Abstract
Objective
Individual academic achievement is a well-known predictor of adult health, and addressing education inequities may be critical to reducing health disparities. Disparities in school quality are well documented. However, we lack nationally representative studies evaluating the impact of school quality on adult health. We aim to determine whether high school quality predicts adult health outcomes after controlling for baseline health, socio-demographics and individual academic achievement.
Methods
We analyzed data from 7,037 adolescents who attended one of 77 high schools in the Unites States and were followed into adulthood from the National Longitudinal Study of Adolescent to Adult Health. Selected school-level quality measures—average daily attendance, school promotion rate, parental involvement, and teacher experience—were validated based on ability to predict high school graduation and college attendance. Individual adult health outcomes included self-rated health, diagnosis of depression, and having a measured BMI in the obese range.
Results
Logistic regressions controlling for socio-demographics, baseline health, health insurance, and individual academic performance demonstrated that school quality significantly predicted all health outcomes. As hypothesized, attending a school with lower average daily attendance predicted lower self-rated health (Adjusted Odds Ratio (AOR) 1.59, p=0.003) and higher odds of depression diagnosis (AOR 1.35, p=0.03); and attending a school with higher parent involvement predicted lower odds of obesity (AOR 0.69, p=0.001). However, attending a school with higher promotion rate also predicted lower self-rated health (AOR1.20, p <0.001).
Conclusions
High school quality may be an important, but complex, social determinant of health. These findings highlight the potential inter-dependence of education and health policy.
Keywords: School quality, health, depression, obesity
Introduction
Educational attainment is a powerful predictor of lifelong health.1,2 More years of education are associated with longer life expectancy and lower rates of depression and obesity.3-5 While the causal pathway linking education and health has not been established, studies suggest several mechanisms such as improved health literacy, higher income, more work-related benefits, healthier social networks, and improved social standing. 3,6-8 Given this evidence, addressing inequities in education may be critical to reducing health disparities.9
Although numerous studies describe these associations between education and health at the individual and interpersonal levels, few address associations at the organizational and community levels. Hence it remains unknown whether higher quality schools contribute to healthier populations.10 School quality may influence each of the pathways through which educational attainment is hypothesized to impact health. High school (HS), for instance, might offer a critical opportunity to shape long-term health trajectories as this near-universal exposure to school occurs during adolescence, a sensitive developmental period when many adult health-related behaviors commence.11,12
Few studies investigate whether selected school-level characteristics are associated with adult health. Accumulation of educational advantage, such as attending a HS with a higher proportion of wealthy or white students, has been associated with a lower risk of health-related work limitations later in life.13 Higher per pupil spending during adolescence has predicted better self-rated health in adulthood.14 Finally, improvements in school quality—measured by changes in pupil-teacher ratio, teacher salary, and length of the school year—have been associated with lower levels of adult disability and stronger associations between educational attainment and self-rated health, obesity, and mortality among black students attending Southern schools before and after segregation.15,16
These studies suggest that school quality constitutes an important driver of adult health outcomes and health disparities. However, there are no representative, longitudinal studies to systematically examine aspects of HS quality and long-term health in the context of modern U.S. society.10 The current study uses longitudinal, nationally representative data to determine whether HS quality is associated with adult health after controlling for individual, family, neighborhood, and school contextual factors.
Methods
We analyzed data from the National Longitudinal Study of Adolescent Health (Add Health). This is a nationally representative sample of 20,745 adolescents in grades 7-12 in the United States during the 1994-95 school year,17 followed into adulthood with four in-home interviews, the most recent in 2008, when the sample was ages 24-34. Participants were recruited using a stratified random sample of all US high schools. Eighty high schools and 65 feeder schools were enrolled in the study. Wave I (collected in 1994-1995) included in-school and in-home participant surveys, as well as an in-home parent survey. Additionally, a school administrator was surveyed from 79 of the 80 sample high schools. Wave IV (collected in 2007-2008) consisted of a follow-up interview for 15,500 of the initial participants.
We sought to identify whether measures of HS quality were associated with adult health, controlling for other contextual factors. Hence we restricted our sample to participants in grades 9-12 at Wave I who attended a sample HS for which the school administrator survey was available, and who completed the Wave IV survey (Figure 1). The resulting analytic sample includes 7,037 students from 77 high schools.
Figure 1. Definition of Analytic Sample.
School Quality Measures
The literature is mixed regarding how to define HS quality.18,19 Given that the primary goal of schools is to produce a more educated population, we selected school-level measures previously associated with improved academic outcomes. To validate this approach, we tested whether these measures were associated with HS graduation and college attendance in our sample, after controlling for socio-demographics.
