Adverse Childhood Experiences (ACEs) in English Electronic Health Records of Linked Mothers and Children: Validation Study Using a Multistage Risk-Prediction Model

Abstract

Background: Electronic health records (EHRs) are readily available and increasingly used for identifying adverse childhood experience (ACEs), yet well-developed indicators for ACEs remain limited. We aimed to develop clinically relevant indicators of ACEs for routine EHRs of mothers and children using multi-stage risk prediction models of child maltreatment (CM) and maternal intimate partner violence (mIPV). Methods: We developed a representative population-based birth cohort of 211 ·393 linked mother-child pairs (422·786 total patients) followed from 2-years pre-birth up to 5-years post-birth (2002-2018) across the Clinical Practice Research Datalink (primary care), Hospital Episode Statistics (secondary care), and the Office for National Statistics mortality register. We randomly assigned the cohort into a derivation cohort (2/3 of sample) and a validation cohort by GP practice. The primary outcome (reference standard) was any CM or mIPV in the mother's or child's record from 2-years pre-birth (mIPV only) up to 5-years post-birth. We used 8 prediction models combined with expert ratings to systematically developed indicators. We validated the final indicators by integrating results from machine learning models, survival analyses and clustering analyses in the validation cohort. Findings: We included 63 indicators in six ACE domains: maternal mental health problems (mMHPs), maternal substance misuse (MSM), adverse family environments (AFE), CM, mIPV, and high-risk presentations for CM (HRP-CM) . Excluding the 7 indicators in the reference standard, 56 indicators showed high discriminative validity for family violence (validation cohort, area under the receiver operating characteristic curve: 0·85, 95% CI: 0·84-0·86). The prevalence over the 2-year period before and after birth were 39·1% for any of the 63 ACE indicators, 22·2% for mMHPs, 15·7% for AFE, 8·1% for HRP-CM, 6·9% for MSM, and 3·0% for any CM (2·4%) and mIPV (1·0%). 74·1% of ACEs were recorded only in primary care. Interpretation: We developed and quantified clinically relevant indicators for identifying ACEs using EHRs of mothers and children presenting to health care during pregnancy and early childhood. Details of all ACEs and indicators are publicly available for further evaluation (https:/ACEsinEHRs.com). Funding: None to declare. Declaration of Interest: None to declare. Ethical Approval: The study was approved by the MHRA (UK) Independent Scientific Advisory Committee