Challenges Associated With Rural‐Urban Stratification for Generalizing Birth Outcomes: Insights From the ECHO Cohort

Résumé fourni par la source

PURPOSE: Efforts to generalize findings from the Environmental influences on Child Health Outcomes (ECHO) Cohort across rural and urban areas are challenged by limitations in both sample composition and the classification schemes used to define place. We evaluated how rural-urban stratification affects the interpretation and generalizability of preterm birth (PTB) prevalence proportions in the ECHO Cohort compared to national benchmarks. METHODS: We used a population data science approach to compare bootstrap estimates of PTB prevalence in ECHO (2017-2019, 2020-2022) to county-level prevalence from the National Center for Health Statistics, stratified by rural-urban classification (RUCC, UIC, NCHS), race/ethnicity, education, and income. We applied post-stratification weights and conducted sensitivity analyses. FINDINGS: Overall PTB prevalence in ECHO was statistically similar to that in US live births. Estimates varied by rural-urban classification scheme but showed no consistent directional difference. Stratifying by race and education revealed variability in PTB differences and gaps in subgroup representation within the analytic sample. Post-stratification increased PTB estimates slightly and stabilized rural estimates. Two predominantly rural cohort sites strongly influenced rural means; excluding one reversed the direction of rural-urban difference while excluding the other increased it. Supplemental analyses showed regional variability in PTB prevalence and suggested that living above 130% of the federal poverty level may be protective. CONCLUSIONS: Rural-urban stratification alone, without accounting for the context of rural places, limits generalizability and may obscure differences between samples drawn from large cohort studies and the broader population. Context-aware stratification may improve validity and equity in population health research.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.