pmc03765076800Public HealthPublic HealthPublic health0033-35061476-5616399086031188499610.1016/j.puhe.2025.01.019NIHMS2054705ArticleUtilize administrative data to assess impact of individual-level and neighborhood-level socioeconomic status on childhood health in a longitudinal studyMeeksHuong D.12YuDavid Zhe2FraserAlison2MaloneyThomas N.3CurtinKaren24NewmanMike2SmithKen R.5Department of Pediatrics, University of Utah School of Medicine, Salt Lake City, UTPedigree and Population Resources, Huntsman Cancer Institute, Salt Lake City, UTDepartment of Economics, University of Utah, Salt Lake City, UTDepartment of Internal Medicine, University of Utah School of Medicine, Salt Lake City, UTDepartment of Family and Consumer Studies, University of Utah, Salt Lake City, UTCorresponding author: Huong D. Meeks, huong.meeks@hsc.utah.edu, Department of Pediatrics, Critical Care Division, University of Utah Spencer F. Eccles School of Medicine, 295 Chipeta Way, Sale Lake City, UT 8410813220253202504220250132026240154160Objectives:

Our study aims to 1) propose an approach to standardize individual-level and neighborhoodlevel socioeconomic status (SES) measures from administrative records across three decades, 2) determine the degree of agreement between individual-level and neighborhood-level SES measures, and 3) compare the associations of individual-level and neighborhood-level SES measures and childhood health.

Study Design:

Retrospective observational study

Methods:

Our cohort included children born in Utah during 1978-2008. Parental SES indicators including education, occupation, and area household income corresponding to maternal addresses were collected from birth certificates. Spearman rank correlation tests were used to determine the correlation between SES measures. Cox regression models and causal mediation analyses were used to assess the association between three childhood health outcomes- mortality, failure-to-thrive, and asthma- and three SES

measures:

highest education level and occupational status score of parent(s) and area household income.

Results:

819,108 children were included in the final analysis. We observed moderate agreement (r ranges 0.43-0.49) between individual-level SES measures (education and occupation) and weak agreement (r ranges 0.19-0.25) between these measures and neighborhood-level SES measures. Parental education had the strongest association with childhood health. The associations of parental occupation and area household income with childhood health, and their mediating effects on parental education, varied by the severity of health outcomes.

Conclusions:

We introduced an approach to standardize SES measures from administrative records for longitudinal analyses. We confirmed that individual-level and neighborhood-level SES measures are distinct indicators of social standing and validated the importance of the choice of SES measures when analyzing childhood health.

Graphical Abstract

Administrative recordsSocioeconomic StatusChildhood HealthLongitudinal StudyOccupational Status ScoreACSEducationFundamental Cause TheoryCausal Mediation AnalysesIntroduction:

Administrative data, including electronic health and vital records, have proven useful in epidemiologic research for several reasons [1]. First, they are readily available and inexpensive to acquire as opposed to data traditionally collected by prospective studies, which are costly and time-consuming. Second, they typically include rich longitudinal data on entire regional populations, which make tracking of disease onset and progression possible. Third, they allow comprehensive observation of individuals across systems from birth to adulthood. Digitization of administrative data has allowed researchers to perform linkages not only between multiple records but also among household members and to contextual data via geographic information systems [2], which is especially useful for studying complex diseases.

One challenge of using administrative data for epidemiologic research is that they were originally intended for other purposes. Individual-level and neighborhood-level socioeconomic status (SES) is rarely directly evaluated despite being shown to be consistently associated with childhood health via its influence on household residential choices, access to and use of healthcare services, and parenting practices [3–5]. Common SES measures from administrative data are education and occupation as proxies for individual-level SES and characteristics of residential areas drawn from census data. Occupation in administrative records is stored as text strings and thus is difficult and expensive to codify. Income is typically collapsed into large categories and top-coded, which then requires complex adjustments to make year-to-year comparisons meaningful in longitudinal studies. Due to the lack of availability and arduous efforts to access individual-level SES measures in administrative records, neighborhood-level SES measures have commonly been used as surrogates [6]. Several studies have assessed the agreement between individual-level and neighborhood-level SES measures [7–9], but few have focused specifically on childhood health [10, 11] and none has utilized SES measures reported in administrative records.

When utilizing SES measures to examine direct effects of SES on health outcomes or as effect modifiers of other risk factors, many longitudinal studies do not account for the instability and inconsistency of these measures across time and places. Researchers often take a cross-sectional approach to describe neighborhood-level SES such as using SES variables from cross-sectional public surveys including the decennial censuses and the American Community Survey (ACS) [12]. Researchers also adopt a categorical approach where individual-level SES is depicted in broad categories, such as unemployed versus employed, or manual versus skilled occupations [13].

Neighborhood-level SES is dynamic [14], and aggregating occupations into small numbers of broad categories or classes assumes that workers within a large category are relatively homogenous and that they were members of these categories for the same amount of time [15]. Various methodologies have been developed to ensure the comparability of neighborhood-level SES measures across populations, places, and time [16], but have not been utilized often in pediatric health research [17]. Kachmar et al (2019) conducted a scoping review of SES measures in pediatric studies in the United States between 1999 and 2018 [6]. Of 228 studies, only 22 (9.6%) included employment/occupation as a measure of SES, aggregated into groupings such as employment status and occupation types, and 58 (25.4%) included census-derived income. No study employed occupational measures or harmonized census-derived variables across long periods of time.