Average Daily attendance
School-wide average daily attendance is associated with improved academic performance, even for students without individual absenteeism.20-22 High absentee rates are hypothesized as a marker for low student and family accountability and may result in teachers taking time from instruction to remediate absentee students. 22,23 In Add Health, school administrators estimated the school's average daily attendance level. Due to the response distribution, we collapsed the original 5-category variable into a 3-level variable with the categories of 75%-89%, 90%-94% and ≥95%. A sensitivity analysis using the original variable yielded similar results.
Student Promotion Rate
Grade retention and school drop-out rates, both of which are associated with poor academic outcomes at the classroom and school levels, have also been used as measures of HS quality.24,25 Low school promotion rates may be a marker for a school's inability to engage or support struggling students.26 School administrators reported the proportion of students in each grade who were held back and the proportion of students in each grade who dropped out of school during the 1993-1994 school year. Correlation between there variables was moderate to high and so, due to concerns for colinearity, we used these responses to calculate the percentage of students in each grade who either were promoted on to the next grade or completed HS and averaged this rate over all grades taught in each school to generate the overall student promotion rate. For ease of interpretation, we standardized the promotion rate such that 1 unit corresponds to 1 standard deviation. A sensitivity analysis using the average grade retention and drop-out rates in separate models yielded similar results.
Parental Involvement
School-level parental involvement is positively associated with academic outcomes.27 One marker for parental involvement is parent participation in school organizations, such as a Parent-Teacher Association.28 This measure is thought to better reflect school-level, as opposed to family-level, educational involvement, as it may convey greater normative social control.29 School administrators estimated the percentage of children with family members in a parent-teacher or other parent organization at school, as a continuous variable. Eleven percent of participants attended a school with no parent organization. To ensure those participants were included in the analysis, a 4-category measure was constructed for the percent of parents involved in school, based on the response distribution (0-14% of participation, 15-29% participation, 30%-100% participation, and no parent organization). Sensitivity analyses using different cut-points as well as with the continuous variable yielded similar results.
Teacher Experience
Teachers' skills have emerged as a particularly important factor in discussions of school quality.30 While studies are mixed regarding how to quantify teacher effectiveness at the individual level, having both experienced teachers and low teacher turnover are associated with improved student achievement at the school level.31,32 In Add Health, school administrators estimated the percentage of full-time classroom teachers that had worked at their school for five years or more, which addresses both teacher experience and teacher turnover. The percent reported was standardized, for ease of interpretation, such that 1 unit corresponds to one standard deviation and served as a continuous measure of teacher experience.
Other Potential School Quality Measures
School administrators were asked to estimate the school's average class size, the percent of students testing below, at, and above grade level, the percent of full-time classroom teachers with a master's degree or higher, and the percent of 12th graders who go on to attend either a 2-year or 4-year college. We also explored these measures as possible HS quality indicators.
Academic Outcomes
Individual academic outcomes were used to validate the school quality measures. High school transcripts, collected during Wave III (2001-2002), reported whether a student graduated HS with a diploma, obtained a GED, or neither. We constructed a dichotomous measure of HS diploma or GED versus no diploma and a 3-category outcome of HS diploma versus GED versus neither, to look for dose response patterns. In Wave IV, participants reported their highest level of education. Based on their responses we created both a dichotomous measure of attending college and a 4-category ordered outcome of no college, some college, college completion, and post-college education.
Health Outcomes
We chose outcomes for their high prevalence, impact on morbidity and mortality, and documented associations with education attainment.3-5 Our primary outcome is general health, which may encompass both physical health and psycho-social well-being. Secondary outcomes of obesity and depression were selected to assess for differential associations between school quality and physical versus mental health domains.
Self-Rated Health
Self-rated health is a well-studied general health measure associated with morbidity and mortality.33 In Waves I and IV, participants rated their health as excellent, very good, good, fair or poor. Given the response distributions in this relatively young and health population, we used a dichotomous measure of low self-rated health for responses of good, fair, or poor. A sensitivity analysis using the 5-category outcome revealed similar results.
Obesity
Obesity is estimated to affect 35% of US adults and is associated with higher health care utilization, lower quality of life, and increased mortality.34,35 In Add Health, self-reported height and weight were collected at Wave I, and measured height and weight at Wave IV. Based on these values we calculated participant body mass index (BMI) and BMI z-score for age and gender. A BMI at or above the 95th percentile in Wave I and ≥ 30 in Wave IV was considered obese.
