Oral Health and Academic Achievement of Children in Low-Income Families

Abstract

Low-income children have higher rates of unmet oral health needs. Prior research suggests that poor oral health is associated with lower academic performance but uses cross-sectional and mostly parent-reported measures. This study examined the association between oral health during the first 5 y of life and subsequent academic achievement for low-income children. Birth certificates of children born in Iowa in 1999–2009 were linked to Medicaid enrollment and dental claims data in 1999–2014 and reading and math standardized school test scores for grades 2 through 11. The following oral health measures were examined: having minor dental treatments (mostly surface fillings), major dental treatments (mostly crowns and pulpotomy) or extractions, and comprehensive dental exams during the first 5 y of life. Regression models were estimated adjusting for sociodemographic factors, early infant health, and school district effects. The sample included 28,859 children and 127,464 child-grade observations. In total, 21%, 12%, and 62% of children had at least 1 minor dental treatment, 1 major treatment or extraction, and 1 comprehensive dental exam in the first 5 y of life, respectively. Children who received a minor dental treatment had higher reading and math scores by 1 percentile (95% CI, 0.09–1.9) and 0.9 percentiles (95% CI, 0.02–1.8), respectively. Children who had a major dental treatment or extraction had lower reading and math scores by 2.4 (95% CI, −3.5 to −1.4) and 1.8 (95% CI, −2.8 to −0.8) percentiles. Children who had a comprehensive oral exam had higher reading and math scores by 0.7 (95% CI, 0.06–1.4) and 1.2 (95% CI, 0.6–1.9) percentiles. The findings suggest that children’s oral health before school age is associated with academic achievement later during school years.

Introduction

Oral health disparities by income emerge early in life in the United States. In 2011–2016, children aged 2 to 5 y below 200% of the federal poverty line had more than twice the rate of untreated dental decay in primary teeth (13.9% vs. 6%) than children in higher-income households (Centers for Disease Control and Prevention [CDC] 2019). Children aged 3 to 5 y with public insurance also have higher rates of dental problems and unmet dental needs than those with private insurance (Shariff and Edelstein 2016). These disparities continue later in childhood in permanent teeth (Shariff and Edelstein 2016; CDC 2019). Tooth decay remains a leading chronic condition among children (CDC 2021).

Poor oral health and unmet dental care needs are risk factors for reduced school performance and lower psychosocial well-being (Guarnizo-Herreno and Wehby 2012). On one hand, oral health problems could result in pain, sleep disruption, poor nutrition, reduced playing time and activity, and worse psychosocial well-being through adverse effects on speaking, smiling, and other social interactions. These changes could in turn reduce attention, learning, and school performance. On the other hand, third variables, including socioeconomic factors, may contribute to this association. Oral health problems have been associated with lower school performance and more absence in the United States (Jackson et al. 2011; Guarnizo-Herreno and Wehby 2012; Seirawan et al. 2012; Guarnizo-Herreno et al. 2019) and other countries (Petridou et al. 1996; Muirhead and Marcenes 2004; Muirhead and Locker 2006; Garg et al. 2012; Paula et al. 2016; Rebelo et al. 2019; Ruff et al. 2019). However, those studies examined contemporaneous measures of oral health and academic outcomes that did not adequately separate oral health and academic outcomes in time. As such, less is known on the association between oral health early in life, before school, and subsequent academic achievement. Furthermore, most studies used parent-reported data for both oral health and academic outcomes, increasing the possibility of reporting bias, measurement error, and confounding. Finally, the literature has largely focused on oral health problems and less on the association between preventive dental care and achievement.

This study examined the association between oral health of low-income children during the first 5 y of life and subsequent academic achievement. The study employed a unique population-based linkage between Medicaid claims data, birth certificates, and standardized school tests and evaluated differences between elementary and higher grades. The study adds new knowledge on the association between early oral health and learning. Early childhood is a critical period for establishing dental care routines and home-based oral hygiene practices but also a time when oral health problems and income disparities emerge. Income disparities in children’s academic achievement are large (Reardon 2013), but whether early oral health is associated with those disparities is not known. The study sheds light on this potential association.

Methods

Data

The data came from 3 sources: 1) the Medicaid Analytical eXtract (MAX) Personal Summary (PS) and Other Services (OT) files, obtained for each year from 1999 through 2014 from the Centers for Medicare & Medicaid Services (CMS) for children born in Iowa between 1999 and 2009 and for mothers enrolled in Medicaid at delivery (Research Data Assistance Center 2021a, 2021b); 2) birth certificates from the Iowa Department of Public Health (IDPH); and 3) standardized school tests from the Iowa Testing Programs (ITP). The annual MAX PS files included information on month-by-month enrollment in Medicaid, an indicator for a claim for inpatient delivery, and basic demographic information (such as birth dates and gender). The annual MAX OT files included claims submitted for outpatient services, including dental services. Birth certificates provided information on infant health and maternal demographic and socioeconomic factors at time of birth.

