Predictive value of indicators for identifying child maltreatment and intimate partner violence in coded electronic health records: a systematic review and meta-analysis

Predictive value of indicators for identifying child maltreatment and intimate partner violence in coded electronic health records: a systematic review and meta-analysis

Link to publication: http://dx.doi.org/10.1136/archdischild-2020-319027

Introduction:

Intimate partner violence (IPV) and child maltreatment (CM) are forms of family violence that often go unnoticed by services, despite recommendations to improve monitoring efforts by the World Health Organization (WHO). CM and IPV refer to any act that causes biopsychosocial harm to a child, a future child, or a partner. In the UK, statutory definitions of CM include fetal alcohol syndrome (FAS) and neonatal abstinence syndrome (NAS) due to neglect or harm during pregnancy. Assessing health records for detailed information on family violence is time-consuming and expensive. To address this, studies and services are increasingly using routinely coded electronic health records (EHRs) to assess family violence. Coded EHRs offer longitudinal population-based assessments, automated early warning systems, and identification of high-risk populations at relatively low costs. However, the validity of these coded indicators remains uncertain due to reported quality issues and lack of external validation. This meta-analysis aims to estimate the positive predictive values (PPVs) of coded indicators for different forms of family violence, including CM, prenatal neglect (NAS or FAS), and IPV, based on external independent reference standards.

Methods:

The study followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses for Diagnostic Test Accuracy Studies and the Meta-analysis of Observational Studies in Epidemiology guidelines. The review protocol was published in the PROSPERO registry. The researchers searched 18 electronic databases and 20 selected journals for relevant studies published between 1 January 1970 and 24 May 2020. Three independent reviewers screened abstracts, full texts, and reference lists of articles using Covidence’s systematic review software. Disagreements over study inclusions were resolved by a family violence expert. The included studies provided data to calculate the PPV of a coded EHR indicator for specific family violence outcomes, were conducted in primary care, paediatric units, or general hospital settings, published in English/Swedish/German, and distinguished family violence cases from non-cases.

Results:

The meta-analysis included 65 cross-sectional and 23 longitudinal studies, involving 20 indicators and 3,875,183 individuals from 11 different countries. The pooled PPV of primary diagnoses for NAS was 80.9%, for FAS was 39.3%, for CM (0-18 years) was 87.8%, and for IPV among women (12-50 years) was 86.1%. The pooled PPVs for specific CM indicators ranged from 88.3% for rib fractures to 19.6% for multiple burn injuries in children under 5 years. Injury-related presentations of IPV provided relatively low PPVs. The proportion of misclassifications (false positives) due to coding errors was on average 2.1%. In assault-coded cases among women, 28.0% had no recorded perpetrator information in the underlying medical charts (missing data).

Discussion:

The study highlights the utility of using routinely coded medical data to evaluate services for at-risk groups exposed to family violence. The consistently high PPVs for CM and IPV indicators in EHRs suggest the potential for early identification and support of high-risk individuals. However, estimates varied depending on the indicator and outcome, with substantial heterogeneity. Coded injury patterns could be considered as a broader measure of CM to identify high-risk groups. Still, the study has limitations, including the complexity of identifying CM and IPV in practice and potential missed diagnoses or misclassifications in EHRs.

Implications:

Using EHRs to identify family violence can improve targeted care and support for at-risk individuals. Coded indicators of family violence could be integrated into computerized clinical decision support systems to flag potential at-risk cases. Linking family members’ EHRs allows for a “Think-Family” approach, aiding in identifying vulnerable children through mothers or vice versa. Despite potential benefits, careful consideration of potential harms, such as stigma, legal consequences, trust, and reduced help-seeking, is necessary before implementing automated EHR systems for family violence identification.

Dr Shabeer Syed, Clinical Psychologist & Senior Research Associate

---