Effects of Enrollment in Medicaid Versus the State Children’s Health Insurance Program on Kindergarten Children’s Untreated Dental Caries

Abstract

Objectives. We compared levels of untreated dental caries in children enrolled in public insurance programs with those in nonenrolled children to determine the impact of public dental insurance and the type of plan (Medicaid vs State Children’s Health Insurance Program [SCHIP]) on untreated dental caries in children.

Methods. Dental health outcomes were obtained through a calibrated oral screening of kindergarten children (enrolled in the 2000–2001 school year). We obtained eligibility and claims data for children enrolled in Medicaid and SCHIP who were eligible for dental services during 1999 to 2000. We developed logistic regression models to compare children’s likelihood and extent of untreated dental caries according to enrollment.

Results. Children enrolled in Medicaid or SCHIP were 1.7 times (95% confidence interval [CI] = 1.65, 1.77) more likely to have untreated dental caries than were nonenrolled children. SCHIP-enrolled children were significantly less likely to have untreated dental caries than were Medicaid-enrolled children (odds ratio [OR]=0.74; 95% CI=0.67, 0.82). According to a 2-part regression model, children enrolled in Medicaid or SCHIP have 17% more untreated dental caries than do nonenrolled children, whereas those in SCHIP had 16% fewer untreated dental caries than did those in Medicaid.

Conclusions. Untreated tooth decay continues to be a significant problem for children with public insurance coverage. Children who participated in a separate SCHIP program had fewer untreated dental caries than did children enrolled in Medicaid.

The State Children’s Health Insurance Program (SCHIP), created by Congress in 1997, expanded eligibility for public dental insurance to children of the working poor and has grown to include more than 5 million children. SCHIP has provided states with the flexibility to experiment with new health care delivery models that may overcome long-standing obstacles to low-income populations obtaining dental care.1 Currently, 18 states operate separate SCHIP programs, 12 offer Medicaid expansions, and 20 have combination programs.2 Furthermore, the implementation of SCHIP had a spillover effect on Medicaid in some states that led to simplification of their enrollment processes and thus increased enrollment in Medicaid.3 Information about the effects of SCHIP on children’s access to dental services is just beginning to emerge.4^,5

There have been only 3 studies, all using data from the National Health Interview Survey (NHIS), that have considered the impact of SCHIP at the national level. Wang et al. found that for children with low incomes, those with either Medicaid or SCHIP were less likely to have unmet needs for dental care by 8% and more likely to have had a dental visit within the last 12 months by 23% than were those who were uninsured.6 Davidoff et al. found that SCHIP expansions increased the probability of a dental visit among children with chronic conditions by 4.5% and decreased unmet treatment needs by 7.4%.7

The third study, by Duderstadt et al., found that children whose family incomes were consistent with SCHIP eligibility and who were insured for a full year visited the dentist about as often as did children with private insurance.8 State-specific studies of SCHIP’s effects have found favorable results for a number of self-reported indicators, including usual source of dental care, any number of dentist visits, timeliness in obtaining care, and unmet treatment needs.4^,7^,9

In North Carolina, SCHIP is a separate program administered by Blue Cross and Blue Shield of North Carolina (BCBSNC). At the time of this study, SCHIP in North Carolina reimbursed dental providers at rates comparable to those of private insurance, whereas Medicaid reimbursed at rates of 44% to 62% of usual fees. Providers submitted Medicaid claims and negotiated settlements through the Department of Medical Assistance, a governmental agency, whereas, for their patients with SCHIP (with the nonprofit company BCBSNC), providers submitted and negotiated claims just as they did for their privately insured patients. Medicaid and SCHIP provided a similar set of dental benefits, including preventive, diagnostic, and restorative services.

The implementation of SCHIP in North Carolina appears to have improved access to dental care for children with low incomes. On the basis of caregivers’ reports, the number of school-aged children who received dental services increased from 47% in the year before enrollment to 64% in the year after enrollment, and perceived unmet dental needs decreased from 43% to 18%, respectively.10 In an analysis of reimbursement claims, Brickhouse et al. documented a 20% greater use of dental services among pre-school-aged children enrolled in SCHIP than among those enrolled in Medicaid.11

With this investigation, we extend our ongoing evaluation of the effects of enrollment in public insurance on North Carolina children and are the first, to our knowledge, to directly examine the impact of SCHIP on clinically determined tooth decay and compare it to that of Medicaid. Health status is an important indicator of the effectiveness of policies to improve access to care.12 We sought to answer 2 questions. First, is the number of untreated dental caries different among children enrolled in public insurance plans from those not enrolled? Second, what is the impact of the type of public insurance plan (Medicaid vs SCHIP) on the number of untreated dental caries?