Depression
Depression is among the most common mental health disorders and is associated with poor economic and health outcomes, including increased mortality.36,37 At Wave I, respondents completed the Center for Epidemiologic Studies Depression Scale (CES-D) short-form, a 10-item screening for depression symptoms in the previous 7 days.38 A score indicating high risk for depression (≥11) served as a dichotomous measure of high risk for depression during HS. In addition, respondents indicated whether they had seriously considered suicide in the previous 12 months. At Wave IV, participants reported whether they had ever been diagnosed with depression.
Covariates
Covariates were selected for their potential associations with both school quality and adult health. School factors included Wave I indicators for school size, type (public or private), urbanicity, and region. Socio-demographics included Wave I measures of participant age, race/ethnicity, gender, household income, highest level of parental education, family structure, and household language. Neighborhood material deprivation was measured using the 1990 census unemployment rate by block group. To account for the potential for school quality to impact adult health through improved access to health insurance, we controlled for whether the participant lacked health insurance during adolescence (Wave I), young adulthood (Wave III) and middle adulthood (Wave IV). To account for the possibility that improved school quality might impact adult health purely by improving individual academic performance, or that school quality is merely a proxy for the sum of individual academic performances, we controlled for participants' cumulative HS grade point average (GPA), as reported on their school transcripts. GPA might be considered a measure of both cognitive and non-cognitive abilities, both of which might contribute to the associations between education and health.39
Data Analysis
Missing data represented less than 5% of observations for all variables in our analysis, with the exception of household income and health insurance at Wave I, which were missing in 25% and 14% respectively. Because these items were likely to be missing in a non-random fashion, we included a category for missing for these two measures. A sensitivity analysis using multiply imputed data for these variables yielded similar results. Additionally, over 26% of the sample attended a school where no standardized testing was performed. Given the high degree of missing and likelihood for there to be both non-random and heterogeneous reasons for missing we eliminated this variable from our model. Analysis was conducted using the “svy” suite of commands in STATA Corp (Version 12) to account for the three survey design elements: stratification, clustering, and weighting. In particular, our data represent a single-stage design in which clustering occurs at the school level. Standard errors of regression coefficients from the following analyses were estimated using the Taylor linearization method that took into account data clustering, as well as stratification and weights. We used the “subpop” identifier to specify our analytic sample.
Validation of School Quality Measures
To determine whether HS quality measures were associated with HS graduation and college attendance, we used logistic regression for the dichotomous HS graduation and college attendance outcomes and multinomial regression for the 3-category HS graduation and 5-category college attendance outcomes. We examined the associations between potential HS quality measures and each academic outcome, after controlling for school (school size, school type, urbanicity, and region) and individual demographic covariates (age, gender, race/ethnicity, household income, parental level of education, family structure, home language). Finally, all potential school quality variables were included in the same model to examine the independence of the school quality variables. We selected variables for our main analysis that were significantly associated with HS graduation or college attendance with a two-sided p-value ≤ 0.1 in the expected direction and with the expected dose-response trend.
Main Analysis
We tested whether the selected measures of HS quality at Wave I were associated with adult health outcomes at Wave IV, controlling for Wave I adolescent health status. Each health outcome was modeled using survey weights and controlling for school and demographic covariates, insurance status, individual academic performance, and baseline health in the relevant domain.
Results
Demographics of our analytic sample were similar to the overall Add Health sample (Table 1) but differed slightly with respect to socio-economic status. The analytic sample had significantly greater percentages of female and white participants; were more likely to come from English-speaking households; and were less likely to come from households earning less than $25,000/year or have parents who did not graduate from HS. A sizable proportion of participants (38%) reported not having health insurance at one of the three time points. Approximately one-third reported less than very good health during adolescence, and nearly 42% reported low self-rated health in adulthood. At Wave I, 34% of adolescents reported symptoms of depression. However, less than 16% of adults in Wave IV reported ever being diagnosed with depression. By Wave IV, nearly 38% were obese. There were no statistically significant differences in the frequencies of the health outcome variables between our analytic sample and those excluded from our analysis.
