Methods We used the Clinical Records Interactive Search (CRIS) system hosted at the NIHR Maudsley Biomedical Research Centre to search the clinical records for patients with catatonia. An initial free-text search was refined by use of a natural language processing app. The results of the app were validated by three of the authors, who included patients in the analysis only if a clinician had made a diagnosis of catatonia and two or more items of the Bush-Francis Catatonia Screening Instrument were in evidence. Demographics, disease-related variables and blood-based biomarkers could then be extracted for these patients and compared, where relevant, to non-catatonic psychiatric patients.
Results The natural language processing app extracted the records of 2766 patients with at least one mention of catatonia in their records. The majority of cases identified by the app could be validated by the researchers. A high proportion of patients had more than one episode of catatonia.
Full results will be available in time for the presentation.
Conclusions This study demonstrates that catatonia is not very rare, even relying on clinician identification. The frequency of recurrence is interesting, as it suggests that catatonia might indicate an underlying trait, rather than merely a transient state.