Could socioeconomic inequity in bronchiolitis admissions be reduced through intervening on gestational age? A principled analysis of cohort data

Abstract

BackgroundAbout 4% of infants in England are admitted to hospital for bronchiolitis, an acute lung infection, and the rates are unequal among socioeconomic groups. However, the causal pathways of this inequity are yet to be determined. We aimed to establish the extent to which socioeconomic inequity in hospital admissions for bronchiolitis would be reduced if disparities in gestational age were eliminated. MethodsHospital episode statistics were used to create a cohort of all singleton children born in English National Health Service hospitals between 2011 and 2016. Infants were followed up until their first birthday, and all emergency admissions to hospital for bronchiolitis (identified in hospital records by ICD-10 code J21) were linked to birth records. The index of multiple deprivation 2010, a neighbourhood measure of relative deprivation, was used to represent socioeconomic deprivation (high or low). We focused on estimating the randomised interventional analogue indirect effect of gestational age as the counterfactual difference in bronchiolitis admissions rates if children born into the high deprivation group had the same distribution of gestational age of children from the low deprivation group. Unlike traditional mediational analysis methods this recent generalisation does not require a continuous outcome or linearity between variables. Estimation was by g-computation, controlling for a-priori confounders of the mediator and outcome association—presence of congenital anomalies (yes or no) and infant sex. Ethics approval was not required since all data were de-identifiable. FindingsThe cohort consisted of 3 948 819 children (51·7% male). Between 2012 and 2016, 155 607 (60·6% male) were admitted to hospital for bronchiolitis (45·0 admissions per 1000 person-years, 95% CI 45·0–45·0). With the assumption of no unmeasured confounding, if socioeconomic disparities could be removed the rate of admissions would decrease by 9·4 per 1000 person-years (95% CI 9·0–9·8); if instead disparities remained but the distribution of gestational age could be equalised across deprivation groups, 12·0% (11·3–12·7) of this effect could be removed. InterpretationThere is some evidence that, in England, interventions aimed at ensuring infants are born at term might help reduce a small proportion of inequity in bronchiolitis morbidity. Future research should explore the contribution of other likely mediators, including birthweight, tobacco smoke exposure, and housing conditions. FundingMedical Research Council (grant number MR/N013867/1) (for KML).