Table 1. Participant Demographics and Descriptive Statisticsa.
| Measure | Percent (number)/Mean (SD) | Measure | Percent (number)/Mean (SD) |
|---|---|---|---|
| Gender | High School Graduation Status | ||
| Female | 51.9% (3760) | High school diploma | 90.7% (6400) |
| Male | 48.1% (3277) | GED | 5.3% (353) |
| Age | No diploma or GED | 4.1% (283) | |
| 13-14 | 5.5% (367) | Highest Educational Attainment | |
| 15-16 | 47.6% (3409) | More than college | 12.3% (893) |
| 17-18 | 44.1% (3094) | College degree | 23.0% (1658) |
| 19+ | 2.8% (165) | Some college | 34.4% (2433) |
| Race/Ethnicity | No college | 30.3% (2052) | |
| White Non-Hispanic | 70.9% (3913) | Mean Cumulative High School GPA | 2.6 (0.9) |
| African American | 13.6% (1277) | High School Size | |
| Hispanic | 9.6% (1098) | 1001-4000 students | 64.5% (4879) |
| API/Native Amer./Other | 5.9% (729) | 401-1000 students | 25.7% (1611) |
| Home Language | 400 or fewer students | 9.8% (547) | |
| English | 93.1% (6233) | High School Urbanicity | |
| Spanish/Other | 6.9% (804) | Urban | 25.3% (1908) |
| Annual Household Income | Suburban | 56.2% (3752) | |
| $0-$24,000 | 18.7% (1341) | Rural | 18.5% (1377) |
| $25,000-$49,000 | 26.1% (1778) | High School Type | |
| $50,000-$74,000 | 19.5% (1318) | Public | 93.2% (6532) |
| $75,000 or more | 12.6% (836) | Private | 6.8% (505) |
| Missing | 23.1% (1764) | Region | |
| Highest Parental Education | West | 17.2% (1793) | |
| Less than high school degree | 10.5% (787) | Midwest | 33.7% (1972) |
| High school graduate/GED | 30.9% (1884) | South | 36.4% (2440) |
| Some college | 22.8% (1487) | Northeast | 12.7% (832) |
| College degree or more | 35.8% (2560) | Average % of experienced teachers at the high school | 68.6% (20.5) |
| Family Structure | Average high school promotion rate | 91.2 (7.9) | |
| 2 Biological parents | 58.3% (3958) | Average high school daily attendance | |
| 1 Biologic/1 step-parent | 16.0% (1208) | 95% or more | 28.3% (1845) |
| Single parent | 19.8% (1486) | 90%-94% | 55.7% (3522) |
| Other | 5.9% (385) | 75%-89% | 16.0% (1670) |
| Health Insurance Status | Average % of parents involved in school PTA | ||
| No health insurance at Wave I | 10.0% (705) | 30-100% | 18.5% (1407) |
| No health insurance at Wave III | 23.7% (1598) | 15-29% | 18.3% (1037) |
| No health insurance at Wave IV | 18.4% (1224) | 0-14% | 48.6% (3820) |
| No PTA at the school | 14.5% (773) | ||
| Adolescent Health | Adult Health | ||
| Low self-rated health | 32.7% (2282) | Low self-rated health | 41.6% (2933) |
| Positive symptoms of depression | 33.6% (2442) | Diagnosis of Depression | 15.5% (1004) |
| Suicidal ideation | 15.2% (999) | Obesity | 38.5% (2650) |
| Obesity | 9.9% (683) |
Percentages reflect survey weights, SD= standard deviation, PTA=Parent teacher association.
Most participants graduated from a public HS and went on to pursue higher education. On average, school administrators reported more than two-thirds of the teachers had taught at their school for ≥5 years and over 91% of students were promoted to the next grade. Approximately 28% of participants attended schools with high average daily attendance and 16% attended schools with low average daily attendance. Parent involvement at participants' schools ranged from 0-100% but nearly half the participants (49%) attended a school where less than 15% of the parents participated.
School Quality and Educational Attainment
School-level average class size, the percent of highly educated teachers, and the percent of 12th graders who go on to attend either a 2-year or 4-year college did not independently predict individual HS graduation or college attendance after controlling for school and demographic covariates. The remaining candidate HS quality measures (average daily attendance, promotion rate, percent of experienced teachers, and parental involvement) all predicted HS graduation or college attendance in both bivariate and multivariate models (results not shown). Logistic regressions of Wave III HS graduation and college attendance on Wave I HS quality (Table 2) revealed that higher promotion rate and exposure to more experienced teachers were associated with higher odds of earning a HS degree or attending college, while lower average daily attendance was associated with lower odds of earning a HS degree or attending college. Attending a HS with no parent-association was associated with lower odds of earning a HS degree. However, attending a school with extremely high levels of parental involvement was also associated with lower odds of earning a HS degree. Multinomial regressions (eTable1) revealed the hypothesized dose response relationship for all school quality variables, again with the exception of parental involvement, as attending a school with extremely high or low parental involvement was associated with lower odds of graduating from HS.
