Determinants of follow-up participation in the Internet-based European influenza surveillance platform Influenzanet

Abstract

Background: "Influenzanet" is a network of Internet-based platforms aimed at collecting real-time data for influenza surveillance in several European countries. More than 30,000 European volunteers participate every year in the study, representing one of the largest existing Internet-based multicenter cohorts. Each week during the influenza season, participants are asked to report their symptoms (if any) along with a set of additional questions.

Objective: Focusing on the first influenza season of 2011-12, when the Influenzanet system was completely harmonized within a common framework in Sweden, the United Kingdom, the Netherlands, Belgium, France, Italy, and Portugal, we investigated the propensity of users to regularly come back to the platform to provide information about their health status. Our purpose was to investigate demographic and behavioral factors associated with participation in follow-up.

Methods: By means of a multilevel analysis, we evaluated the association between regular participation during the season and sociodemographic and behavioral characteristics as measured by a background questionnaire completed by participants on registration.

Results: We found that lower participation in follow-up was associated with lower educational status (odds ratio [OR] 0.80, 95% CI 0.75-0.85), smoking (OR 0.64, 95% CI 0.59-0.70), younger age (OR ranging from 0.30, 95% CI 0.26-0.33 to 0.70, 95% CI 0.64-0.77), not being vaccinated against seasonal influenza (OR 0.77, 95% CI 0.72-0.84), and living in a household with children (OR 0.69, 95% CI 0.65-0.74). Most of these results hold when single countries are analyzed separately.

Conclusions: Given the opportunistic enrollment of self-selected volunteers in the Influenzanet study, we have investigated how sociodemographic and behavioral characteristics may be associated with follow-up participation in the Influenzanet cohort. The study described in this paper shows that, overall, the most important determinants of participation are related to education and lifestyle: smoking, lower education level, younger age, people living with children, and people who have not been vaccinated against seasonal influenza tend to have a lower participation in follow-up. Despite the cross-country variation, the main findings are similar in the different national cohorts, and indeed the results are found to be valid also when performing a single-country analysis. Differences between countries do not seem to play a crucial role in determining the factors associated with participation in follow-up.

Keywords: Internet; influenza; participatory surveillance.

Figure 1

Panel A: Illustration of study definitions for enrollment and follow-up participation. Definition for enrollment: first symptoms questionnaire (SQ) at least 60 days before end of monitoring season and at least one more SQ within 15 days from first SQ; Definition for follow-up participation: at least two SQs during time window of 30 to 60 days after first SQ. Panel B: The numbers of Influenzanet users and enrolled participants, obtained applying the enrollment definition given in Panel A, are shown. The final sample on which the multilevel regression was performed, after the complete-subject exclusion of individuals with missing records, was of 22,968 enrolled individuals who provided information for all the variables. Among these, after applying the definition for follow-up participation given in Panel A, 16,161 were respondent participants and 6,807 had a lower participation in the follow-up.

Figure 2

Country-specific distribution of independent variables of enrolled participants: gender, age, education level, household composition, smoking, presence of a chronic condition, and vaccination status (SE = Sweden, UK = United Kingdom, NL = the Netherlands, BE = Belgium, FR = France, IT = Italy, PT = Portugal).

Figure 3

Crude (left panels) and adjusted (right panels) odds ratio by country. An odds ratio greater than 1 is associated with higher likelihood of participation in follow-up ((SE = Sweden, UK = United Kingdom, NL = the Netherlands, BE = Belgium, FR = France, IT = Italy, PT = Portugal).