MISCLASSIFICATION AND SELECTION BIAS IN CASE-CONTROL STUDIES USING ANAUTOMATED DATABASE

Automated databases are increasingly being used for pharmacoepidemiolo gical research and fieldwork is often carried out to supplement and va lidate information held within them. In MEMO's case-control studies, p atients are identified using computerized ICD9 diagnosis codes, the or iginal medical records are retrieved and checked, and only patients fu lfilling case inclusion criteria are used. The 20-30% of patients for whom medical records cannot be found are usually excluded. The aim is to eliminate misclassification bias. However, selection bias may be in troduced if availability of medical records is associated with exposur e. This investigation was therefore carried out to assess the relative importance of misclassification and selection bias. Data from four pr evious case-control studies, investigating the associations between NS AIDs and hospitalization for colitis, acute renal failure, appendiciti s and colorectal cancer, were used. To assess misclassification, odds ratios (with 95% CI) for recent exposure to NSAIDs were compared in re peated studies that used all patients identified by ICD9 codes, with s tudies using validated cases only. Selection bias was assessed by comp aring results in studies using patients for whom records could and cou ld not be found. Results were plotted and the graphs indicated that mi sclassification bias was relatively unimportant, but that selection bi as could be introduced into a study in this way. (C) 1997 John Wiley & Sons, Ltd.