Table 2. Logistic Regressions of High School Graduation and College Attendance on School Quality Measuresa.
| High School Graduation or GED | College Attendance | |||
|---|---|---|---|---|
| Odds Ratio | 95% CI | Odds Ratio | 95% CI | |
| Average Daily Attendance | ||||
| 95% and above | Reference | |||
| 90%-94% | 0.71 | 0.41 – 1.22 | 0.92 | 0.67 – 1.26 |
| 75%-89% | 0.41 | 0.17 – 1.03 | 0.69 | 0.44 – 1.07 |
| Promotion Rate (unit=1 SD) | 1.30 | 0.99 – 1.69 | 1.18 | 0.98 – 1.41 |
| % Experienced Teachers (unit=1 SD) | 1.62 | 1.29 – 2.02 | 1.24 | 1.10 – 1.39 |
| Parental Involvement | ||||
| 0%-14% | Reference | |||
| 15%-29% | 0.81 | 0.30 – 2.15 | 0.98 | 0.72 – 1.34 |
| 30%-100% | 0.60 | 0.33 – 1.07 | 1.26 | 0.85 – 1.85 |
| No PTA | 0.27 | 0.13 – 0.56 | 0.85 | 0.65 – 1.11 |
95% CI= 95% Confidence Interval, SD= Standard deviation, PTA= Parent teacher association. Models control for school size, school type, urbanicity, region, age, gender, race/ethnicity, household income, parental level of education, family structure, and home language.
School Quality and Adult Health
School average daily attendance, promotion rate, parent involvement, and teacher experience at Wave I were then used as predictors of adult health outcomes at Wave IV (Table 3, eTable2). After we controlled for socio-demographics, baseline health status, health insurance, and individual school performance, HS quality was significant associated (either positively or negatively) with all adult health outcomes. Attending a HS with low average daily attendance was associated with higher odds of poor self-rated health (adjusted odds ratio (AOR) 1.59, p-value 0.003) and higher odds of being diagnosed with depression (AOR 1.35, p-value=0.03) in adulthood. Higher promotion rate was associated with higher odds of low self-rated health (AOR 1.20, p-value <.001), and higher parent involvement at school was associated with decreased odds of adult obesity (AOR 0.69, p-value 0.001).
Table 3. Associations between High School Quality and Adult Health Outcomes Adjusted for Covariatesa.
| Low Self-Rated Health | Diagnosis of Depression | Obesity | ||||
|---|---|---|---|---|---|---|
| Odds Ratio | 95% CI | Odds Ratio | 95% CI | Odds Ratio | 95% CI | |
| Average Daily Attendance | ||||||
| 95% and above | Reference | Reference | ||||
| 90%-94% | 1.09 | 0.87 – 1.36 | 1.35* | 1.04 - 1.75 | 1.00 | 0.85 - 1.19 |
| 75%-89% | 1.59** | 1.17 – 2.16 | 1.36 | 0.92 - 2.02 | 1.19 | 0.90 – 1.57 |
| Promotion Rate (unit=1 SD) | 1.20*** | 1.09 - 1.32 | 1.16 | 1.00 - 1.35 | 1.02 | 0.90 - 1.17 |
| % Experienced Teachers (unit=1 SD) | 0.95 | 0.87 - 1.03 | 1.10 | 0.98 - 1.24 | 0.95 | 0.87 - 1.03 |
| Parental Involvement | ||||||
| 0%-14% | Reference | Reference | Reference | |||
| 15%-29% | 1.18 | 0.94 - 1.48 | 1.18 | 0.85 - 1.63 | 1.06 | 0.80 - 1.39 |
| 30%-100% | 1.07 | 0.83 - 1.38 | 0.85 | 0.60 - 1.21 | 0.69** | 0.55 – 0.85 |
| No PTA | 1.04 | 0.80 - 1.34 | 1.18 | 0.85 - 1.63 | 1.06 | 0.80 - 1.39 |
p<.05;
p<.01;
p<.001.
95% CI=95% Confidence Interval, SD= Standard deviation, PTA= Parent teacher association. Each of the three models control for baseline health outcomes, school size, school type, urbanicity, region, age, gender, race/ethnicity, household income, parental level of education, family structure, home language, neighborhood unemployment, lack of health insurance, and individual grade point average.