Our study aims were to: 1) propose a new approach to standardize individual-level and neighborhoodlevel SES measures from administrative records across multiple decades to support longitudinal analyses; 2) determine the degree of agreement between individual-level and neighborhood-level SES measures; and 3) compare the association of individual-level and neighborhood-level SES measures and childhood health outcomes.

Material and Methods:Data:

This retrospective observational study utilized data drawn from the Utah Population Database (UPDB), one of the world’s richest sources of linked population-based information [18]. Currently, the UPDB contains demographic, genealogical, and health data on over 11 million individuals from the late 18th century to the present, representing Utah’s population that appear in a range of administrative records. UPDB data is continually updated and linked at the individual-level and family-level as annual statewide birth and death certificates, hospitalizations, ambulatory surgeries, and driver licenses become available. This study is approved by the University of Utah’s Resource for Genetic and Epidemiologic Research, the body that governs use of UPDB data, and its Institutional Review Board (IRB #00043524).

Study Population:

Eligible children included those born in Utah from 1978 to 2008 with informative parents, defined as those with known birth year and duration of Utah residency. We chose the study period of 1978-2008 because parental education and occupation information as well as maternal residence and mailing addresses were available from Utah birth certificates, allowing us to measure family and neighborhood SES from these records. Additional inclusion criteria were: 1) Utah residency; 2) known birth month/year and sex; and 3) known birth order, birth county, and maternal marital status. We excluded stillbirths and adoptions, as data for these are not consistently available in the UPDB. Children whose parents were informative but lacked information on education, occupation, and maternal addresses were excluded because this information was deemed essential in determining SES. We also excluded children from military families because of their unique life experiences due to factors such as reassignments, deployments, and the frequency of location changes. Eligible children were followed from birth until death, age 18, or last date residing in Utah, whichever occurred first.

Outcomes:

Death month/year and month/year of first asthma and failure-to-thrive (FTT) diagnoses were collected for all eligible children. Childhood mortality was determined from Utah death certificates and was defined as death prior to age 18. Asthma diagnosed prior to age 18 and FTT diagnosed prior to age 2 were of interest and chosen specifically because they are among the most common issues in children for which socioeconomic status has been demonstrated to be a risk factor [19–21]. These diagnoses were determined from UPDB’s healthcare facility database using the appropriate International Classification of Disease codes version 9 and 10 (Supplemental Table 1).

SES measures:

Our study used parental education attainment(s) and occupation(s) to measure individual-level SES and Census median household income estimates to measure neighborhood-level SES. All SES measures were ascertained from birth certificates and via their addresses. Parental education included the maximum of paternal and maternal education and was categorized into five groups: less than high school, high school graduate, some college, college graduate, and post college. Parental occupation was derived from parental occupational status and the Nam-Powers-Boyd occupation status scores (NPBOSS). Children were assigned into quartiles based on the highest NPBOSS of the two parents; children from households where both parents were unemployed were categorized into a distinct category. Area household income was categorized into quartiles based on the median household income associated with the maternal residential addresses at the time of the child’s birth. Further details on the imputations of SES measures were provided in the Appendix.

Confounders:

The list of confounders was informed by prior literature and included the following demographic characteristics of the child: birth year, sex, race (whether White), ethnicity (whether Hispanic), birth order (whether first born), and the following maternal characteristics at the child’s birth: residential county classification (whether urban, rural, or frontier), age, and marital status (whether married) [22].

Statistical Analysis:

Spearman rank correlation tests were used to determine the association between the following SES measures: maximum education obtained by each parent individually and combined, parental occupation, and area household income. All parental education measures were ranked from one to five, with one being the lowest level (less than high school). Parental occupation was ranked from one to five, with families whose parents were unemployed ranked as one and families in the fourth quartile ranked as five. Area household income was ranked from one to four, with neighborhoods in the first quartile received the ranking of one.

We investigated the relationship between childhood health and the three SES measures. First, we treated these SES measures as continuous variables and performed seven Cox regression models with increasing levels of adjustment for each outcome: models 1-3 adjusted for each SES measure separately, models 4-6 adjusted for pairs of SES measures, and model 7 included all three SES measures simultaneously. Second, we conducted causal mediation analyses based on a counterfactual framework to evaluate whether parental occupation and area household income mediated the effect of parental education [23]. Single mediation models were performed for each mediator, along with a multiple mediation model incorporating both mediators. Each mediation analysis included two components: a Cox regression model estimating the direct effect of parental education adjusting for mediator(s), and a linear regression model assessing the association between parental education and the mediator(s). Sensitivity analyses incorporating exposure-mediator interactions were performed for all models. Third, findings from the causal mediation analyses were used to inform the development of Cox regression models examining the relationship between SES measures, treated as categorical variables, and childhood health.

Cox regression was chosen in order to take time and observation periods of differing lengths into consideration and fitted with the sandwich estimator to account for clustering induced by multiple births within families. We did not consider fixed-effect models as parental measures within sibship clusters do not vary. All models were adjusted for covariates described earlier. Statistical analyses were conducted using R version 4.3 (R Core Team, 2022) with the packages tidyverse [24] and CMAverse [25].

