Efficiency of hospital reporting systems in detecting head injury admissions.

Objectives: To examine the completeness of data on admission with head injury at a Regional Neuroscience Centre.

Design: A comparative study using retrospective and prospective data sources.

Subjects: All adults admitted to all specialties with all severities of head injury.

Methods: Adult admissions with 'head injury' in a single month were identified from two sources: (a) prospective ED reports using ED codes for head injury + site 'Head/Face' and 'Reason for attendance', (b) retrospective ICD-10 coding reports of codes S00-09. Data from both sources were compared and reasons for non-capture analysed. Admissions from both reports were combined to obtain a more complete number of admissions with head injury.

Results: A total of 112 admissions with head injury were identified from both sources. Completeness from ED reports was 70% (n = 78) and 83% (n = 93) from ICD-10 reports. In total, 53% (n = 59) of admissions were captured in both reports. The main reasons for non-capture in the ED reports were non-coding of head injury (56%, n = 19), and admission via another routes, for example, transfers (44%, n = 15). The main reasons for non-capture in the ICD-10 reports were non-coding of head injury or not the primary diagnosis (74%, n = 14).

Conclusions: Reliable epidemiological data is required for planning and commissioning services, however, identification of head injury admissions is difficult due to ambiguity in terminology and patient population. More than one source is necessary for completeness of data, however, data issues such as comprehensiveness, reliability and inclusion criteria/possible bias need to be considered.