Validity and completeness of the General Practice Research Database for studies of inflammatory bowel disease

Purpose The objective of this study was to evaluate the validity and completeness of the General Practice Research Database (GPRD) as a tool for research into inflammatory bowel disease epidemiology. Methods Patients diagnosed with inflammatory bowel disease were identified from GPRD. Mailed surveys were sent to the general practitioner caring for a stratified random sample of 170 of these patients and collected information on the diagnosis of inflammatory bowel disease and the most recent surgery and hospitalization. Results Usable surveys were returned for 157 patients (92%). The inflammatory bowel disease diagnosis was highly probable or probable in 144 (92%, 95% CI 86 to 96%). Among the 53 patients with a potentially incident inflammatory bowel disease diagnosis, 33 (62%) had the first recorded diagnosis in GPRD within 30 days of the date reported in the survey (median difference -8 days interquartile range 0 to -81days). Of 12 surgeries and 25 hospitalizations reported in the survey, 11 (92%) and 19 (76%) were identified in GPRD, respectively. Conclusions The diagnosis of inflammatory bowel disease in GPRD appears reliable for most patients. Important medical events such as hospitalizations and surgery are recorded at a high rate, although algorithms to identify these events are complex. Copyright (C) 2002 John Wiley Sons, Ltd.