Results:

Of the initial 1,234,088 children born in Utah between 1978 and 2008, 1,144,250 matched our eligibility criteria. After excluding those with missing key socioeconomic information, a total of 819,108 children were included in the final analysis (Supplemental Figure 1).

The mean (standard deviation [SD]) follow-up time is 15.9 (4.1) years. There were 8,134 (1.0%) children who died prior to age 18, 25,281 (3.1%) children diagnosed with FTT prior to age 2, and 35,594 (4.4%) children diagnosed with asthma prior to age 18. Sociodemographic characteristics at birth are presented in Table 1. For 29.0% children, maximum parental education was high school graduate or less. Most children came from dual-income and single-income families (49.0% and 47.7% respectively); among single-income families, the majority had fathers being the sole workers outside of the home (44.5%).

Spearman Correlations of SES measures:

Individual-level SES measures were moderately correlated with each other and weakly correlated with neighborhood-level SES measure [26] (Figure 1; all p<0.001). Paternal education was moderately correlated with maternal education (r=0.63), with fathers generally reporting higher education than mothers. Parental occupation was moderately correlated with all three measures of parental education (r ranges 0.43-0.49). Area household income was weakly correlated with parental education measures (r ranges 0.16-0.19) and occupation (r=0.25).

Associations between SES measures and childhood health:

Tables 2 summarizes the associations of parental education, occupation, and area household income with childhood health. Overall, children from lower SES backgrounds showed significantly higher rates of mortality and morbidity compared to those from higher SES backgrounds. The magnitudes of hazard ratios (HRs) were generally consistent whether SES measures were adjusted individually or jointly, except for parental occupation, which was significant in models with asthma as the outcome and excluding parental education but not in those including it.

Parental occupation mediated the association between parental education and childhood mortality (Proportion Mediated [PM] [95% Confidence Interval {CI}] = 20.6 [14.3-27.9], p<0.001) and FTT (PM [95%CI] = 8.9 [3.2-15.4], p<0.001) but not childhood asthma (Supplemental Table 2). Area household income mediated the association between parental education and all health outcomes, though with marginal effect (PM ranges 0.6-3.1%). Sensitivity analyses confirmed the mediating effect of parental occupation and area household income on the relationship between parental education and childhood mortality, with reduced estimates (PM [95%CI] = 13.6 [7.2-21.3] and 2.2 [1.2-3.6], respectively), but not on other outcomes (Figure 2, Supplemental Table 3).

All three SES measures were included in the final Cox models (Table 3). Compared to children of parents with a high school degree, children of parents with less than a high school education had higher HRs of mortality and FTT while those whose parents had some college education or more had lower HRs of mortality and morbidity. Compared to children with parents of highest occupational status, children with unemployed parents had higher HRs for mortality but not FTT while those with parental NPBOSS in the 1st quartile had higher HRs for both mortality and FTT. We did not observe significant association between parental occupation and childhood asthma. Children living in the poorest neighborhoods consistently had higher HRs for both mortality and morbidity compared to children living in the wealthiest neighborhoods.

Discussion:

This study is one of only a few studies that standardized multiple individual-level and neighborhoodlevel socioeconomic measures collected in administrative records across a three-decade study period to examine their associations with childhood health. We observed moderate agreement between parental education and occupation and weak agreements between these measures and area household income. The magnitude of the relationship between SES measures at birth and childhood health differed depending on not only the choices of SES measures but also the outcomes, with parental education showing the most consistent effects.

Our observed low to moderate correlations between SES measures at birth are consistent with previous literature [8, 27] and confirm that SES measures, either between individual-level and neighborhoodlevel or within the same level often do not overlap fully and are distinct indicators of social standing. Parental education, parental occupation, and area household income each represent unique aspects of a child’s socioeconomic background. Parental education indicates the level of knowledge and skills obtained through formal education, which can affect parents’ ability to support their child’s development. Parental occupation reflects the nature of employment, which is consistently linked to income security, earnings stability, and economic prospects. It might also act as a proxy for social standing, potential networks, and the resources accessed through the occupational social environment, while also reflecting job-related exposures that a parent may pass on to their child. Area household income provides a measure of the overall economic environment for a neighborhood and can impact child health through various pathways, including access to healthcare and environmental quality. Relying on a single SES indicator will likely risk overlooking the diverse ways in which socioeconomic background influences childhood health [28, 29].

Parental education was associated with all three childhood outcomes, whereas the association of parental occupation and area household income with childhood health varied by the outcomes. Children of highly educated parents have lower mortality and morbidity compared to children of parents with lower educational attainment. This finding supports the Fundamental Cause Theory (FCT) hypothesis, which suggests that people of higher SES have more health advantages due to flexible resources, irrespective of health conditions [30]. Our study lends credence to the use of FCT as a conceptual framework for the exploration of health disparities in early childhood [31–33]. Interestingly, our study found a significant difference in mortality but not morbidity between children with unemployed parents and those with parents of highest occupational status. We hypothesized that this is due to the underdiagnoses in children of unemployed parents and the overdiagnoses in children with parents of higher occupational status. Our finding of higher mortality and morbidity risks in children in areas with lowest household income suggests the presence of structural disparities likely resulting from a lack of resources in less affluent neighborhoods [34].

