Developing a prediction rule from automated clinical databases to identifyhigh-risk patients in a large population with diabetes

OBJECTIVE - To develop and validate a prediction rule for identifying diabe tic patients at high short-term risk of complications using automated data in a large managed care organization. RESEARCH DESIGN AND METHODS - Retrospective cohort analyses were performed in 57,722 diabetic members of Kaiser Permanente, Northern California, aged greater than or equal to 19 years, Data from 1994 to 1995 were used to mode l risk for macro- and microvascular complications (n = 3,977), infectious c omplications (n = 1,580), and metabolic complications (n = 316) during 1996 , Candidate predictors (n = 36) included prior inpatient and outpatient dia gnoses, laboratory records, pharmacy records, utilization records, and surv ey data. Using split-sample validation, the risk scores derived from logist ic regression models in half of the population were evaluated in the second half. Sensitivity, positive predictive value, and receiver operating chara cteristics curves were used to compare scores obtained from full models to those derived using simpler approaches. RESULTS - History of prior complications or related outpatient diagnoses we re the strongest predictors in each complications set. For patients without previous events, treatment with insulin alone, serum creatinine - greater than or equal to 13 mg/dl, use of two or more antihypertensive medications, HbA(1c) > 10%, and albuminuria/microalbuminuria were independent predictor s of two or all three complications. Several risk scores derived from multi variate models were more efficient than simply targeting patients with elev ated HbA(1c) levels for identifying high-risk patients. CONCLUSIONS - Simple prediction rules based on automated clinical data are useful in planning care management for populations with diabetes.