Defining paediatric abusive head trauma using routinely collected patient datasets: Scoping review and testing in a...

Suboptimal coding and underreporting of abusive head trauma (AHT) makes cases difficult to identify from routinely collected patient data, particularly if not admitted to hospital.

To identify existing AHT definitions for routine patient data and compare their impact on AHT incidence estimates when applied to a UK primary care cohort.

Cohort participants were all infants in a large primary care database, who were under 1 year old between January 2013 and June 2021, registered at a General Practice, with linkages to secondary care and Accident and Emergency data.

A scoping review to identify studies that used diagnostic codes or routinely collected data to define AHT in children under 5 years from electronic health records. We used the definitions identified to calculate incidence per 100,000 person-years in a UK primary care cohort to explore variability of incidence estimates.

We identified 40 eligible studies (29 definitions), with most definitions based on hospital data using ICD-9-CM and ICD-10 codes. None applied to a UK population or primary care data sources. Incidence of AHT in our cohort ranged from 9 to 15 per 100,000 person-years using "probable" or "presumptive" definitions, and up to 49 per 100,000 person-years using broader definitions including "possible" AHT.

The wide range of hospital AHT incidence calculated in our cohort using the definitions identified in the scoping review indicates a need for a standardised definition suited to UK datasets including primary care.