To merge the 3 data sets, a multistep matching process was used (Fig. 1). The birth certificates and the ITP data were matched via child’s name and birth dates. The birth certificate–ITP matched data set included pseudo-IDs unique to each child. The Medicaid PS and OT files did not have names. Therefore, the Medicaid PS files were matched to birth certificates based on unique combinations of full maternal date of birth, full child’s date of birth, and child’s gender. Birth certificates and Medicaid files were matched based on the following steps. First, newborns were matched to their mothers in the MAX PS files based on the first 9 digits of the unencrypted state case ID, which is shared between mothers and newborns. Mothers were selected from the PS files based on having an inpatient delivery claim in the same year in which the child was born. Therefore, newborns were matched to their mothers only if the mothers were enrolled in Medicaid at time of delivery and if the newborns were enrolled in Medicaid in their birth year. Then, newborns linked to their mothers in the MAX PS files were matched to their birth certificates based on unique combinations of full maternal birth date, full child’s birth date, and child’s gender. Children in the birth certificate–Medicaid matched sample were then matched to the ITP data using the pseudo-ID that links the birth certificates and ITP data sets.

Figure 1.

Sample construction flowchart.

Medicaid PS and OT files assign a unique ID to each beneficiary that does not change over time, which was used to link Medicaid PS and OT for the same child over time. In order to accurately measure dental services use during the first 5 y of life from the Medicaid claims data, the analytical sample was further limited to children enrolled in Medicaid for at least 9 mo in each year of the first 5 y of life. This ensured that the child was enrolled for most of this period and reduced the chance unmeasured dental services (not having a claim) due to being uninsured.

Oral Health Status Measures

Dental problems were measured based on the intensity and severity of dental treatments that children received derived from the Code on Dental Procedures and Nomenclature (CDT) codes in the claims data (OT file) similar to prior studies (Chi et al. 2013; Wehby et al. 2017). Dental treatment codes were categorized into minor treatments, major treatments, and tooth extractions. Minor treatments included primarily surface fillings (over 99% of cases). Major treatments included mostly crowns and pulpotomy (over 99% of cases). Major treatments and extractions were combined in the main model since they both indicate major dental problems but separated in additional models. Other treatment codes (not specified, less severe, and less common codes such as analgesia, hospital call, or space maintainer) were excluded. Binary indicators were coded for any minor treatments, any major treatments, and any extractions during the first 5 y of life (first 60 mo).

The main measure of preventive services use was an indicator for whether the child received a comprehensive dental evaluation (CDT codes D0150 and D0120). Comprehensive exams can identify dental problems for treatment to stop disease and control complications. They can also identify concerns about oral health and home prevention habits, initiate counseling to parents and children about oral health–promoting activities such as teeth brushing and diet, and help in initiating a visit routine. Another reason to focus on comprehensive exams was that the majority of children (over 97%) who received other preventive services, including a prophylaxis, fluoride application, or dental sealants, also received a comprehensive exam. In an alternative model, another indicator was used for receiving both a comprehensive exam and at least 1 other preventive service such as prophylaxis, fluoride application, or dental sealants (with these 3 services representing almost all received preventive services). Similar to treatment measures, a binary indicator was coded for whether the child received any comprehensive exam during the first 5 y of life (or alternatively a comprehensive exam and at least 1 other preventive service).

Birth Certificate Covariates

Several demographic, socioeconomic, and infant health covariates were included from birth certificate data. These were maternal marital status (married or not), highest schooling level (less than high school, high school, and some college or college graduate), age (<20, 20–24, 25–29, and ≥30 y) at the child’s birth, race (White or non-White), and ethnicity (Hispanic or non-Hispanic). Also included were child’s gender, number of prior live births (0, 1, 2, 3, and ≥4), child’ birth weight (in grams) and gestational age (in weeks), and maternal smoking during pregnancy (yes or no).