Discussion
Several indicators for HS quality significantly predicted adult health outcomes, even after controlling for many likely confounders. These results align with previous longitudinal studies suggesting school quality may be an under-recognized social determinant of health.13-16 The effect sizes found in this study are moderate, yet comparable to associations between self-rated health and access to health care and obesity.40 For example, a decrease in average daily attendance from 95% and above to 90-94% was associated with a 35% increased odds of depression diagnosis in adulthood. Compared to other variables in the models, the effect sizes for the school quality measures were equal or greater in magnitude than those for parental level of education and having health insurance.
HS quality measures each predicted at least one of our three examined adult health outcomes (self-rated health, depression, obesity), in either the expected or unexpected direction. As expected, lower average daily attendance predicted lower self-rated adult health and higher odds of depression diagnosis, while more school-level parental involvement predicted lower odds of obesity. With respect to obesity, although the school quality validation analyses suggest that some schools with extremely high levels of parental involvement may be schools requiring more parental oversight, parental involvement may nevertheless signify greater overall parent engagement in school, which might result in healthier nutrition and physical activity patterns. The relationship between promotion rate and health, however, appears more complex. A higher school-level promotion rate was associated with low self-rated health in adulthood. It is unclear what underlies this finding. It is possible, for instance, that high promotion rates in the absence of other essential aspects of individual academic achievement or school quality (“social promotion”) might not significantly benefit adult health. Future studies might further explore these findings by delineating the specific mechanisms through which various measures of high school quality might contribute to long-term health.
These results reinforce the notion of schools as critical platforms for population health and the intimate connection between education and health policy. The outcomes associated with school quality in this analysis (self-rated health, depression, and obesity) produce considerable long-term health and economic costs to society. If school quality might be leveraged to prevent negative health outcomes, even small individual level effects might have large impacts across a population. Identifying specific aspects of school quality that might be leveraged to produce health can inform policy-makers, advocates, educators, and health providers. This may require developing robust school quality measures that go beyond standardized test performance and are specifically focused on population health. Such measures might be used to both identify children at risk for poor adult health outcomes and to develop and evaluate interventions to improve school quality. Given the international conversation about value-based spending, these findings also highlight the need to consider the inter-connectedness of the education and health sectors when calculating societal returns on investment. For example, policies supporting school attendance might be evaluated in terms of both their education and health impacts. Given the recent emphasis on tracking and addressing rates of chronic absenteeism in the United States41, these results suggest that health advocates might also be engaged in evaluating the benefits of such initiatives.
This study is limited by the school quality measures available in Add Health. In particular, it is possible that school administrators incorrectly estimated the measures included here and that alternative measures, such as per pupil spending or access to school-based health care, might yield different results. Additionally, although we control for many socio-demographic and contextual factors, we cannot exclude the possibility that findings appear significant due to unmeasured variables. In particular, although we controlled for neighborhood level poverty, we were unable to control for the percent of low-income students attending each school. Using HS transcript and school administrator data substantially reduced our sample from the initial study population, which could have introduced bias. Individuals including in our analytic sample were more like to be white and to come from families that were English speaking and of higher socio-economic status. As a result, although our analysis controls for these variables, our findings may not be generalizable to low-income, minority populations. Of note, however, there were no differences in health outcomes comparing those included with those excluded in this analysis. Finally, this study might underestimate the true impact of school quality on health due to both an inability to account for the school environment prior to HS and the lack of follow-up past the 4th decade.
Despite these limitations, this analysis challenges us to think holistically about how education supports population health. Schools are critical public institutions that may have the potential to impact a child's long-term health trajectory. As health advocates, health care providers and public health practitioners may need to consider the degree to which they should engage at school, district, state, and federal levels to support school quality as a mechanism for improving lifelong health and reducing health disparities.
Supplementary Material
Highlights.
Individual academic achievement is a well-known predictor of health.
Few studies investigate whether school quality is also associated with long-term health.
We analyzed Add Health to test whether high school quality predicted adult health.
High school quality predicted self-rated health, depression, and obesity.
School quality may be an under-recognized social determinant of health.
Acknowledgments
This study was funded by a grant from the NIH/National Center for Advancing Translational Sciences (UL1TR000124) and the UCLA Children's Discovery and Innovation Institute.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Contributor Information
Bergen B. Nelson, Email: bnelson@mednet.ucla.edu.
Tumaini R. Coker, Email: tcoker@mednet.ucla.edu.
Christopher Biely, Email: cbiely@mednet.ucla.edu.