The magnitudes of the effects of SES measures vary by outcomes, emphasizing the importance of tailoring the choice of measure when investigating connections between SES and specific outcomes [6]. Among the three SES measures at birth, parental education consistently has the strongest effect regardless of the outcomes. Parental occupation and area household income explained about 14.3% of the mortality risk related to parental education. Previous research on the mediating effect of occupation on education has primarily focused on adult populations, with reported PM ranging from 11.0% to 49.9% [35–39]. The modest mediating effects of parental occupation and area household income on the relationship between parental education and childhood mortality- and the absence of such effects on childhood morbidities- observed in our study could be due to how SES impacts child health differently from adult health. Our recommendation for future research on childhood health that seeks to control for or examine the influence of socioeconomic background is to apply a multidimensional SES perspective which incorporates multiple and distinct resource indicators. Additionally, employing advanced statistical models that enhance the causal interpretation is warranted for a deeper understanding of both the direct and indirect effects of each SES dimension as well as the overall SES-outcome relationship.

Limitations of our study include the absence of household-level income data [40] and the use of census tract-level median household income as the sole area-based SES measure. Using a variety of area-based measures of education, poverty, and occupation in addition to income or a summary score of area socioeconomic environment to characterize neighborhood would help increase our understanding of the impact of childhood residency on health outcomes [41, 42]. Another limitation is the lack of adjustments for place-based built, social, and natural environments, particularly applicable to childhood asthma [43]. However, given the interrelations between multiple area-based measures [44] and the long study period, standardizing multiple SES measures or developing a valid summary score harmonized over three decades is outside the scope of this study.

The study offers several advantages and strengths. Our approach of examining SES takes advantage of the publicly available harmonized SES measures, which is useful in longitudinal research due to its simplicity and ease of interpretability. In addition, our cohort is taken from a very large and comprehensive population database that includes all persons in Utah. Thus, we are able to control for important confounding and mediating factors while assessing the role of SES at birth on childhood health.

In conclusion, our study introduced an approach to standardize individual-level and neighborhood-level SES measures from administrative data to support longitudinal analyses and confirmed the importance of choosing SES measures attentively when analyzing childhood health. We aim to investigate potential variations in the correlations between individual-level and neighborhood-level SES measures over time in future research. In addition, we encourage the routine collection of socioeconomic data in state and federal vital statistics to monitor and reduce social inequalities in health.

Supplementary MaterialAcknowledgements and Funding:

We thank the Pedigree and Population Resource of Huntsman Cancer Institute, University of Utah (funded in part by the Huntsman Cancer Foundation) for its role in the ongoing collection, maintenance, and support of the Utah Population Database (UPDB). We also acknowledge partial support for the UPDB through grant P30 CA2014 from the National Cancer Institute, University of Utah and from the University of Utah’s program in Personalized Health and Center for Clinical and Translational Science. We would like to thank our team of coders who made this project possible: Diana Lane Reed, Darlene Evans, Derek Voeller, Jillian Newman, and Cassandra Palmer.

Research was supported by the NCRR grant, “Sharing Statewide Health Data for Genetic Research” (R01 RR021746, G. Mineau, PI), the NIA grant, “Early Life Conditions, Survival, and Health: A Pedigree-Based Population Study” (R01 AG022095, K.R. Smith PI) with additional support from the Utah Department of Health and the University of Utah.

This study was also supported by the National Institute of Occupational Safety and Health of the National Institutes of Health under award number T24OH008414 (H. Meeks, PI)

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 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.

Declaration of Competing Interest:

