Participant profiles according to recruitment source in a large Web-based prospective study: experience from the Nutrinet-Santé study.

J Med Internet Res. 2013 15(9):e205.

Kesse-Guyot E, Andreeva V, Castetbon K, Vernay M, Touvier M, Méjean C, Julia C, Galan P, Hercberg S.

UREN, Inserm U557; Inra U1125, Cnam, Sorbonne Paris Cité, Université Paris 13, Bobigny, France. 

Interest in Internet-based epidemiologic research is growing given the logistic and cost advantages. Cohort recruitment to maximally diversify the sociodemographic profiles of participants, however, remains a contentious issue.
The aim of the study was to characterize the sociodemographic profiles according to the recruitment mode of adult volunteers enrolled in a Web-based cohort.
The French NutriNet-Santé Web-based cohort was launched in 2009. Recruitment is ongoing and largely relies on recurrent multimedia campaigns. One month after enrollment, participants are asked how they learned about the study (eg, general newscast or a health program on television, radio newscast, newspaper articles, Internet, personal advice, leaflet/flyers) The sociodemographic profiles of participants recruited through operative communication channels (radio, print media, Internet, advice) were compared with the profiles of those informed through television by using polytomous logistic regression.
Among the 88,238 participants enrolled through the end of 2011, 30,401 (34.45%), 16,751 (18.98%), and 14,309 (16.22%) learned about the study from television, Internet, and radio newscasts, respectively. Sociodemographic profiles were various, with 14,541 (16.5%) aged ≥60 years, 20,166 (22.9%) aged <30 years, 27,766 (32.1%) without postsecondary education, 15,397 (19.7%) with household income <€1200/month, and 8258 (10.6%) with household income €3700/month. Compared to employed individuals, unemployed and retired participants were less likely to be informed about the study through other sources than through television (adjusted ORs 0.56-0.83, P<.001). Participants reporting up to secondary education were also less likely to have learned about the study through radio newscasts, newspaper articles, Internet, and advice than through television (adjusted ORs 0.60-0.77, P<.001).
Television broadcasts appear to permit the recruitment of e-cohort participants with diverse sociodemographic backgrounds, including socioeconomically disadvantaged individuals who are usually difficult to reach and retain in long-term epidemiologic studies. These findings could inform future Web-based studies regarding the development of promising targeted or general population recruitment strategies.