Ning Li, Email: nli@biomath.ucla.edu.
Lynne C. Wu, Email: lynne.chang@gmail.com.
Paul J. Chung, Email: paulchung@mednet.ucla.edu.
References
- 1.Winkleby J, Frank, Fortmann Socioeconomic status and health: How education, income and occupation contribute to risk factors for cardiovascular disease. American Journal of Public Health. 1992;82:816–820. doi: 10.2105/ajph.82.6.816. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Eide ER, Showalter MH. Estimating the relation between health and education: What do we know and what do we need to know? Economics of Education Review. 2011;30(5):778–791. [Google Scholar]
- 3.Cutler DM, Lleras-Muney A. Education and health: evaluating theories and evidence. National Bureau of Economic Research; 2006. [Google Scholar]
- 4.Cutler DM, Lleras-Muney A. Understanding differences in health behaviors by education. Journal of health economics. 2010;29(1):1–28. doi: 10.1016/j.jhealeco.2009.10.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Cohen AK, Rai M, Rehkopf DH, Abrams B. Educational attainment and obesity: a systematic review. Obesity Reviews. 2013;14(12):989–1005. doi: 10.1111/obr.12062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Herd P. Education and Health in Late-life among High School Graduates. Journal of Health and Social Behavior. 2010 Dec 1;51(4):478–496. doi: 10.1177/0022146510386796. 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Dewalt DA, B ND, Sheridan S, Lohr KN, Pignone MP. Literacy and health outcomes: a sytematic review of the literature. Journal of General Internal Medicine. 2004;19:1228–1239. doi: 10.1111/j.1525-1497.2004.40153.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mazzonna F. The long lasting effects of education on old age health: Evidence of gender differences. Social Science & Medicine. 2014;101(0):129–138. doi: 10.1016/j.socscimed.2013.10.042. [DOI] [PubMed] [Google Scholar]
- 9.Mechanic D. Disadvantage, Inequality, And Social Policy. Health Affairs. 2002 Mar 1;21(2):48–59. doi: 10.1377/hlthaff.21.2.48. 2002. [DOI] [PubMed] [Google Scholar]
- 10.Walsemann KM, Gee GC, Ro A. Educational Attainment in the Context of Social Inequality: New Directions for Research on Education and Health. American Behavioral Scientist. 2013 Aug 1;57(8):1082–1104. 2013. [Google Scholar]
- 11.Liang W, Chikritzhs T. Age at first use of alcohol predicts the risk of heavy alcohol use in early adulthood: A longitudinal study in the United States. International Journal of Drug Policy. doi: 10.1016/j.drugpo.2014.07.001. 0. [DOI] [PubMed] [Google Scholar]
- 12.Brook JS, Lee JY, Brook DW, Finch SJ. Determinants of Obesity: Results from a Longitudinal Study of Adolescents and Adults Living in an Urban Area1, 2. Psychological Reports. 2013 Dec 01;113(3):717–733. doi: 10.2466/15.13.PR0.113x26z6. 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Walsemann KM, Geronimus AT, Gee GC. Accumulating Disadvantage Over the Life Course: Evidence From a Longitudinal Study Investigating the Relationship Between Educational Advantage in Youth and Health in Middle Age. Research on Aging. 2008 Mar 1;30(2):169–199. 2008. [Google Scholar]
- 14.Johnson RC. The Health Returns of Education Policies from Preschool to High School and Beyond. The American Economic Review. 2010;100(2):188–194. doi: 10.1257/aer.100.2.188. [DOI] [PubMed] [Google Scholar]
- 15.Frisvold D, Golberstein E. School quality and the education–health relationship: Evidence from Blacks in segregated schools. Journal of health economics. 2011;30(6):1232–1245. doi: 10.1016/j.jhealeco.2011.08.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Frisvold D, Golberstein E. The Effect of School Quality on Black-White Health Differences: Evidence From Segregated Southern Schools. Demography. 2013 Dec 01;50(6):1989–2012. doi: 10.1007/s13524-013-0227-z. 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Klein JD. The National Longitudinal Study of Adolescent Health: Preliminary Results: Great Expectations. JAMA, The Journal of the American Medical Association. 1997;278:864–865. [PubMed] [Google Scholar]
- 18.Card D, Krueger AB. Labor market effects of school quality: Theory and evidence. National Bureau of Economic Research; 1996. [Google Scholar]
