Análise de viabilidade da realização da vigilância de fatores de risco e proteção para doenças crônicas não transmissíveis por meio de painel online

Abstract

Introduction. The growth of non-communicable chronic diseases (NCDs) has led to conduct population surveys to monitor their risk and protective factors. While household or telephone data collection is more common, the internet has been suggested as a lower time and cost alternative. This study aims to analyze the feasibility of conducting NCD surveillance among Brazilian adults through an online panel. Methodology. The study compared results collected through an online panel (pilot study) with those obtained from telephone interviews (control study) and an official survey on the topic (Vigitel). For the pilot study, data collection was conduct via an online panel with a non-probabilistic sample of at least 500 adults (≥ 18 years old) residing in Belo Horizonte. For the control group, telephone interviews were conduct, limited to a single capital city, with a minimum probabilistic sample of 500 adults (≥ 18 years old) residing in Belo Horizonte and having telephone access. Indicators of lifestyle, health conditions and sociodemographic variables were compared. Weighting factors were employed to correct unequal selection probabilities in the control group and to equal the sociodemographic distribution of the samples to the city's demographics. Panel bias was estimated and compared, using the control group and Vigitel 2023 as references. Regression models assessed the indicators across the different data collection modes. The project obtained approval from the Research Ethics Committee of the School of Nursing at the Federal University of Minas Gerais (UFMG). Results: The studies showed a similar distribution between genders, but the online panel overestimated the population aged 35 to 54 and those with higher education (58.7% vs. 35.3% and 59.8% vs. 37.0%, respectively). The weighting factors reduce the relative biases of most indicators, resulting in similar estimates between the studies, except for the consumption of ultra-processed foods, which was overestimated online (34.8% vs. 26.2%, respectively). There was also similarity with Vigitel, except for ultra-processed consumption and excessive alcohol (37.5% vs. 18.8% and 34.8% vs. 22.4%, respectively), both overestimated online. The likelihood of reporting ultra-processed food consumption or excessive alcohol consumption by telephone was, respectively, 72% and 28% lower than online. The online panel required 10 invitations to obtain a complete questionnaire and the average completion time was 10 minutes. Furthermore, data collection by panel was 2.3 times cheaper than telephone. Conclusion. Non-probabilistic online panels are an option for health research that demands swift data collection or linked to probabilistic surveys. However, when continuous, repeated, and probabilistic data collection is necessary, online methodologies with lower costs and probabilistic procedures take precedence.