None declared

List of Abbreviations and Acronyms:ACS

American Community Survey

CI

Confidence Interval

FTT

Failure-to-thrive

ICD

International Classification of Disease

HR

Hazard Ratio

NPBOSS

Nam-Power-Boyd Occupational Status Score

PM

Proportion Mediated

SD

Standard Deviation

SES

Socioeconomic Status

UPDB

Utah Population Database

References:CaseyJA, , Using Electronic Health Records for Population Health Research: A Review of Methods and Applications. Annual Review of Public Health, Vol 37, 2016. 37: p. 61–81.SchinasiLH, , Using electronic health record data for environmental and place based population health research: a systematic review. Annals of Epidemiology, 2018. 28(7): p. 493–502.29628285 De LepeleereS, , Parenting Practices as a Mediator in the Association Between Family Socio-Economic Status and Screen-Time in Primary Schoolchildren: A Feel4Diabetes Study. International Journal of Environmental Research and Public Health, 2018. 15(11).FlouriE, MavroveliS, and TzavidisN, Modeling risks: effects of area deprivation, family socio-economic disadvantage and adverse life events on young children’s psychopathology. Social Psychiatry and Psychiatric Epidemiology, 2010. 45(6): p. 611–619.19629362 Molina-GarciaJ, , Neighborhood built environment and socio-economic status in relation to multiple health outcomes in adolescents. Preventive Medicine, 2017. 105: p. 88–94.28863871 KachmarAG, , Socioeconomic Status in Pediatric Health Research: A Scoping Review. Journal of Pediatrics, 2019. 213: p. 163–170.31300308 BuajittiE, ChiodoS, and RosellaLC, Agreement between area- and individual-level income measures in a population-based cohort: Implications for population health research. Ssm- Population Health, 2020. 10.GeyerS, , Education, income, and occupational class cannot be used interchangeably in social epidemiology.: Empirical evidence against a common practice. Journal of Epidemiology and Community Health, 2006. 60(9): p. 804–810.16905727 MossJL, , Comparisons of individual- and area-level socioeconomic status as proxies for individual-level measures: evidence from the Mortality Disparities in American Communities study. Population Health Metrics, 2021. 19(1).MillerM, SaldarriagaEM, and Jones-SmithJC, Household socioeconomic status modifies the association between neighborhood SES and obesity in a nationally representative sample of first grade children in the United States. Preventive Medicine Reports, 2020. 20.Pardo-CrespoMR, , Comparison of individual-level versus area-level socioeconomic measures in assessing health outcomes of children in Olmsted County, Minnesota. Journal of Epidemiology and Community Health, 2013. 67(4): p. 305–310.23322850 DunlopAL, , Racial and geographic variation in effects of maternal education and neighborhood-level measures of socioeconomic status on gestational age at birth: Findings from the ECHO cohorts (vol 16, e0245064, 2021). Plos One, 2022. 17(5).ElovainioM, , Socioeconomic status and the development of depressive symptoms from childhood to adulthood: A longitudinal analysis across 27 years of follow-up in the Young Finns study. Social Science & Medicine, 2012. 74(6): p. 923–929.22305468 RosenthalSS, Old homes, externalities, and poor neighborhoods. A model of urban decline and renewal. Journal of Urban Economics, 2008. 63(3): p. 816–840.SongX and XieY, Occupational Percentile Rank: A New Method for Constructing a Socioeconomic Index of Occupational Status. Sociological Methods & Research, 2023.OakesJM and RossiPH, The measurement of SES in health research: current practice and steps toward a new approach. Social Science & Medicine, 2003. 56(4): p. 769–784.12560010 PoulainT, , Associations Between Socio-Economic Status and Child Health: Findings of a Large German Cohort Study. International Journal of Environmental Research and Public Health, 2019. 16(5).SmithKR, , The Utah Population Database. A model for linking medical and genealogical records for population health research. Historical Life Course Studies, 2022. 12: p. 58–77.KozyrskyjAL, , Association Between Socioeconomic Status and the Development of Asthma: Analyses of Income Trajectories. American Journal of Public Health, 2010. 100(3): p. 540–546.19696386 LeeWS, , The relationship between childhood asthma and socioeconomic status: a Korean nationwide population-based study. Frontiers in Public Health, 2023. 11.TangMN, , Failure to Thrive or Growth Faltering: Medical, Developmental/Behavioral, Nutritional, and Social Dimensions. Pediatrics in Review, 2021. 42(11): p. 590–603.34725219 KelaherM, , The impact of different measures of socioeconomic position on the relationship between ethnicity and health. Annals of Epidemiology, 2008. 18(5): p. 351–356.18346910 VanderWeeleTJ and TchetgenEJT, Mediation analysis with time varying exposures and mediators. Journal of the Royal Statistical Society Series B-Statistical Methodology, 2017. 79(3): p. 917–938.28824285 WickhamHA, M.; BryanJ; ChangW; McGowanL, Welcome to the Tidyverse. Journal of Open Source Software, 2017. 4(43): p. 1686.ShiBY, , CMAverse: A Suite of Functions for Reproducible Causal Mediation Analyses. Epidemiology, 2021. 32(5): p. E20–E22.34028370 AkogluH, User’s guide to correlation coefficients. Turk J Emerg Med, 2018. 18(3): p. 91–93.30191186 FestinK, , Choice of measure matters: A study of the relationship between socioeconomic status and psychosocial resources in a middle-aged normal population. Plos One, 2017. 12(8).BukodiE and GoldthorpeJH, Decomposing ‘Social Origins’: The Effects of Parents’ Class, Status, and Education on the Educational Attainment of Their Children. European Sociological Review, 2013. 29(5): p. 1024–1039.ErolaJ, JalonenS, and LehtiH, Parental education, class and income over early life course and children’s achievement. Research in Social Stratification and Mobility, 2016. 44: p. 33–43.CloustonSAP and LinkBG, A Retrospective on Fundamental Cause Theory: State of the Literature and Goals for the Future. Annual Review of Sociology, Vol 47, 2021, 2021. 47: p. 131–156.BalajM, , Parental education and inequalities in child mortality: a global systematic review and meta-analysis. Lancet, 2021. 398(10300): p. 608–620.34119000 LiXJ, SundquistK, and SundquistJ, Parental Occupation and Risk of Hospitalization for Asthma in Children and Adolescents. Journal of Asthma, 2009. 46(8): p. 815–821.SchulerA, , Assessment of Value of Neighborhood Socioeconomic Status in Models That Use Electronic Health Record Data to Predict Health Care Use Rates and Mortality. Jama Network Open, 2020. 3(10).BrownAF, , Structural Interventions to Reduce and Eliminate Health Disparities. American Journal of Public Health, 2019. 109: p. S72–S78.30699019 ElsenburgLK, , Mediation of the parental education gradient in early adult mortality by childhood adversity: a population-based cohort study of more than 1 million children. Lancet Public Health, 2022. 7(2): p. E146–E155.35122758 FujishiroK, , The Role of Occupation in Explaining Cognitive Functioning in Later Life: Education and Occupational Complexity in a US National Sample of Black and White Men and Women. Journals of Gerontology Series B-Psychological Sciences and Social Sciences, 2019. 74(7): p. 1189–1199.28958077 LaineJE, , Reducing socio-economic inequalities in all-cause mortality: a counterfactual mediation approach (vol 49, pg 497, 2020). International Journal of Epidemiology, 2020. 49(2): p. 707–707.32048712 LuJP, , Educational inequalities in mortality and their mediators among generations across four decades: nationwide, population based, prospective cohort study based on the ChinaHEART project. Bmj-British Medical Journal, 2023. 382.SheikhMA, AbelsenB, and OlsenJA, Education and health and well-being: direct and indirect effects with multiple mediators and interactions with multiple imputed data in Stata. Journal of Epidemiology and Community Health, 2017. 71(11): p. 1037–1045.28900028 HanleyGE and MorganS, On the validity of area-based income measures to proxy household income. Bmc Health Services Research, 2008. 8.ForerB, , The Canadian Neighbourhood Early Childhood Development (CanNECD) Socioeconomic Index: Stability and Measurement Invariance Over Time. International Journal of Population Data Science, 2022. 7(3).Putnam-HornsteinE, PrindleJJ, and RebbeR, Community disadvantage, family socioeconomic status, and racial/ethnic differences in maltreatment reporting risk during infancy. Child Abuse & Neglect, 2022. 130.CurtisDS, , Social inequities in neighborhood health amenities over time in the Wasatch Front Region of Utah: Historical inequities, population selection, or differential investment? Cities, 2024. 145.Diez-RouxAV, , Area characteristics and individual-level socioeconomic position indicators in three population-based epidemiologic studies. Ann Epidemiol, 2001. 11(6): p. 395–405.11454499 BoydM and NamCB, The Nam-Powers-Boyd Occupational Scale for the OCC2010 Classification in the IPUMS American Community Surveys. 2016: www.npb-ses.info.

