Assessing inequalities in paediatric emergency department admissions in the Northwest of England: a retrospective cohort study.

Introduction: Prior to the UK's first COVID-19 related lockdown (2020), the number of paediatric emergency department (ED) attendances and admissions from ED had increased. Predictors of admission to a paediatric ED include patient characteristics and reasons for admission. With evidence of a social gradient in paediatric emergency hospital attendances and admissions, the rise in admissions could also reflect socio-economic conditions (SECs). Methods: To investigate the effect of SECs on admission to hospital from ED, we analysed data on all children presenting to the ED at Alder Hey Children's Hospital in Northwest England, between 2016 and 2019 (206 918 attendances). Logistic regression models and generalised additive models assessed the association between SECs (using Index of Multiple Deprivation (IMD)) and risk of admission from the ED, conditional on ED attendance. We further explored the social patterning of other important factors, such as referral source, diagnosis and triage category. Logistic regression models additionally adjusted for these important variables to investigate if these factors helped to explain the association between deprivation and risk of admission from the ED. Results: Admission occurred in approximately 16% of attendances among children living in areas of lower SECs compared with 24% among their most affluent counterparts. In unadjusted analyses, increasing deprivation was associated with lower odds of admission conditional on ED attendance (OR per one-unit increase in IMD 0.996, 95% CI 0.995 to 0.996). After adjustment for demographic, case-mix and clinical presentation variables, the association reversed direction (OR 1.002, 95% CI 1.000 to 1.003). However, in generalised estimating equation sensitivity analyses accounting for repeated attendances, the fully adjusted association attenuated towards the null. Conclusions: Associations between SECs and admission risk conditional on ED attendance were sensitive to adjustment for case-mix and clinical presentation variables and to modelling strategy. Results reflect differences in clinical decision-making, risk tolerance or unmeasured social factors influencing admission decisions.