Accuracy of identifcation of traumatic brain injuries in routine hospital data

This study assessed the performance of an Australian classification system (Pozzatto et al 2019), using a standardised approach to ICD-10 codes to identify cases of likely TBI in routine hospital discharge data.MethodsThe original study was done on hospital data from New South Wales. We replicated their approach using Irish hospital data, held by Health Intelligence, from 2013 to 2020. Cases not classified as TBI by this system, but with codes, such as loss-of-consciousness, skull fracture or intra-cranial injury were manually reviewed.ResultsAll 98,419 discharges with any code in S00 to S99 were reviewed. 27,851 (28.3%) had a skull fracture or intracranial injury. 12,106 (12.3%) had loss-of-consciousness and/or post-traumatic amnesia. 11,976 (98.9%) of these (12.2% of the total) had either a skull fracture or an intra-cranial injury reported. 26,085 (26.5%) of the original 98,419 cases were classified as TBI using the NSW classification. Manual review of 1.3% (1,356) cases added a maximum of 0.32% (321) further possible cases of TBI, suggesting a sensitivity of the classification of 98.8% (95% CI 98.6% - 98.9%).DiscussionThe main limitation is that it is not possible to identify false positive cases - those coded as TBI, but where no TBI was present. This approach to identifying TBI works well, and is feasible for wider implementation. It provides comparability between different studies.Pozzato I et al. (2019), Epidemiology of hospitalised traumatic brain injury in the state...
Source: The European Journal of Public Health - Category: General Medicine Source Type: research