Spearman Correlation matrix for measures of parental socioeconomic status.

Mediation of the associations between parental education and childhood health outcomes by parental occupation and area household income including exposure-mediator interaction.

Abbreviation: PM: Proportion Mediated, HR: Hazard Ratio, CI: Confidence Interval

Note:

1. Single mediation models were performed for each mediator, along with a multiple mediation model incorporating both mediators. Each mediation analysis included two components: a Cox regression model estimating the direct effect of parental education adjusting for mediator(s), and a linear regression model assessing the association between parental education and the mediator(s). All models were additionally adjusted for child’s birth year, sex, race (whether Caucasians), ethnicity (whether Hispanic), birth order (whether first born), and maternal characteristics, which include her residential county classification (whether urban, rural, or frontier), age at birth, and marital status at time of birth (whether married).

2. The indirect effect is the effect between exposure (parental education) and outcome (childhood mortality, failure to thrive, and childhood asthma) caused by the mediating variable(s) (parental occupation and area household income). The direct effect is the effect of the exposure (parental education) on the outcome (childhood mortality, failure to thrive, and childhood asthma) absent of the mediator(s). The total effect is the sum of the indirect effect and the direct effect. Proportion mediated is how large part of the total effect that is explained by the indirect effect.

Sociodemographic characteristics as reported in birth certificates of analysis cohort