METHODS

Study Design and Data Sources

We used a retrospective cohort design to compare the prevalence of untreated dental caries among students enrolled in kindergarten from September 2000 through May 2001 who had been enrolled in public insurance the previous year (October 1999 to September 2000) with those who had not been enrolled during that period. The main exposure variable was program enrollment (nonenrolled, SCHIP, or Medicaid) obtained from Medicaid and SCHIP enrollment and reimbursement claims files. The outcome variable for untreated dental caries was the number of decayed primary teeth (i.e., baby teeth) for each child; we obtained this from the North Carolina Surveillance of Dental Caries system (NCSoDC).

The Medicaid and SCHIP enrollment files provided the enrollment status for each child. These files also provided demographic information about each enrolled child (date of birth, gender, race, and county of residence) and enrollment status for every month during the 12 months. Duration of enrollment was controlled in the analysis by the use of a variable consisting of the number of months enrolled in each plan during the 12-month study period. The Medicaid and SCHIP dental claims files (1999–2000) contained all paid dental claims for an enrolled child and provided a usable measure of dental services utilization.

The NCSoDC system provides annual public health surveillance of dental caries of almost all kindergarten students in the state. Oral health professionals who receive annual training and standardization collect information from open-mouth dental screenings.13^,14 The reliability of these professionals compared with standard examiners was determined in a separate study to be good to excellent (mean κ = 0.86).15 The NCSoDC indicators used for our study were the numbers of decayed, missing, and filled primary teeth, which together represent the lifetime caries experience for a child. We included only molars in the treated-by-extraction category because anterior teeth may have been exfoliated naturally and their inclusion would inflate this estimate of caries experience. We used the variable of missing and filled teeth to control for a child’s lifetime experience of dental treatment, which includes the child’s expressed demand for dental care. The number of dental caries for each child was considered separately as the primary outcome variable and represented untreated dental caries.

We manually matched kindergarten students in the NCSoDC with the Medicaid and SCHIP enrollment or claims files, using a computer program developed to match the child’s first and last names, age, and county. The resulting data set was supplemented with county-level provider characteristics created from the Medicaid and SCHIP dental claims files along with the metropolitan status (metropolitan vs rural) of the county in which the child lived.

Analysis Strategy

We quantified summary statistics and comparisons among exposure groups as predicted probabilities of dental caries. We developed multivariate regression models to compute adjusted estimates of the relation between a child’s public insurance enrollment status and untreated dental caries. The first analysis of effect included the entire sample; the second was limited to those with public insurance.

We used a 2-part regression model (logistic and ordinary least squares) to estimate the differences in actual levels of the outcome (untreated dental caries) and to compare enrolled versus nonenrolled children and SCHIP versus Medicaid.16 The first analysis in the 2-part model used a logistic regression model to predict the probability of a child having any untreated dental caries. The second part of the model used ordinary least squares regression to predict the number of teeth with untreated dental caries, conditional on there being untreated carie present. The models contained the same covariates. The marginal effects of the plans were calculated from the 2-part model with methods proposed by Duan et al.17 Both logistic regression models were adjusted for gender (girl vs boy), race (Black, Hispanic, American Indian, and other vs White), insurance use (dental claims vs no claims), missing and filled teeth, length of enrollment (months), and the ratio of dental providers providing care to children enrolled in Medicaid or SCHIP (5–10 per 1000 and > 10 per 1000 vs < 5 per 1000). We estimated statistical calculations and marginal effects using Stata 9.0.18

RESULTS

Descriptive Statistics

For the sample of children screened (n = 79731), descriptive statistics measuring dental caries are displayed in Table 1 ▶. Approximately 85% of enrolled kindergarten children were screened according to the North Carolina Department of Public Instruction (94350 average daily membership from 2000 to 2001); 30% (n = 23 936) of these children were publicly insured by either SCHIP (10%) or Medicaid (90%). Overall, kindergarten children had a mean of 0.73 (SE = 0.01) untreated dental caries per child. Twenty-three percent had 1 or more dental caries, and these children had a markedly higher mean number of dental caries per child (dental caries = 3.15; SE = 0.02) than did the overall sample.

