Web-based survey attracted age-biased sample with more severe illness than paper-based survey

To assess how web-based recruitment is comparable to postal surveys.In 2002, we invited female users of major Norwegian websites to join a women's health study on the Internet. The results of this study on the prevalence of urinary incontinence (UI) were compared with similar data collected by post in a previous epidemiological study, EPINCONT (Epidemiology of Urinary Incontinence in Nord-Trøndelag).Altogether 1,812 web respondents compared with 27,936 postal respondents from the EPINCONT study. The Internet sample was younger than the EPINCONT sample (37 vs. 48 years, P0.05). The proportion of women 60 years or older was 3.3% in our study and 29.0% in the EPINCONT study. Unadjusted prevalence of UI was lower in our study (20%) than in the EPINCONT study (25%), but stratified prevalence rates were higher in all individual age groups. In the Internet sample, we found less slight UI in all age groups, and more moderate (30-39 and 50-59-year age groups) and severe UI (30-39, 40-49, and 50-59-year age groups).We attracted a younger population with more severe UI than the EPINCONT study. Web-based approaches are less appropriate for studies on conditions concerning the older population than postal methods.