AllChildhood mortalityFailure-to-thriveChildhood asthma
N819,1088,13425,28135,594
Age of outcome occurrence-3.5 (5.7)0.1 (0.2)8.2 (5.6)
  Median (IQR)-0.3 (0, 4.3)0 (0, 0)7.3 (2.8, 13.5)
Demographic characteristics
Male420,733 (51.4%)4,848 (59.6%)13,468 (53.3%)21,011 (59.0%)
Birth year1995.3 (9.3)1992.2 (9.6)2004.1 (3.5)1999.0 (6.4)
  Median (IQR)1997 (1986, 2003)1993 (1983, 2001)2005 (2002, 2007)2000 (1996, 2004)
Whether Caucasian
  Yes713,243 (87.1%)7,602 (93.5%)21,484 (85.0%)31,820 (89.4%)
  No36,344 (4.4%)453 (5.6%)1,265 (5.0%)2,517 (7.1%)
  Unknown69,521 (8.5%)79 (1.0%)2,532 (10.0%)1,257 (3.5%)
Whether Hispanic
  No565,184 (69.0%)7,273 (89.4%)17,703 (70.0%)26,943 (75.7%)
  Yes86,994 (10.6%)731 (9.0%)4,001 (15.8%)5,429 (15.2%)
  Unknown166,930 (20.4%)130 (1.6%)3,577 (14.1%)3,222 (9.0%)
Whether first-born260,814 (31.8%)2,487 (30.6%)9,857 (39.0%)11,533 (32.4%)
Maternal residence county classification1
  Urban698,354 (85.3%)6,903 (84.9%)20,912 (82.7%)29,698 (83.4%)
  Rural108,530 (13.2%)1,110 (13.6%)3,964 (15.7%)5,197 (14.6%)
  Frontier12,224 (1.5%)121 (1.5%)405 (1.6%)699 (2.0%)
Whether born to single mothers54,432 (6.6%)676 (8.3%)3,267 (12.9%)4,557 (12.8%)
Maternal age27.3 (5.3)26.9 (5.7)27.6 (5.4)26.9 (5.4)
  Median (IQR)26.7 (23.4, 30.8)26.2 (22.7, 30.6)27.0 (23.6, 31.1)26.3 (22.9, 30.4)
Socioeconomic characteristics
Paternal education attainment
  Less than high school72,015 (8.8%)970 (11.9%)3,180 (12.6%)4,855 (13.6%)
  High school graduate229,042 (28.0%)2,672 (32.8%)7,360 (29.1%)12,471 (35.0%)
  Some college236,087 (28.8%)2,305 (28.3%)6,631 (26.2%)9,348 (26.3%)
  College graduate149,301 (18.2%)1,145 (14.1%)4,227 (16.7%)4,649 (13.1%)
  Post college132,663 (16.2%)1,042 (12.8%)3,883 (15.4%)4,271 (12.0%)
Maternal education attainment
  Less than high school82,645 (10.1%)1,211 (14.9%)3,283 (13.0%)5,438 (15.3%)
  High school graduate264,119 (32.2%)2,985 (36.7%)7,821 (30.9%)13,085 (36.8%)
  Some college270,796 (33.1%)2,418 (29.7%)7,416 (29.3%)10,319 (29.0%)
  College graduate141,380 (17.3%)1,065 (13.1%)4,369 (17.3%)4,376 (12.3%)
  Post college60,168 (7.3%)455 (5.6%)2,392 (9.5%)2,376 (6.7%)
Parental education attainment
  Less than high school42,140 (5.1%)591 (7.3%)1,952 (7.7%)2,781 (7.8%)
  High school graduate195,759 (23.9%)2,445 (30.1%)6,388 (25.3%)11,281 (31.7%)
  Some college240,439 (29.4%)2,421 (29.8%)6,708 (26.5%)10,318 (29.0%)
  College graduate180,610 (22.0%)1,403 (17.2%)5,161 (20.4%)5,732 (16.1%)
  Post college160,160 (19.6%)1,274 (15.7%)5,072 (20.1%)5,482 (15.4%)
Parental occupational status
  Both unemployed26,306 (3.2%)296 (3.6%)749 (3.0%)948 (2.7%)
  Only mother employed26,609 (3.2%)276 (3.4%)784 (3.1%)838 (2.4%)
  Only father employed364,784 (44.5%)3,868 (47.6%)10,638 (42.1%)14,134 (39.7%)
  Both employed401,409 (49.0%)3,694 (45.4%)13,110 (51.9%)19,674 (55.3%)
Parental occupation2
  Unemployed26,306 (3.2%)296 (3.6%)749 (3.0%)948 (2.7%)
  NPBOSS Quartile 1 (1-38)210,978 (25.8%)2,655 (32.6%)7,196 (28.5%)10,865 (30.5%)
  NPBOSS Quartile 2 (39-61)185,804 (22.7%)1,920 (23.6%)5,101 (20.2%)8,493 (23.8%)
  NPBOSS Quartile 3 (62-85)246,942 (30.1%)2,097 (25.8%)7,595 (30.0%)9,699 (27.2%)
  NPBOSS Quartile 4 (86-100)149,078 (18.2%)1,166 (14.3%)4,640 (18.4%)5,589 (15.7%)
Area household income3
  Quartile 1204,856 (25.0%)2,256 (27.7%)6,909 (27.3%)9,397 (26.4%)
  Quartile 2204,190 (24.9%)2,153 (26.5%)6,107 (24.2%)9,319 (26.2%)
  Quartile 3205,257 (25.1%)1,961 (24.1%)6,145 (24.4%)8,817 (24.8%)
  Quartile 4204,805 (25.0%)1,764 (21.7%)6,120 (24.2%)8,061 (22.6%)

Classified by Utah Department of Health and Human Services using 2018 data. Urban counties have a population density of > 100 people per sq. mile; Rural counties have a population density of <99 and > 6 people per sq. mile and Frontier counties have < 6 people per sq. mile. (https://ruralhealth.health.utah.gov/portal/county-classifications-map/)

Parental occupation based on parental occupational status and Nam-Power-Boyd occupational status score (NPBOSS). Detailed on how this variable was derived can be found in the Method and Appendix sections.

Area household income based on maternal residential addresses (or mailing addresses if residential addresses are missing). Detailed on how this variable was derived can be found in the Method and Appendix sections.

Note: N (%) was reported for categorical variables; Mean (Standard Deviation [SD]) and Median (Interquartile Range [IQR]) was reported were continuous variables.

Associations between parental socioeconomic status (SES) measures as continuous variables with childhood health outcomes.