TABLE 1—

Descriptive Statistics of Dental Caries in Kindergarten Children, by Dental Insurance Enrollment: North Carolina, 2000–2001

				Unadjusted Probability of Dental Cariesa		Adjustedb Probability of Dental Cariesa
	No.	Carie Free,c %	Missing and Filled Teeth,d %	%	Mean (SD)	%	Mean (SD)
Nonenrolled	55 795	67	19	20	0.63 (0.01)
Enrolled	23 936	55	23	30	0.90 (0.01)	30	0.96 (0.01)
SCHIP	2094	58	25	24	0.74 (0.04)	24	0.92 (0.05)
Medicaid	21 842	55	23	31	0.98 (0.01)	36	1.19 (0.05)
All children	79 731	63	20	23	0.73 (0.01)

Note. SCHIP = State Children’s Health Insurance Program.

^aAt least 1 decayed tooth.

^bThe predicted probability of dental caries adjusted for covariates (length of enrollment, utilization, missing or filled teeth, race, and public provider ratio).

^cNo dental caries or missing or filled teeth.

^dAt least 1 missing or filled tooth.

Among children enrolled in public insurance, 30% had untreated dental caries, with an overall mean of 0.90 (SE = 0.01) per child. Medicaid-enrolled children had the highest percentage and largest mean number of untreated dental caries (31%; dental caries = 0.98; SE = 0.01) followed by SCHIP-enrolled children (24%; dental caries = 0.74; SE = 0.04) and nonenrolled children (20%; dental caries = 0.63; SE = 0.01). During the study period, the mean enrollment time in public insurance was 10 months, with 35% having dental claims. Twenty-three percent of enrolled children had 1 or more missing or filled teeth. Additional summary characteristics of the children enrolled in public insurance programs are displayed in Table 2 ▶.

TABLE 2—

Percentages and Means (SE) of Untreated Dental Caries in Kindergarten Children (n = 23 936) Enrolled in SCHIP or Medicaid: North Carolina, 2000–2001

Plan Enrollment	Outcome
SCHIP, % (SE)	8.7 (0.001)
Medicaid, % (SE)	91.3 (0.001)
Outcome variable
Untreated tooth decay,a % (SE)	30.0 (0.003)
Decayed teeth, mean (SE)	0.96 (0.003)
Control variables
Missing and filled teeth,a % (SE)	23.4 (0.003)
Dental claim,a % (SE)	35.3 (0.003)
Length of enrollment, mos, mean (SE)	10.0 (0.02)
Race/Ethnicity, % (SE)
White	41.1 (0.003)
Black	44.9 (0.003)
Hispanic	6.8 (0.002)
American Indian	2.2 (0.001)
Other	5.0 (0.001)
Ratio of public dental care providers, % (SE)
< 5	20.7 (0.003)
5–10	51.2 (0.003)
> 10	28.1 (0.003)

Note. SCHIP=State Children’s Health Insurance Program.

^a Yes or no.

Complete Sample

Kindergarten children enrolled in public insurance programs were 1.71 (95% CI=1.65, 1.77) times more likely to have untreated dental caries than were nonenrolled children, whereas SCHIP-enrolled children were less likely (odds ratio [OR]=0.74; 95% confidence interval [CI]=0.67, 0.82) than were Medicaid-enrolled children to have untreated dental caries (Table 3 ▶). As noted in Table 1 ▶, the base case probabilities of having untreated dental caries for a nonenrolled child and enrolled child were 20% and 30%, respectively. The marginal effects suggest that if a child changed from Medicaid to SCHIP, he or she would have 11% fewer untreated dental caries.

TABLE 3—

Unadjusted Results of the 2-Part Regression Model Predicting Untreated Dental Caries in Kindergarten Children Enrolled in Medicaid or SCHIP: North Carolina, 2000–2001

	Logistic Regression Modela
	b (SE)	OR (95% CI)	OLS model,b b (SE)
Nonenrolled children (Ref)	. . .	1.00	. . .
Enrolled children	0.5339 (0.02)	1.71 (1.65, 1.77)	0.3397 (0.01)c
SCHIP	−0.2960 (0.05)	0.74 (0.67, 0.82)	−0.2449 (0.05)d
Medicaid (Ref)	. . .	1.00	. . .

