Body mass index: comparing mean values and prevalence rates from telephoneand examination surveys

Background: Cost effective means of assessing the levels of risk factors in the population have to be defined in order to monitor these factors over t ime and across populations. This study is aimed at analyzing the difference in population estimates of the mean levels of body mass index (BMI) and th e prevalences of overweight, between health examination survey and telephon e survey. Methods: The study compares the results of two health surveys, one by telep hone (N=820) and the other by physical examination (N=1318). The two survey s, based on independent random samples of the population, were carried out over the same period (1992-1993) in the same population (canton of Vaud, Sw itzerland). Results: Overall participation rates were 67% and 53% for the health interv iew survey (HIS) and the health examination survey (HES) respectively. In t he HIS, the reporting rate was over 98% for weight and height values. Self- reported weight was on average lower than measured weight, by 2.2kg in men and 3.5kg in women, while self-reported height was on average greater than measured height, by 1.2cm in men and 1.9cm in women. As a result, in compar ison to HES, HIS led to substantially lower mean levels of BMI, and to a re duction of the prevalence rates of obesity (BMI>30kg/m(2)) by more than a h alf. These differences are larger for women than for men. Conclusion: The two surveys were based on different sampling procedures. Ho wever, this difference in design is unlikely to explain the systematic bias observed between self-reported and measured values for height and weight. This bias entails the overall validity of BMI assessment from telephone sur veys.