- 19.Hanushek EA. Some simple analytics of school quality. National Bureau of Economic Research; 2004. [Google Scholar]
- 20.Roby DE. Research On School Attendance And Student Achievement: A Study Of Ohio Schools. Educational Research Quarterly. 2004;28(1):3–16. [Google Scholar]
- 21.Lamdin DJ. Evidence of student attendance as an independent variable in education production functions. The Journal of Educational Research. 1996;89(3):155–162. [Google Scholar]
- 22.Gottfried MA. Absent peers in elementary years: The negative classroom effects of unexcused absences on standardized testing outcomes. Teachers College Record. 2011;113(8):1597–1632. [Google Scholar]
- 23.Lazear EP. Educational Production. The Quarterly Journal of Economics. 2001 Aug 1;116(3):777–803. 2001. [Google Scholar]
- 24.Rumberger RW, Thomas SL. The distribution of dropout and turnover rates among urban and suburban high schools. Sociology of Education. 2000:39–67. [Google Scholar]
- 25.Gottfried MA. The Spillover Effects of Grade-Retained Classmates: Evidence from Urban Elementary Schools. American Journal of Education. 2013;119(3):405–444. [Google Scholar]
- 26.Lee VE, Burkam DT. Dropping out of high school: The role of school organization and structure. American Educational Research Journal. 2003;40(2):353–393. [Google Scholar]
- 27.Griffith J. Relation of Parental Involvement, Empowerment, and School Traits to Student Academic Performance. The Journal of Educational Research. 1996;90(1):33–41. [Google Scholar]
- 28.Georgiou SN. Parental involvement: Definition and outcomes. Social Psychology of Education. 1996;1(3):189–209. [Google Scholar]
- 29.Sui-Chu EH, Willms JD. Effects of parental involvement on eighth-grade achievement. Sociology of Education. 1996:126–141. [Google Scholar]
- 30.Rivkin SG, Hanushek EA, Kain JF. Teachers, schools, and academic achievement. Econometrica. 2005;73(2):417–458. [Google Scholar]
- 31.Ronfeldt M, Loeb S, Wyckoff J. How Teacher Turnover Harms Student Achievement. American Educational Research Journal. 2013 Feb 1;50(1):4–36. 2013. [Google Scholar]
- 32.Lubienski Sarah Theule, Lubienski Christopher, Crane Corinna Crawford. Achievement Differences and School Type: The Role of School Climate, Teacher Certification, and Instruction. American Journal of Education. 2008;115(1):97–138. [Google Scholar]
- 33.Idler EL, Benyamini Y. Self-Rated Health and Mortality: A Review of Twenty-Seven Community Studies. Journal of Health and Social Behavior. 1997 Mar;38(1):21–37. 1997. [PubMed] [Google Scholar]
- 34.Visscher TL, Seidell JC. The Public Health Impact of Obesity. Annual Review of Public Health. 2001;22(1):355–375. doi: 10.1146/annurev.publhealth.22.1.355. [DOI] [PubMed] [Google Scholar]
- 35.Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of childhood and adult obesity in the United States, 2011-2012. JAMA. 2014;311(8):806–814. doi: 10.1001/jama.2014.732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Zheng D, Macera CA, Croft JB, Giles WH, Davis D, Scott WK. Major depression and all-cause mortality among white adults in the United States. Annals of Epidemiology. 1997;7(3):213–218. doi: 10.1016/s1047-2797(97)00014-8. [DOI] [PubMed] [Google Scholar]
- 37.Control CfD, Prevention. QuickStats: Prevalence of Current Depression* Among Persons Aged≥ 12 Years, by Age Group and Sex—United States, National Health and Nutrition Examination Survey, 2007-2010. MMWR Morb Mortal Wkly Rep. 2012;60(51):1747. [Google Scholar]
- 38.Bradley KL, Bagnell AL, Brannen CL. Factorial Validity of the Center for Epidemiological Studies Depression 10 in Adolescents. Issues in Mental Health Nursing. 2010;31(6):408–412. doi: 10.3109/01612840903484105. [DOI] [PubMed] [Google Scholar]
- 39.Conti G, Hansman C. Personality and the education–health gradient: A note on “Understanding differences in health behaviors by education”. Journal of health economics. 2013;32(2):480–485. doi: 10.1016/j.jhealeco.2012.07.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Phillips LJ, Hammock RL, Blanton JM. Predictors of Self-rated Health Status Among Texas Residents. Preventing Chronic Disease. 2005;2(4):A12. [PMC free article] [PubMed] [Google Scholar]
- 41.Force MBsKT. Report to the President. 2014:A-7–A-6. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.