School Test Scores

The ITP data included scores on standardized school tests administered to students in Iowa, with reading comprehension and math being the most commonly tested areas. The tests included the Iowa Tests of Basic Skills and the Iowa Tests of Educational Development before the 2011–2012 academic year and the Iowa Assessments after that. The tests are widely recognized for their reliability and validity (Hoover et al. 2003) and have been included in research studies (e.g., Wehby et al. 2014, 2015; Gallagher et al. 2017). Scores can be standardized as percentiles on a national sample of scores. Other variables were grade level, month and year of testing, and school district. Because of less frequent testing in elementary grades K, 1, and 12, the analysis focused on tests during grades 2 through 11.

Statistical Analysis

The following regression was estimated:

$$\begin{array}{l}Scor{e}_{it}={\beta }_0+{\beta }_{1}Mino{r}_i+{\beta }_{2}Majo{r}_i\\ +{\beta }_{3}Preventio{n}_i+{\mathbf{X}}_i\lambda +{\mathbf{Z}}_{it}\delta +e_i.\end{array}$$ (1)

Score was a reading or math test score for child i tested in year t. Minor was a 0/1 indicator for whether the child had any minor dental treatments during the first 5 y of life (1 if yes, 0 if no). Major was a 0/1 indicator for receiving any major dental treatments or tooth extractions during the first 5 y (1 if yes, 0 if no). Prevention was a 0/1 indicator for whether the child received at least 1 comprehensive dental exam during those first 5 y (1 if yes, 0 if no). β₁, β₂, and β₃ were estimates of the associations of minor treatments, major treatments, and comprehensive exams with test scores, respectively. X included child and mother characteristics from birth certificates described above in addition to fixed effects (0/1 indicators) for child’s birth year and birth month. Z included fixed effects (0/1 indicators) for child’s grade level, test year, test month, and school district.

The model was first estimated pooling all grades together and then separately for elementary grades (pooling grades 2 through 6) and middle/high school (grades 7 through 11). The regression was estimated using ordinary least squares. Since the model included multiple tests (grades) per child, standard errors were clustered at the child level.

Alternate models were also estimated. First, major treatments and extractions were separated into 2 indicators. Second, the measure of comprehensive exams was replaced by a measure for receiving a comprehensive exam plus at least 1 other preventive service. Third, because some children who did not receive treatments might still have had dental problems that were not treated, the associations of minor and major treatments with test scores were estimated only for children who received at least 1 comprehensive exam. That way, comparing children who received treatments and those who did not only included children with at least some measured access for comprehensive dental care. Finally, another model included separate measures of oral health at ages 1 to 3 and 4 to 5 y to check if associations with test scores varied by age at which dental services were received.

Results

Sample Description

The analytical sample included 28,859 unique children matched across the 3 data sets (Medicaid, birth certificates, and school tests) with 127,464 child-grade observations for the reading score analyses, as well as 28,852 unique children and 126,775 child-grade observations for the math score analyses. Table 1 reports descriptive statistics for the analytical sample. Nineteen percent of children had at least 1 minor dental treatment, and 11% had at least 1 major treatment during the first 5 y of life. Fifty-eight percent had at least 1 comprehensive dental exam in those years. Mean standardized scores were the 46th percentile for reading and 45th percentile for math. Eighty-four percent of mothers were White, and 13% were Hispanic. Thirty-five percent of mothers had less than high school education at birth, and 42% had high school education.

Table 1.

Summary Statistics of the Analytical Sample.

Characteristic	Value
Oral health measures, %
Minor treatment	18.9
Major treatment or extraction	10.9
Comprehensive exam	57.9
School test scores, mean (SD)
Reading comprehension score	46.1 (27.7)
Math score	44.7 (26.6)
Demographic, socioeconomic, and infant health variables
Child’s sex, %
Female	49.1
Male	51.0
Number of prior live births, %
0	34.9
1	30.8
2	19.7
3	9.0
4 or more	5.6
Maternal age at child’s birth (years), %
<20	19.8
20–24	42.8
25–29	23.2
≥30	14.2
Maternal race, %
White	83.6
Non-White	16.4
Maternal Hispanic ethnicity, %
Yes	13.4
No	86.7
Maternal marital status at child’s birth, %
Married	29.1
Not married	70.9
Maternal education at child’s birth, %
Less than high school	34.9
High school	41.6
Some college or college graduate	23.5
Maternal smoking during pregnancy, %
Yes	37.5
No	62.5
Gestational age, mean (SD), wk	38.8 (1.8)
Birth weight, mean (SD), g	3,305 (539)

The descriptive statistics for reading scores, sociodemographic and infant health variables, and oral health measures are based on the analytical sample with data on reading and all model covariates (N = 127,464 child-grade observations; 28,859 unique children). The mean and standard deviation (SD) for math are based on the sample with data on math tests and all model covariates (N = 126,775 child-grade observations; 28,852 unique children). The descriptive statistics for reading and math scores are based on the sample of child-grade observations. The descriptive statistics of the sociodemographic and infant health variables are based on the sample of unique children with at least 1 reading test. Math and reading test scores are in national percentile rankings (child’s test ranking in percentiles on a national distribution of scores). The descriptive statistics are rounded to the first decimal (except for birth weight). Therefore, the sum of the percentages of subgroups of a variable may slightly exceed 100%.