Hazard Ratio (95% Confidence Interval)
Model 1Model 2Model 3Model 4Model 5Model 6Model 7
Childhood mortality
Parental education0.87 (0.85, 0.89)*--0.90 (0.88, 0.92)*0.88 (0.86, 0.90)*-0.90 (0.88, 0.92)*
Parental occupation-0.89 (0.97, 0.91)*-0.92 (0.90, 0.94)*-0.90 (0.88, 0.91)*0.93 (0.91, 0.95)*
Area household income--0.92 (0.90, 0.94)*-0.93 (0.91, 0.95)*0.94 (0.92, 0.96)*0.94 (0.92, 0.96)*
 
Failure to thrive
Parental education0.93 (0.92, 0.94)*--0.94 (0.92, 0.95)*0.93 (0.92, 0.94)*-0.94 (0.92, 0.95)*
Parental occupation-0.96 (0.95, 0.97)*-0.98 (0.97, 0.99)#-0.96 (0.95, 0.97)*0.98 (0.97, 1.00)≠
Area household income--0.97 (0.96, 0.98)*-0.98 (0.97, 0.99)*0.98 (0.97, 0.99)*0.98 (0.97, 0.99)*
 
Childhood asthma
Parental education0.87 (0.86, 0.88)*--0.87 (0.86, 0.88)*0.87 (0.86, 0.88)*-0.87 (0.86, 0.88)*
Parental occupation-0.95 (0.94, 0.96)*-1.00 (0.99, 1.01)-0.95 (0.94, 0.96)*1.00 (0.99, 1.01)
Area household income--0.97 (0.97, 0.98)*-0.99 (0.98, 1.00)#0.98 (0.97, 0.99)*0.99 (0.98, 1.00)#

denotes p-value < 0.001;

denotes p-value < 0.01;

denotes p-value < 0.05

Note: All Cox-proportional hazard models were additionally adjusted for child’s birth year, sex, race (whether Caucasians), ethnicity (whether Hispanic), birth order (whether first born), and maternal characteristics, which include her residential county classification (whether urban, rural, or frontier), age at birth, and marital status at time of birth (whether married) and fitted with the sandwich estimator to account for clustering induced by multiple births within families. Models 1-3 adjusted for individual SES measure, models 4-6 adjusted for pairs of SES measures, and model 7 included all three SES measures simultaneously.

Associations between parental socioeconomic status (SES) measures as categorical variables with childhood health outcomes.

Childhood mortalityFailure to thriveChildhood asthma
HR (95% CI) p-value HR (95% CI) p-value HR (95% CI) p-value
Parental education
  Less than high school1.23 (1.12, 1.35)<0.0011.12 (1.06, 1.18)<0.0011.00 (0.96, 1.05)0.860
  High school graduate (ref)1.001.001.00
  Some college0.86 (0.81, 0.91)<0.0010.92 (0.89, 0.95)<0.0010.83 (0.81, 0.85)<0.001
  College graduate0.75 (0.69, 0.80)<0.0010.82 (0.78, 0.85)<0.0010.63 (0.61, 0.65)<0.001
  Post college0.79 (0.73, 0.85)<0.0010.85 (0.81, 0.89)<0.0010.67 (0.65, 0.70)<0.001
Parental occupation
  Both parents unemployed1.29 (1.13, 1.47)<0.0011.00 (0.92, 1.08)0.9170.95 (0.89, 1.02)0.167
  NPBOSS Quartile 1 (1-38)1.22 (1.13, 1.32)<0.0011.06 (1.02, 1.11)0.0070.98 (0.95, 1.02)0.384
  NPBOSS Quartile 2 (39-61)1.04 (0.96, 1.13)0.2810.98 (0.94, 1.03)0.4001.01 (0.98, 1.05)0.484
  NPBOSS Quartile 3 (62-85)1.00 (0.93, 1.08)0.9141.01 (0.97, 1.05)0.7090.97 (0.94, 1.01)0.131
  NPBOSS Quartile 4 (86-100) (ref)1.001.001.00
Area household income
  Quartile 11.20 (1.12, 1.28)<0.0011.06 (1.03, 1.10)0.0011.04 (1.01, 1.08)0.012
  Quartile 21.12 (1.05, 1.19)<0.0010.98 (0.95, 1.02)0.4231.04 (1.01, 1.07)0.013
  Quartile 31.05 (0.98, 1.12)0.1470.99 (0.95, 1.02)0.4241.01 (0.98, 1.04)0.608
  Quartile 4 (ref)1.001.001.00

Abbreviation: Nam-Power-Boyd Occupational Status Scores (NPBOSS), HR: Hazard Ratio, CI: Confidence Interval

Note: All Cox-proportional hazard models were additionally adjusted for child’s birth year, sex, race (whether Caucasians), ethnicity (whether Hispanic), birth order (whether first born), and maternal characteristics, which include her residential county classification (whether urban, rural, or frontier), age at birth, and marital status at time of birth (whether married) and fitted with the sandwich estimator to account for clustering induced by multiple births within families.

Highlights:

Introduce approaches to standardize multiple SES measures from administrative data across multiple decades to support longitudinal analyses.

Individual-level and neighborhood-level SES measures, though correlated, are distinct indicators of social standing.

The impact of SES measures on childhood health depends both on the outcomes and the specific SES measures.

Parental occupation and area household income mediate the effect of parental education on childhood mortality but not failure-to-thrive and childhood asthma.