Maternal Educational Attainment and Child Health in the United States

Abstract

Purpose:

To identify how child health status differs by mother’s educational attainment for the overall U.S. population and by race/ethnicity, and to assess whether these disparities have changed from 2000 to 2017.

Design

Repeated cross-sectional data from the 2000-2017 National Health Interview Survey (NHIS).

Setting

United States.

Subjects

Children aged 1-17 from a nationally representative sample of the noninstitutionalized U.S. population (N=199,427).

Measures

Reported child health status, mother’s educational attainment, child’s race/ethnicity, and control variables were measured using the NHIS.

Analysis

Using logistic regression models, we assessed the relationship between maternal education and child health. Missing data were imputed.

Results

Children whose mothers had less than a high school education (OR=3.84 95% CI: 3.62, 4.07), high school diploma or equivalent (OR=2.57 95% CI: 2.44-2.70), or some college (OR=1.90 95% CI: 1.80, 2.00) had worse reported health status compared to children whose mothers graduated college. These associations were strongest among non-Hispanic White children, with significantly (p<.05) smaller associations observed for non-Hispanic Black, Mexican Origin, and other Hispanic children. The associations were smaller but persisted with inclusion of controls. From 2000 to 2017, child health disparities slightly narrowed or remained stagnant among White, non-Hispanic Black, and other Hispanic children, but widened for Mexican Origin children (p<.05).

Conclusion

Maternal education disparities in child health are wide and have persisted.

Purpose

Educational disparities in health are widening among U.S. adults,¹ but much is unknown about the magnitude, trends, and variations in disparities in child health. This study identifies the associations between maternal educational attainment and child health status and determines whether these associations have changed from 2000-2017. We use child health status reported by a household adult (usually the mother) because it allows us to identify broad patterns and avoid issues of healthcare access and diagnostic bias connected with common child health conditions. We focus on maternal educational attainment given prior research demonstrating the importance of this measure.^2-3 In line with previous research,⁴ we conduct separate analyses for non-Hispanic White, non-Hispanic Black, Mexican Origin, and other Hispanic children.

Methods

Design

We employ the 2000-2017 National Health Interview Survey (NHIS),⁵ obtained from the Integrated Public Use Microdata Series (IPUMS).⁶ The NHIS is suitable because it has a repeated cross-sectional design, has a very large sample, and provides nationally representative data on the health and family circumstances of U.S. children. We examine children aged 1-17 in the Sample Child files and incorporate information on their family members from the NHIS person files. This public, de-identified dataset does not require IRB approval. See NHIS for information on informed consent processes.⁵

Sample

Our sample includes 199,427 children aged 1-17 who are in the Sample Child Core for 2000-2017 and have valid data for health status. We address item missingness with multiple imputation.⁷

Measures

All data are reported by a knowledgeable adult in the household.⁵ We dichotomize respondent-reported child health status into poor, fair, or good versus very good or excellent health. We categorize mother’s education into less than high school diploma (<HS), high school diploma/GED (HS), some college/associate’s degree, and four-year college degree or higher (BA+; referent). Covariates include survey year, child’s age, child gender, child race/ethnicity, child/parents’ nativity status, region of residence, mother’s age, number of children in the family, family structure, poverty status, homeownership status, mother’s employment status, mother’s health status, and child birthweight. Measurement schemes for all variables are provided in Online Supplement Table A1.

Analysis

We use logistic regression to assess the relationship between maternal educational attainment and child health, adjusting for complex sampling design.⁵ Because significant interaction terms indicate differences in the education-health association by race/ethnicity, we present results from separate racial/ethnic models. We first interact mother’s education with survey year and race/ethnicity to determine whether these temporal patterns differ, and then we interact mother’s education and survey year for each racial/ethnic group. Interaction terms do not show any significant differences in the associations between mother’s education and either child gender or age; thus, we present models including boys and girls of all ages. Section 1 of the Online Supplement contains methodological details.

Results

Descriptive statistics (available in Online Supplement Section 2) demonstrate that children whose mothers have higher educational attainment have generally better health status. However, children with more highly educated mothers also have more advantageous socioeconomic and health circumstances, necessitating multivariable analysis. Thus, Table 1 displays odds ratios and 95% confidence intervals for children’s health status by mother’s education for the full sample (Panel A) and by race/ethnicity (Panels B-E), based on logistic regression models. Overall, child health status is strongly graded by maternal education. The strong relationship in the base model (Model 1) attenuates with each group of covariates, with smaller differences in the full model (Model 4). Among racial/ethnic groups, differences are strongest among non-Hispanic Whites. Compared to White children whose mothers graduated college, White children whose mothers did not graduate high school have nearly five-fold odds of good/fair/poor health (OR=4.93, Model 1, Panel B). For each of the racial/ethnic groups, differences attenuate with the inclusion of covariates, although moderate differences persist in full models.

TABLE 1.

