Objectives: Hospital Episode Statistics data are used for healthcare planning and hospital reimbursements. Reliability of these data is dependent on the accuracy of individual hospitals reporting Secondary Use Services (SUS) which includes hospitalisation. The number and coding accuracy for Parkinson's disease hospital admissions at a tertiary centre in Birmingham was assessed.
Study Design: Retrospective, routine-data-based study
Methods: A retrospective electronic database search for all Parkinson's disease patients admitted to the tertiary hospital over a 4 year period (2009-2013) was performed on the SUS database using International Classification of Disease (ICD-10) codes, and on the local inpatient electronic prescription database, Prescription and Information Communications System (PICS), using medication prescriptions. Capture-recapture methods were used to estimate the number of patients and admissions missed by both databases.
Results: From the two databases, between July 2009 and June 2013, 1,068 patients with Parkinson's disease accounted for 1,999 admissions. During these admissions Parkinson’s disease was coded as a primary or secondary diagnosis. 91% of these admissions were recorded on the SUS database. Capture-recapture methods estimated that the number of Parkinson's disease patients admitted during this period was 1,127 patients (95% confidence interval: 1,107 -1,146). A supplementary search of both SUS and PICS was undertaken using the hospital numbers of these 1,068 patients. This identified another 479 admissions. SUS database under-estimated Parkinson's disease admissions by 27% during the study period.
Conclusion: The accuracy of disease coding is critical for healthcare policy planning and must be improved. If the under-reporting of Parkinson's disease admissions on the SUS database is repeated nationally, expenditure on PD admissions in England is under-estimated by approximately £61 million per year.
- Parkinson’s disease
- Hospital Episode Statistics
- service evaluation
- clinical coding