Pleural empyema accuracy on Hospital Episode Statistics (HES): a validation study

Introduction: Increasingly, research is performed on routinely recorded clinical data, such as hospital episode statistics (HES). In uncommon diseases like pleural empyema, this data would be useful in assessing epidemiology. Also, empyema has a significant healthcare tariff and accurate coding ensures appropriate reimbursement. We aimed to assess the accuracy of coding for empyema across three separate UK hospitals. Methods: HES data (J86 code) was extracted for 1 year (Aug 2017 to Aug 2018) from 3 UK hospitals. Cases were verified to assess accuracy using medical notes, radiology, and microbiology. Empyema was defined using BTS criteria. To assess coding sensitivity, data from screened participants in the Pleural Infection Longitudinal Outcome (PILOT) study, for which our hospital was a recruitment centre, was extracted. The coding of these ‘confirmed cases’ was cross checked with hospital data from the same period. Results: Overall, 182 patients with a coded diagnosis of empyema were identified: 124 from a surgical centre. This coding was accurate in 158 cases (86%). Limiting the codes to exclude elective admissions increased the accuracy of coding (to 93%). This also excluded the vast majority (14/21) of empyema secondary to cardiothoracic or upper GI surgery. In patients with a confirmed empyema from the PILOT study, coding was 72% sensitive (18/25). The most common replacement code was ‘effusion otherwise unspecified’. Conclusion: Empyema coding was specific, with >85% of HES codes recorded accurately. However, in a prospectively collected cohort of patients, known clinically to have empyema, coding was around 70% sensitive. Excluding codes for elective admissions removed 2/3 of post-surgical cases.