Validity of Impairment Ratings Made from Medical Records and from Personal Knowledge

The extent to which medical records supply reliable and valid data about a patient's condition is unknown. Recent emphasis on evaluation of quality of care by record review makes this an important issue. One hundred and three male patients (mean age 65) discharged from hospital to extended care were studied. At hospital discharge, the patient was rated by his physician on a 13-item impairment scale. Another physician, using only medical records, rated the same patients on the same scale. Similarity between ratings was found to be low. Validity was studied by following patients six months later to classify them as living (75) or dead (28). Each set of ratings was then used to predict death, on the assumption that degree of impairment would be related to subsequent death. Both sources predicted death at a highly significant level; however, slightly different predictors were elicited depending on whether records or personal knowledge were used. Common predictors were respiratory, hepatic, and neurologic impairment, with in person ratings adding upper gastrointestinal and record ratings adding vascular and psychiatric impairment. What might be concluded is that both sources were valid, but both incomplete. Records had missing data, and physicians using personal knowledge tended to think only of the patient's treated condition.