Associations of Oral Health and School Test Scores

Table 2 reports the regression estimates for the associations of the oral health measures during the first 5 y of life with reading and math test scores combining grades 2 through 11 (detailed regression results are in Appendix Table 1). The 3 measures of dental services were simultaneously included in the regression. Children who received a minor dental treatment had higher reading and math scores by 1 percentile (95% confidence interval [CI], 0.09–1.9) and 0.9 percentiles (95% CI, 0.01–1.8), respectively (P < 0.05). In contrast, children who had a major dental treatment or extraction had lower reading and math scores by 2.4 (95% CI, −3.5 to −1.3) and 1.8 (95% CI, −2.8 to −0.7) percentiles, respectively (P < 0.01). Finally, children who had a comprehensive oral exam had higher reading and math scores by 0.7 (95% CI, 0.06–1.4; P < 0.05) and 1.2 percentiles (95% CI, 0.6–1.9; P < 0.01), respectively. The results of the alternate models described above were consistent with the main model and are reported in Appendix B.

Table 2.

Regression Estimates of Associations of Oral Health Measures during the First 5 y of Life and School Test Scores during Grades 2–11.

Oral Health Measure	Reading	Math
Minor treatment	1.0	0.9
	(0.5) {0.032} [0.09–1.9]	(0.4) {0.046} [0.01–1.8]
Major treatment or extraction	−2.4	−1.8
	(0.5) {0.00001} [−3.5 to −1.3]	(0.5) {0.001} [−2.8 to −0.7]
Comprehensive exam	0.7	1.2
	(0.3) {0.034} [0.06–1.4]	(0.3) {0.0003} [0.6–1.9]
N	127,464	126,775

The estimates represent the regression coefficients of the oral health measures with the test score as the outcome. Standard errors are in parentheses. P values are in curly brackets, and 95% confidence intervals are in brackets. N represents the number of child-grade observations in the regression.

Table 3 reports the same estimates separately for elementary grades (2 through 6) and middle/high school grades (7 through 11). Estimates were close between the 2 groups except for minor dental treatment, which was associated with an increase in math scores only during elementary grades by 1.1 percentiles (95% CI, 0.3–2.0; P < 0.05) but had a smaller and statistically nonsignificant estimate at higher grades.

Table 3.

Regression Estimates of Associations of Oral Health Measures during the First 5 y of Life and School Test Scores during Grades 2–6 and 7–11.

Oral Health Measure	Grades 2–6		Grades 7–11
	Reading	Math	Reading	Math
Minor treatment	0.9	1.1	1.1	0.3
	(0.5) {0.044} [0.03–1.8]	(0.4) {0.012} [0.3–2.0]	(0.7) {0.088} [−0.2 to 2.5]	(0.6) {0.653} [−1.0 to 1.6]
Major treatment or  extraction	−2.3	−1.8	−2.5	−1.7
	(0.5) {0.00002} [−3.4 to −1.3]	(0.5) {0.001} [−2.8 to −0.7]	(0.8) {0.002} [−4.1 to −1.0]	(0.8) {0.026} [−3.2 to −0.2]
Comprehensive exam	0.8	1.2	0.7	1.2
	(0.3) {0.027} [0.1–1.4]	(0.3) {0.0003} [0.5–1.8]	(0.6) {0.204} [−0.4 to 1.8]	(0.5) {0.02} [0.2–2.3]
N	96,969	96,338	30,495	30,437

The estimates represent the regression coefficients of the oral health measures with the test score as the outcome. Standard errors are in parentheses. P values are in curly brackets, and 95% confidence intervals are in brackets. N represents the number of child-grade observations in the regression.

Discussion

Using unique linkages between birth certificates, Medicaid enrollment and dental claims for low-income children enrolled in Medicaid in Iowa, and standardized school test scores, the study finds that oral health early in life (before school age) was associated with academic achievement. Specifically, receiving minor dental treatments and comprehensive dental exams during the first 5 y of life was associated with higher reading and math scores combining grades 2 to 11, while receiving major dental treatments was associated with lower scores. Most estimates were similar when separating elementary and middle/high school grades, indicating persistence over age. Because low-income children have higher rates of unmet dental health needs and lower academic performance than higher-income children, the findings suggest that early oral health is associated with income disparities in academic outcomes.

