PT - JOURNAL ARTICLE AU - Cleary, R AU - Beard, R AU - Coles, J AU - Devlin, B AU - Hopkins, A AU - Schumacher, D AU - Wickings, I TI - Comparative hospital databases: value for management and quality. AID - 10.1136/qshc.3.1.3 DP - 1994 Mar 01 TA - Quality in Health Care PG - 3--10 VI - 3 IP - 1 4099 - http://qualitysafety.bmj.com/content/3/1/3.short 4100 - http://qualitysafety.bmj.com/content/3/1/3.full SO - Qual Health Care1994 Mar 01; 3 AB - OBJECTIVES--To establish an accurate and reliable comparative database of discharge abstracts and to appraise its value for assessments of quality of care. DESIGN--Retrospective review of case notes by trained research abstractors and comparison with matched information as routinely collected by the hospitals' own information systems. SETTING--Three district general hospitals and two major London teaching hospitals. PATIENTS--The database included 3905 medical and surgical cases and 2082 obstetric cases from 1990 and 1991. MAIN MEASURES--Accessibility of case notes; measures of reliability between reviewers and of validity of case note content; application of high level quality indicators. RESULTS--The existing hospital systems extracted insufficient detail from case notes to conduct clinical comparative analyses for medical and surgical cases. The research abstractors at least doubled the diagnostic codes extracted. Interabstractor agreement of about 70% was obtained for primary diagnosis and assignment to diagnosis related group. These data were sufficient to create a comparative database and apply high level quality indicators designed to flag topics for further study. For obstetric-specific indicators the rates were comparable for abstractors and the hospital information systems, which in each case was a departmentally based system (SMMIS) producing more detailed and accessible data. CONCLUSIONS--Current methods of extracting and coding diagnostic and procedural data from case notes in this sample of hospitals is unsatisfactory: notes were difficult to access and recording is unacceptably incomplete. IMPLICATIONS--Improvements as piloted in this project, are readily available should the NHS, hospital managers, and clinicians see the value of these data in their clinical and managerial activities.