The accuracy of health risk appraisal in predicting mortality

In order to determine the accuracy of the Centers for Disease Control/Health Risk Appraisal (CDC/HRA) program, the authors compared observed to predicted mortality for the 3,135 persons followed from 1959-79 as part of the Tecumseh Community Health Study. The analysis was limited to smokers and never-smokers aged 25-60 whose 1959 questionnaires included at least the minimal variables for prediction using the CDC/HRA (age, sex, race, height, weight, and smoking habits). For men and women overall and in each age group, the observed proportion dying over 20 years of follow-up increased as the difference between 1959 age and risk age increased. CDC/HRA predicted 10-year risks of mortality appeared to improve upon age-sex-race predicted risks of mortality when compared to the observed proportion dying over 10 years and when predictors were used in a logistic regression model with vital status after 10 years as the dependent variable. Thus, CDC/HRA may be an appropriate method for identifying high-risk populations for health interventions.