Defining paediatric abusive head trauma using routinely collected patient datasets: Definition development for a UK primary care cohort

Background
Existing definitions used to identify abusive head trauma (AHT) from routinely collected patient data are developed for use in secondary care or mortality data. Incidence estimates calculated using these definitions do not include mildly symptomatic cases of AHT presenting to primary care, therefore underestimating the true population incidence.

Objective
To (1) develop AHT definitions for use with primary care, secondary care and accident and emergency data in the UK setting; (2) compare incidence using these definitions in a UK primary care cohort.

Participants and setting
The cohort included all infants in Clinical Practice Research Database (CPRD) Aurum, under 1 year old between January 2013 and June 2021, with linkages to Hospital Episodes Statistics Admitted Patient Care (HES APC), Accident and Emergency (HES AE) data and Diagnostic Imaging Dataset (DID).

Method
This study had two parts, (1) definition development via clinical expert consensus, and (2) comparative testing of definitions in a UK primary care cohort.

Results
Definitions of AHT were developed based on combinations of traumatic brain injury, abuse and head computerised tomography codes present in UK health records. Our definitions for AHT presenting to general practice, hospital and ED identified higher incidence in our cohort than all ICD-10 based definitions identified in the literature.

Conclusion
Inclusion of injuries presenting to GP, ED or admitted to hospital, in our definitions is likely to produce a more accurate estimate of AHT in the population. Caution should be applied, however, regarding potential inclusion of false positives and missed cases due to non-presentation to healthcare.