Note. OR = odds ratios; CI = confidence interval; OLS = ordinary least squares; SCHIP = State Children’s Health Insurance Program.

^aPart 1 of regression model. At least 1 decayed tooth.

^bPart 2 of regression model; results conditional on part 1. Extent of decay for at least 1 tooth.

^cEnrolled children have 17% more untreated dental caries than do nonenrolled children.

^dThe marginal effect of moving a child from Medicaid to SCHIP would be 11% fewer untreated dental caries.

Children Enrolled in Public Insurance

Summary statistics for the predicted probabilities of untreated dental caries in the enrolled population, adjusted for regression model covariates, are also presented in Table 1 ▶. Among children enrolled in public insurance, 30% had untreated dental caries, with an overall mean of 0.96 (SE=0.01) per child. Medicaid-enrolled children had the highest percentage and largest mean number of untreated dental caries (36%; dental caries=1.19; SE=0.05) followed by SCHIP-enrolled children (24%; dental caries=0.92; SE=0.05).

Table 4 ▶ presents the parameter estimates for the final 2-part regression model for children enrolled in 1 of the public insurance programs. Again, children enrolled in SCHIP were significantly less likely to have any untreated dental caries than were those enrolled in Medicaid (OR = 0.73; 95% CI = 0.66, 0.82). The marginal effects suggest that if children changed from Medicaid to SCHIP, they would have 16% fewer untreated dental caries.

TABLE 4—

Adjusted Results of the Final 2-Part Regression Model Predicting Untreated Dental Caries in Kindergarten Children Enrolled in Medicaid or SCHIP: North Carolina, 2000–2001

Variables	Logistic Regression Model,a b (SE)	OLS Model,b b (SE)
Plan
SCHIP	−0.31* (0.05)	−0.27c* (0.05)
Medicaid (Ref)	1.00	1.00
Control variables
Length of enrollment, mos	−0.02* (0.01)	−0.02* (0.004)
Dental claims (yes)	−0.11* (0.03)	−0.01 (0.03)
Missing and filled teeth	0.52* (0.04)	0.19* (0.03)
Race
White (Ref)	1.00	1.00
Black	0.02 (0.03)	−0.06 (0.03)
Hispanic	0.43* (0.06)	0.38* (0.05)
American Indian	0.61* (0.09)	0.52* (0.09)
Other	0.21* (0.07)	0.12 (0.06)
Ratio of public providers
< 5 (Ref)	1.00	1.00
5–10	0.07 (0.04)	−0.01 (0.03)
> 10	−0.16* (0.04)	−0.20* (0.04)
Intercept	−0.82 (0.06)	1.19 (0.05)

Note. OLS = ordinary least squares; SCHIP = State Children’s Health Insurance Program.

^aPart 1 of regression model. At least 1 decayed tooth.

^bPart 2 of regression model; results conditional on part 1. Mean number of dental caries for individuals with at least 1 untreated decayed tooth.

^cThe marginal effect of moving a child from Medicaid to SCHIP would be 16% fewer untreated dental caries.

* P ≤ .001.

DISCUSSION

Our investigation, to our knowledge, is the first to examine the effectiveness of SCHIP in meeting the dental treatment needs of enrolled children. The strengths of this study include its large sample size and linkage of dental health status from a statewide surveillance system with public insurance administrative files containing claims and enrollment data. This approach provided key health status elements that are rarely available in studies of the effectiveness of public insurance at the population level.

The 3 primary findings concern the effect of SCHIP enrollment on the unmet dental treatment needs of North Carolina’s kindergarten children, the burden of untreated dental caries in this population that is associated with being enrolled in Medicaid compared with the SCHIP program, and the effect of public insurance overall on unmet dental treatment needs in children.

The most important finding is the effect on young children’s unmet dental needs of the SCHIP program, which began relatively recently, compared with Medicaid, which has been in existence for almost 40 years. The separate SCHIP program in North Carolina reduces more untreated dental caries than does Medicaid. Marginal effects suggest there would be a 16% improvement if children with sociodemographic characteristics similar to those enrolled in Medicaid were moved into SCHIP (Table 4 ▶).

