Measuring socioeconomic adversity in early life

Abstract

Aim: Early life adversity leads to enduring effects on physical and mental health, school performance and other outcomes. We sought to identify potentially modifiable factors associated with socioeconomic adversity in early life.

Methods: We enrolled 1503 pregnant women aged 16-40 years, without pregnancy complications or pre-existing conditions from Shelby County, Tennessee. Social, familial and economic variables were analysed using principal components (PCs) analyses to generate the Socioeconomic Adversity Index (SAI). This was replicated using the National Survey of Children's Health (NSCH). Health and social outcomes were compared across the quintile groups defined by SAI values at the county, state and national levels.

Results: Significant differences occurred across the SAI Quintile-1 to Quintile-5 groups in marital status, household structure, annual income, education and health insurance. Significantly worse health and social outcomes occurred in the lower versus higher SAI quintiles, including maternal depression, parental incarceration, child's birthweight and potential for child abuse. Maternal age and race also differed significantly across the SAI quintiles.

Conclusion: Modifiable factors contributing to socioeconomic adversity in early life included marital status, household structure, annual income, education and health insurance. Those exposed to greater socioeconomic adversity as defined by SAI values had significantly worse maternal and child outcomes.

Keywords: Outcomes; Perinatal; Poverty; Pregnancy; Socio-economic status.

Figure 1:

CONSORT diagram for the CANDLE Study, showing the numbers of subjects enrolled and those evaluated at each follow-up visit. CANDLE enrolled 1503 women in the second trimester of pregnancy, with 1455 live births and some attrition in the infant/toddler (n=1241, 82.6%) and preschool (n=1208, 80.4%) groups because of missed clinic visits.

Figure 2:

Heatmap of the variables with absolute correlations of >0.1 with the SAI principal components at the perinatal (3 PCs), infant/toddler (5 PCs), and preschool (4 PCs) periods for the CANDLE cohort. Numbers below the PCs denote proportions of the total variation explained by that corresponding PC.

Figure 3:

Distribution of salient factors assessed at different contact points within CANDLE across the quintile (Q1-Q5) groups assigned by the ranks of their SAI scores; (2A) marital status, (2B) household structure, (2C) maternal education, and (2D) maternal income. Mothers in the lower quintile groups were more often unmarried, had more dependents, received less education, and had lower annual incomes than those in the higher quintile groups.

Figure 4:

Density plots showing the bimodal distribution of SAI values for the Perinatal (blue), Infant/Toddler (red), and Preschool (green) time periods in the CANDLE Study, and unimodal distributions of SAI scores for the US (black) and Tennessee (purple) populations in NSCH.

Figure 5:

Receiver Operating Characteristic (ROC) curves comparing the SAI vs. SES measures in predicting maternal/child outcomes in the US NSCH data, with significant differences for maternal depression (p<0.0001), maternal anxiety (p<0.0001), parental incarceration (p<0.0001), maternal drug abuse (p<0.0001) and potential child abuse (p<0.0001).