Major treatments are a marker for complex and severe dental problems associated with pain, decay, and periodontal disease that may disrupt children’s attention, sleep, playing, and learning and reduce their psychosocial well-being. Those treatments, received by nearly 11% of the children at least once in the first 5 y of life, had the largest differences in reading and math scores, including declines by 2.4 and 1.8 percentiles, respectively, or 5.2% and 4% of score sample means. In contrast, the higher scores with minor treatments suggest that some children who did not receive such treatments had untreated cavities or other dental problems that might worsen over time. Finally, the positive association of comprehensive exams with achievement, even when adjusting for minor and major treatments, might suggest benefits from prevention but also that not all children with dental problems appear to be receiving the treatments they need. However, the study provides association rather than causal estimates, and elucidating possible mechanisms in future research is important. Furthermore, oral health might interact with other early life health measures and perinatal risk factors associated with achievement such as birth weight and maternal health behaviors (Figlio et al. 2014; von Hinke Kessler Scholder et al. 2014; Kristjansson et al. 2018), which can also be examined in future work.

The findings add to the evidence on the importance of adequate access to dental care early in life and before school age. Medicaid covers dental preventive and treatment services for children without out-of-pocket cost sharing. However, low-income families face other access barriers, including availability of local dentists who participate in Medicaid and provide care for very young children, and time, information, and transportation costs (Rodgers et al. 2008; CMS 2011). Addressing those access barriers should continue to be the focus of policies and programs aimed at improving children’s dental care utilization and oral health. IDPH implements a dental home initiative (I-SMILE) to support families and individuals in accessing dental services by sharing information, connecting families with providers, finding transportation services, and providing dental screening and fluoride in childcare and preschool programs (IDPH 2021). Continuing and bolstering such programs that consider multiple access constraints are needed to reduce access barriers.

Most general dentists in Iowa see few children younger than 3 y and commonly refer them to pediatric dentists concentrated in urban areas (McQuistan et al. 2005; Wolfe et al. 2006). This barrier exacerbates transportation and time costs to families. Moreover, less than half of general dentists report participating in Medicaid, and of those who participate, most limit acceptance to previous patients who transition to Medicaid coverage (Reynolds et al. 2019). States differ in Medicaid reimbursement rates, estimated to range between nearly 31% and 82% (about 41% in Iowa) of private insurance reimbursement for children’s dental services across states with fee-for-service programs in 2016 (Gupta 2017). Some studies suggest that reimbursement rates are associated with dentist participation and children’s access to dental care (Decker 2011; Buchmueller et al. 2015; Nasseh and Vujicic 2015).

The primary study strengths include measuring oral health and preventive dental care use before school rather than simultaneously with academic outcomes, using objective measures based on existing administrative data, and studying a population-based sample. Most prior studies have used survey and parent-reported outcomes that might involve report errors or bias. The study also separates treatments by intensity/severity and includes preventive care, which captures different aspects of oral health, access, and utilization of dental services.

A weakness of using claims data to measure dental (or other) health problems is that children who do not receive treatment are not necessarily problem free and include a mixture of both children without dental problems and those with untreated problems. The sample included children enrolled for at least 9 mo in each of the first 5 y in Medicaid to ensure continued coverage (and reduce the chance of error in measuring dental services use due to being uninsured). Furthermore, findings for minor and major treatments were similar in an alternate sample limited to children who had at least 1 comprehensive exam to ensure minimum access. Moreover, the regression compared children within the same school district who likely faced similar geographic access barriers compared to children in different districts. However, other factors may influence dental care utilization such as parental knowledge about the child’s dental care needs. Another limitation is that the associations might still reflect residual confounding (such as parental investment in children’s health and learning). The regression adjusted for multiple household factors and school district effects, which reduced this threat. However, the included household factors did not capture changes in the household socioeconomic and child health measures over time.

In conclusion, children’s oral health before school age is associated with academic achievement later in life, with lower reading and math test scores in the presence of major dental problems, as well as better scores with minor dental treatments and comprehensive dental exams. The findings support the importance of addressing barriers to dental care and promoting oral health early in life before school age and suggest that the disproportionally high rates of unmet oral health needs among low-income children are associated with income gaps in academic achievement.

Author Contributions

G.L. Wehby, contributed to conception, design, data acquisition, analysis, and interpretation, drafted and critically revised manuscript. The author gave final approval and agrees to be accountable for all aspects of the work.