Steps in Abstracting Medical Records
A medical record abstraction can be useful when demographic data are simple, commonly recorded medical data are desired, and when attempts to control for reliability are made. Many studies have attempted to abstract relevant information on hypertensive patients in a family medicine clinic and have illustrated the steps required in conducting abstractions of medical records. Multiple drafts of a form are typically prepared. The first few drafts will be revised to reflect the purposes of the study more clearly, as these purposes became more clearly defined through staff decisions. Validity and reliability analyses will be conducted on the next two drafts by having sets of two abstracters jointly abstract 30 records. Reliability indices can be calculated on the 30 jointly abstracted records for each variable, and each pair of abstracters. All cases in which differing values are obtained need to be reviewed against the medical record. Rules can be generated to refine the search or the recording process. On the second set of reliability abstractions, rules for abstraction can be further refined. Variables for which a reliability of abstraction of > 0.80 or higher (kappa) is not achieved can be dropped. A final draft will then be produced for study.