s dental visits in the present study was in line with analogous estimates from the NHIS, though not statistically significant, which may be because of smaller sample sizes in the NHANES. Also consistent with previous research, we found larger effects on children’s outcomes after adult dental coverage policies had been in effect for at least one full year. These findings suggest that adult dental coverage may take time to influence children’s dental care use and oral health. Though, our estimates for lagged dental coverage policies and children’s dental visits were somewhat larger than those implied by NHIS.

In contrast, research examining the Affordable Care Act’s (ACA’s) Medicaid expansion and children’s dental visits did not find an effect, though the authors note that estimates were imprecise.¹⁵ Studies of the association between the ACA’s Medicaid expansion and adult dental care use generally suggest more modest increases than research examining pre-ACA changes to adult dental coverage policies.^16–18 One study suggested a significant increase in dental visits among childless adults but a decline among parents as a result of the ACA’s expansions.¹⁸

We estimated a significant reduction in untreated caries (5 percentage point decline), but a smaller and non-significant change in tooth restorations (2 percentage point increase) when considering lagged adult dental coverage policies. The fact that our estimate for any untreated caries was larger than for any tooth restorations was expected. A child that visits the dentist and receives a filling would initially be categorized as having untreated caries and then as having tooth restorations after the visit. Conversely, a child that stops visiting the dentist may develop untreated caries whether or not he initially had tooth restorations. While preventing the development of caries is optimal, there are many factors that may influence a child’s caries experience including access to dental care, diet, and health behaviors.¹⁹ Since we were unable to distinguish new from existing restorations, the presence of any tooth restorations could reflect a child’s oral health status several years before the NHANES interview. Having untreated caries, however, represents a current unmet health care need that is often easily addressed in a single visit to the dentist.

Our results for the exam-based oral health measures suggested that effects persisted as estimates for the one- and two-year lags of the policy changes were similar in magnitude. In contrast, the association between adult dental coverage and parent-reported oral health was only statistically significant two or more years after a policy change. While research suggests a significant correlation between parent-reported and exam-based caries measures,²⁰ parent-reported outcomes may be subject to social desirability bias and other inaccuracies.²¹ Gaps in parental awareness and reporting may explain differences in the exam-based and parent-report results. However, the exam-based and self-report measures were derived from separate datasets covering different study years, which could also explain these differences.

The lag between changes in adult dental coverage policies and effects on children’s oral health may also depend on whether a state adds or drops coverage. For example, adding adult dental coverage may result in fairly expeditious treatment of existing untreated caries whereas untreated caries may take time to develop in a child who has lost dental care but had it previously. By this logic, oral health improvements might be expected sooner when a state adds adult dental coverage than oral health declines when a state drops dental coverage. Adding and dropping dental coverage may also have different behavioral effects. For example, parents who visit the dentist regularly may receive advice about their children’s oral health care from their provider, and this knowledge is unlikely to be lost immediately when a state drops dental coverage. On the contrary, some Medicaid-enrolled adults may be unaware of their dental benefit status, and may therefore be slower to respond to gaining dental coverage.²² Our analysis was not able to separately identify the effects of adding and dropping coverage given data limitations; future research with larger state-specific samples is warranted.

The prevalence of untreated caries among children has declined over time, and disparities in poor oral health by income and race and ethnicity have narrowed in recent years.²³ However, rates of untreated caries remain higher for Hispanic and non-Hispanic black children than for non-Hispanic white children, and for children with family incomes below the federal poverty level than for children in higher income families.²³ Our subgroup analysis suggested that adult dental coverage was associated with declines in poor oral health among all racial and ethnic subgroups, though there was some evidence that non-Hispanic black children experienced larger and more persistent effects relative to non-Hispanic white children. Further, as expected, the effects of adult dental coverage were concentrated among children in low-income families, with no evidence of effects among children in higher income families. These findings suggest that adult dental coverage may have greater effects on children with the highest unmet oral health care needs.

Conclusion

This analysis builds on previous research that found a positive association between Medicaid adult dental coverage and children’s dental visits by providing evidence that these gains may translate to a reduction in caries among children in low-income families. Medicaid reimbursement rates and the availability of dental providers that accept Medicaid patients are important determinants of dental care access among Medicaid-enrolled adults^9,17 and children^24,25. While not addressed in the present study, these and other state and local factors may amplify or depress the beneficial effects of Medicaid adult dental coverage we document.

According to a recent American Dental Association report, the total cost of offering extensive dental benefits in the 28 states that do not currently provide them would be $836 million per year, or about $4.64 per adult Medicaid enrollee per month.²⁶ While offering adult dental coverage is unlikely to be a cost-effective means to increase dental visits and improve oral health among children if that were the only objective, the combined benefits for adults and children may make adult dental coverage an attractive investment for states.^5,8,9 Future assessments of the costs and benefits of offering Medicaid adult dental coverage may also consider potential effects on the children of adult enrollees. These effects may be long lasting as research suggests that policies that increase oral health care access at early ages are also associated with better adult oral health.^27,28