.J Clin Epidemiol Vol. 46, No. 6, pp. 561-571, 1993 0895-4356193 $6.00 + 0.00 Printed in Great Britain. All rights reserved Copyright 0 1993 Pergamon Press Ltd VALIDITY OF CARDIOVASCULAR DISEASE RISK FACTORS ASSESSED BY TELEPHONE SURVEY: THE BEHAVIORAL RISK FACTOR SURVEY STEVEN J. BowLIN,‘~~* BARBARA D. MORRILL,’ ANNE N. NAFZIGER,~ PAUL L. JENKINS,* CAROL LEWIS* and THOMAS A. F’EARSON*~~ ‘Department of Epidemiology and Biostatistics, School of Medicine, Case Western Reserve University, Cleveland, OH 44106, *Research Institute of the Mary Imogene Bassett Hospital, Cooperstown, NY 13326 and ‘Department of Epidemiology, Columbia University, New York, NY 10027, U.S.A. (Received zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFE in revised form 5 October 1992) Abstract-The Behavioral Risk Factor Surveillance System (BRFSS) collects telephone interview data on behaviors for the leading causes of premature death and disability. Its validity has never been adequately studied. The authors replicated BRFSS method- ology to validate self-reported cardiovascular disease (CVD) risk factors. Nine-hundred and eleven subjects from three upstate New York counties were interviewed between l/89 and 5/90. Interviewees were offered physical examinations and laboratory testing for CVD risk factors; 282 men and 344 women participated. The authors studied validity by comparing objectively measured to self-reported CVD risk factors. Sensitivities for self-reported hypertension, hypercholesterolemia, obesity, smoking, and diabetes were: 43,44, 74, 82 and 75%, respectively. Only smoking sensitivity differed by gender: men, 77%; women, 86%. Specificity was > 85% for all risk factors, except hypercholes- terolemia in men (75%). Prevalence was underreported for hypertension, hypercholes- terolemia, obesity, and smoking by 43, 50, 25 and 17%, respectively. Results suggest telephone survey research includes physiologic measurements for blood pressure, cholesterol, height, weight, and smoking to validate self-reported CVD risk factors. When this is impossible, results such as these can be used, in similar samples, to correct risk factor prevalence rates from telephone surveys for misclassifications. Cardiovascular disease Risk factors Survey methods Telephone Validity INTRODUCTION Collecting health information by telephone surveys is increasingly in demand. Telephone surveys are less costly than in-home inter- views, and generally yield higher participation rates than mail surveys [l]. Assessing the val- idity of telephone survey data is difficult, and involves procedures that increase the cost of *All correspondence should be addressed to: Steven J. Bowlin, D.O., M.P.H., Department of Epidemiology and Biostatistics, School of Medicine, Case Western Reserve University, 2109 Adelbert Road, Cleveland, OH 44106-4945, U.S.A. the research. Validation may include clinic examination of subjects or review of medical records for disease or risk factor confirmation. However, measuring validity is important if the data are to be used confidently. Validation studies of cardiovascular disease risk factors such as smoking, hypertension, hypercholes- terolemia, obesity, diabetes, and physical ac- tivity acquired by telephone survey are rare, as shown by the lack of information in the litera- ture. The Centers for Disease Control’s (CDC) Behavioral Risk Factor Surveillance System (BRFSS), a telephone survey, is the world’s 561