As of 2005, enrollment penetration is high for Medicaid programs. It is possible that larger gains in access to dental care could be made by improving the effectiveness of the public insurance programs themselves rather than focusing on enrolling more children in those programs. In their national study of SCHIP effectiveness, Wang et al. concluded that improving public insurance programs’ effectiveness regarding access and use of dental services by children already enrolled may be more important than increasing enrollment.6

The comparison of untreated dental caries of children enrolled in public insurance programs with those of children of the same age who were not enrolled provides an assessment of the total effect of public insurance on dental health outcomes. This comparison is important because federal guidelines require Medicaid to provide access to dental care for children enrolled in Medicaid equal to that of other children in their communities.19

Although this effect could not be explored beyond a descriptive analysis because we did not have important information such as insurance status for those not enrolled in Medicaid or SCHIP, we found that children enrolled in public insurance programs have more untreated dental caries than do children who are not enrolled. The prevalence and severity of dental caries were greater in children enrolled in public insurance than in those not enrolled. This difference in extent of untreated disease appears to be of clinical and public health significance because of the large number of children enrolled in these public programs nationwide. Children living in poverty continue to have serious levels of untreated dental caries.20

Limitations

Our study had 2 primary limitations. First, we did not randomly assign study participants to enrollment status, and thus, selection bias or other contributions to lack of equivalence of study groups could affect results. There are barriers beyond dental coverage that affect low-income children’s access to dental care that we did not measure, such as ethnicity, overall health and caregivers’ income, education, access to transportation, and dental health literacy.

Because our design was nonrandomized and because enrollment in Medicaid or SCHIP is voluntary, any observed effects of a health plan on untreated dental caries could have been confounded by self-selection. Poor oral health status may independently increase use of dental services and therefore reduce the amount of untreated disease and bias the effect of insurance coverage. Conversely, parents who are diligent about prevention of dental disease in their children may be more likely to enroll their children in public insurance.

Although various analytic approaches have been developed for dealing with problems of selection bias, these techniques generally require additional measures (e.g., instrumental variables) that can be used to predict enrollment but that do not affect the outcome of interest.21 We were unable to undertake more-formal statistical modeling of self-selection bias, because there were no unambiguous markers that we could have used as instrumental variables. Because we did not have good instrumental variables for the multivariate analysis, we relied on the use of as many control measures as possible to try to mitigate any problems of selection bias.

To quantify any selection bias that may have existed, we used a stratified analysis to compare Medicaid and SCHIP enrollees across 4 strata defined by cross-classifying 2 indicators of dental treatment history. One indicator was use of dental care services during the study period (i.e., ≥ 1claim). The other indicator was missing or filled teeth as measured by the dental screening, which distinguished between children who had evidence of past treatment for dental caries and those who did not. Probabilities of untreated dental caries were calculated using predicted probabilities from the final logistic regression model, which also adjusted for age, race, and availability of dental providers. We estimated that SCHIP produced absolute reductions in the probability of dental caries from 3% (for SCHIP children with utilization who had no missing or filled teeth) to 7% (in those with utilization who had 1 or more missing or filled teeth); this confirmed that our results were not solely caused by selection biases.

A second limitation was that children enrolled in Medicaid or SCHIP may have obtained dental care not reimbursed by Medicaid during the time they were enrolled in public insurance. Children enrolled in Medicaid or SCHIP also may be enrolled in other public programs, such as WIC (Special Supplement Nutrition Program for Women, Infants, and Children) and Early Head Start or Head Start, that have dental components that are not measured.22^,23

Policy Implications

SCHIP has played an important role over the past 10 years in reducing the number of uninsured children in America and providing a crucial safety net for families in a time of declining employer-based dental insurance. Recent studies of SCHIP show increased use of dental services and reduced self-reported unmet dental needs for established enrollees.24 We provided additional evidence that SCHIP also can improve clinically determined oral health status. It is unknown at this time how the Deficit Reduction Act of 2005 will affect children’s access to dental care and what the effects will be of actions taken by Congress during reauthorization legislation of the SCHIP program in 2007.25 Nevertheless, both instances provide opportunities to national and state legislators, and Medicaid programs themselves, to improve the effectiveness of public insurance in low-income children’s gaining access to and using dental care.

Currently, SCHIP provides states with the opportunity to creatively expand health insurance for children in low-income families and experiment with innovative programs. The reimbursement and administrative improvements of SCHIP compared with Medicaid affected the outcome, with SCHIP enrolled children in North Carolina having fewer untreated dental caries than Medicaid enrolled children. The findings highlight the importance of using lessons learned from the effects of a separate SCHIP program in improving access to dental care among low-income children.