Pesquisa epidemiológica baseada na web: estratégias e características pessoais associadas à adesão e participação de adultos jovens

Abstract

In the recent years, epidemiologic surveys are being compromised by the reduction on the response rates. This problem can jeopardize the validity of the findings due to the higher probability of occurrence of the nonresponse bias. Several factors contribute to the declining response rates of epidemiologic studies, including the data collection method used (paper 1). With the growing internet coverage and access around the globe, some studies conducted in high-income countries are using the internet to collect epidemiologic data. However, using the internet to conduct surveys in low- and middle-income countries is still in its infancy. Thus, the first objective of this thesis was to develop the coortesnaweb, a platform used to conduct internet-based epidemiologic surveys with the members of the 1993 Pelotas birth cohort. The coortesnaweb cover all the steps through recruitment, data collection and dissemination of results, including some extra features such gamification strategies. Using the platform, we analyzed the efficiency and the inequalities of three online and free-to-use recruitment methods, user to invite the members of the 1993 birth cohort to register into the platform (paper 2). The eligible cohort members were randomly allocated to be recruited using e-mail, Facebook or Whatsapp messages. The recruitment rate was calculated for each method and according to socio demographic characteristics of the sample. The absolute and relative inequalities on recruitment were also analyzed according to schooling and socioeconomic position. The overall recruitment rate was 26.8%, being Facebook the most efficient recruitment method (recruitment rate = 30.6%). The highest inequalities on recruitment according to schooling and socioeconomic position were seen using e-mail as recruitment method. The Facebook was the methods that presented the lowest inequalities. Also using the coortesnaweb platform, we analyzed the influence of questionnaire length and frequency of reminders on the response rates of five questionnaires using a 2x2 factorial design (paper 3). At the moment of registration into the coortesnaweb platform, the participants were randomly allocated to respond to short or long questionnaires, and to receive reminders with low or high frequency. Besides calculating the response rate for each questionnaire, we also analyzed the total number of questionnaires responded according to the socio demographic characteristics of the sample. Considering the five questionnaires together, the mean response rate was 54.3%. Sending reminder more frequently was positively associated with the response rates of the first two questionnaires, but the questionnaire length had no influence on the response rates. Women and participants that studied for 12 years or more had, respectively, 13.0% (95%CI = 1.03, 1.23) e 28.0% (95%CI = 1.09, 1.50) increased risk of responding an additional questionnaire compared to the reference groups. Our findings show that it is feasible to conduct internet-based epidemiologic surveys in the Brazilian context, even though the internet coverage is not universal. Additionally, the higher participation of some specific groups should be considered in order to interpret the findings from web-based survey, especially regarding prevalence estimates.