The present study describes clustering of lifestyle
risk factors in a relatively large population. However,
the response to this study was approximately 44%. The
percentage of smokers in our study population was
somewhat lower in men but similar in women compared with national data (Stivoro, 2000). From nonresponse analyses it was made clear that response was
higher among women, older people (40), and higher
educated subjects. Compared with nonrespondents, respondents had a lower self-rated health status. No
difference was observed for physical activity. A selection in the response may have affected the prevalence
estimates and the clustering analyses. Since the analyses on clustering were performed in subgroups of age,
gender, and education nondifferential bias is reduced.
However, it is unclear to what extent the differential
selection bias may have affected the outcome of the
study.