Reliability and changes in validity of self-reported cardiovascular disease risk factors using dual response: The behavioral risk factor survey

Abstract

The authors previously studied the validity of self-reported cardiovascular disease (CVD) risk factors assessed by telephone surveys, and found the validity low, especially for self-reported hypertension and hypercholesterolemia. One way to improve validity is to combine repeated measurements (dual response) into a single measure. The authors explored this and the reliability of self-reported CVD data collected by the Behavioral Risk Factor Survey in three New York counties from January 1989 to May 1990. Nine hundred and eleven subjects were interviewed by telephone to collect CVD risk factor and health behavior information. Interviewees were offered physical examination and laboratory testing to verify self-reported CVD risk factors; 628 participated. Subjects were also reinterviewed to assess the test-retest reliability of the survey, and to study how validity of self-reported CVD data changes by dual response. Reliability coefficients for CVD risk factors, preventive health practices, and knowledge of risk factor levels ranged from 0.42 to 0.99. Minimal improvement in sensitivity of self-reported risk factors was found using dual response, and it did not improve specificity. Also, for prevalence of risk factors, dual response minimally improved self-reported rates compared to objective estimates. Combining self-reported measurements causes minimal changes in the validity of these variables. Physiological assessment for hypertension and hypercholesterolemia, or correction for misclassification, is needed for valid individual measurement and for community prevalence estimates from telephone surveys. Self-reported cigarette smoking, obesity, and diabetes mellitus have better validity, but physiological assessment or correction for misclassification may supplement these self-reported risk factors.