Odds Ratios and 95% Confidence Intervals for Child Health Status (Good/Fair/Poor vs. Very Good/Excellent) by Mother’s Education Compared to the Referent (College Degree or higher), Ages 1-17, U.S. (2000-2017)

	Model 1		Model 2		Model 3		Model 4
	Panel A: All children (N=199,427)
< High school	3.84	3.62,4.07	3.51	3.30,3.73	2.79	2.62,2.98	1.62	1.51,1.74
High school diploma	2.57	2.44,2.70	2.47	2.34,2.60	2.20	2.09,2.32	1.47	1.38,1.57
Some college/ AA	1.90	1.80,2.00	1.82	1.73,1.92	1.72	1.63,1.82	1.24	1.17,1.31
	Panel B: Non-Hispanic White (N=98,010)
< High school	4.93	4.58,5.31	4.27	3.95,4.62	3.33	3.08,3.59	1.75	1.63,1.89
High school diploma	2.94	2.81,3.07	2.71	2.59,2.84	2.42	2.32,2.52	1.59	1.51,1.66
Some college/ AA	2.03	1.94,2.11	1.88	1.80,1.96	1.77	1.70,1.84	1.27	1.21,1.33
	Panel C: Non-Hispanic Black (N=31,504)
< High school	2.85	2.58,3.14	2.66	2.41,2.94	2.09	1.88,2.32	1.48	1.32,1.67
High school diploma	2.08	1.91,2.27	2.07	1.87,2.29	1.80	1.64,1.99	1.42	1.28,1.57
Some college/ AA	1.61	1.50,1.73	1.60	1.49,1.72	1.49	1.39,1.60	1.23	1.12,1.34
	Panel D: Mexican origin (N=37,616)
< High school	2.89	2.66,3.15	2.70	2.48,2.95	2.22	2.03,2.42	1.38	1.21,1.57
High school diploma	1.96	1.76,2.18	1.91	1.73,2.12	1.72	1.56,1.89	1.21	1.05,1.39
Some college/ AA	1.69	1.50,1.90	1.65	1.46,1.86	1.56	1.39,1.75	1.09	0.93,1.26
	Panel E: Other Hispanic (N=18,667)
< High school	3.37	2.77,4.11	3.01	2.39,3.80	2.42	1.84,3.17	1.61	1.16,2.24
High school diploma	2.37	2.04,2.74	2.25	1.89,2.69	1.97	1.63,2.38	1.55	1.31,1.83
Some college/ AA	1.71	1.51,1.94	1.64	1.42,1.90	1.53	1.30,1.79	1.23	1.00,1.51

SOURCE 2000-2017 NHIS Sample Child Supplement

NOTES Model 1 controls for survey year, age, gender, race/ethnicity (Panel A only), nativity status, and region. Model 2 controls for all variables in Model 1 plus mother’s age, number of children, and family structure. Model 3 controls for all variables in Model 2 plus poverty status, homeownership, and mother’s employment. Model 4 controls for all variables in Model 3 and mother’s reported health status and low birthweight.

Figure 1 depicts predicted probabilities of good/fair/poor health status by mother’s education from 2000-2017 using logistic regression and controlling for age, gender, nativity, survey year, and region. Panel A reveals some maternal education convergence in health status among non-Hispanic White children: health status improved some for non-Hispanic White children whose mothers have not graduated college, but remained similar among children whose mothers have a college degree. Panels B and D demonstrate parallel trends over time by mother’s education among Black and other Hispanic children. Panel C shows that Mexican Origin children whose mothers have college degrees or more and those with less than a high school diploma exhibit steeper declines in the likelihood of worse health status over time compared to children whose mothers have a high school diploma or some college. Section 3 of the Online Supplement provides an extensive set of tables based on sensitivity analyses that take into account different measurement schemes for child health and race/ethnicity, as well as different statistical modeling procedures.

FIGURE 1. Predicted Probabilities of Good/Fair/Poor (Vs. Very Good/Excellent) Health Status Over Time by Race/Ethnicity and Mother’s Education, U.S. Children (2000-2017).

A) Non-Hispanic White (N=98,010)

B) Non-Hispanic Black (N=31,504)

C) Mexican origin (N=37,616)

D) Other Hispanic (N=18,667)

SOURCE 2000-2017 NHIS Sample Child Supplement

NOTES Predicted probabilities calculated from logistic regression models interacting survey year with mother’s education categories, each run separately by racial/ethnic group. Models control for survey year, age categories, gender, nativity status, and region.

Panels A and C use interaction terms between mother’s education and survey year; Panels B and D control for mother’s education and survey year.

Discussion

This study reveals important differences in child health status across levels of maternal education even after considering an array of covariates. The disparities are wide for all racial/ethnic groups considered, largely because of the very high level of very good or excellent health among White children with highly educated mothers.

Across 2000 to 2017, we find some convergence in overall health status by maternal education among White children, with Black and other Hispanic children displaying stable disparities across time. The health of White children in the three highest maternal education groups is not improving, which, along with the generally improving health for Black children, suggests a slow narrowing of Black-White gaps in child health. In contrast, Mexican Origin children whose mothers were in the highest and lowest maternal education groups exhibited improved patterns of health, while those in the middle education groups remained similar.

We identify two limitations. First, this study reports observed associations, but does not provide causal estimates. Longitudinal data for multiple cohorts of children would provide opportunities for more specific causal investigations. Second, child health status is reported by the survey respondent. Through accounting for maternal and family characteristics, we attempted to mitigate but cannot fully eliminate inaccuracies or influence from respondent characteristics.

Poor health in childhood has immediate effects, such as missed school days, and long-term consequences such as compromised health and lower educational attainment in adulthood.⁸ Improving maternal education may be an important intervention for improving child health, and may have beneficial spillover effects, such as improving maternal health or increasing income.³ Also striking were the widely varying levels of relatively poor child health across racial/ethnic groups, demonstrating the centrality of race/ethnicity to the experiences of U.S. children. We therefore implore policymakers to address both socioeconomic and racial/ethnic inequalities to best ensure excellent health for American children of all backgrounds.

SO WHAT?

What is already known on this topic?

Prior research has established that maternal education shapes child well-being and health through multiple mechanisms.

What does this article add?

We uncover large maternal education disparities in child health status that have persisted.

What are the implications for health promotion or practice?

Policies seeking to improve child health should focus on both socioeconomically disadvantaged and racial/ethnic minority children, the latter of whom fare worse than White children regardless of maternal educational levels.