Health scores in Flusurvey participants: findings from the 2012–13 influenza season

Abstract

BackgroundFlusurvey was launched in 2009 and is part of an internet-based Europe-wide survey, Influenzanet, to monitor influenza-like illness. Flusurvey is open to all residents of the UK. At the beginning of each influenza season, participants are asked to fill in a background questionnaire providing information about risk factors, including public transport use, background health problems, socioeconomic factors, and household information. Subsequently, participants are asked to complete a weekly survey and provide data for respiratory symptoms. Although Flusurvey had been used to understand incidence of influenza-like illness and risk factors, we asked participants to report health status in the form of health scores, to understand what factors affected health status. MethodsParticipants were recruited by social media, news coverage, and word of mouth. All UK residents of any age were eligible for participation. Each week, all participants (irrespective of whether they were ill or not) were asked to report how they were feeling by health scores on a scale of 0–100, where 0 is the worst imaginable health state and 100 represents perfect health. The weekly reports of all individuals (healthy individuals and individuals with influenza-like illness) were assessed to see which factors affected their overall health scores. Only participants with at least two reports were included. Health scores were converted into odds, and linear regression with robust standard errors was done on the log of the odds. To investigate the association of potential risk factors on health scores, we did a multivariate regression that included age, sex, region, household characteristics, presence of underlying health problems (diabetes, asthma, and kidney and heart disease), public transport use, vaccination, employment status, living with pets, living with children, working with children, socioeconomic class, and smoking. These variables were based on self-reported factors at baseline. FindingsBetween November, 2012, and April, 2013, 5943 participants were recruited, 4532 of whom had at least two reports, with nearly 40 000 total reports. In individuals reporting an influenza-like illness, female participants (odds ratio 0·92, 95% CI 0·83–1·01), participants with underlying health problems (0·82, 0·72–0·93), and those taking public transport (0·86, 0·77–0·97) reported the lowest health scores. Participants from London (1·39, 1·09–1·79) reported the highest health scores. In participants reporting no symptoms, individuals older than 65 years (1·53, 1·17–2·00) and those from the southeast coast region (1·26, 1·00–1·57) reported the highest health scores, and those with underlying health problems (0·72, 0·64–0·81), allergies (0·89, 0·82–0·97), and taking public transport (0·89, 0·82–0·98) reported the lowest health scores. InterpretationOur results show that examination of what factors contribute to how an individual self-reports a health score is possible. Not surprisingly, participants who have underlying health problems are more likely than participants without underlying health problems to report lower health scores, regardless of whether they report symptoms. Slightly more surprising is that older individuals report higher health scores when reporting no symptoms than younger age groups. Public transport use was associated with low health scores. Finally, males and females did not differ in their health scores. Limitations of the study include a non-random self-selecting sample, undersampling of children and diagnoses of influenza-like illness, and health scores being based on self-reported symptoms. FundingThe project was funded by EPIWORK as part of the EU Seventh Framework Programme (project reference 231807). They had no role in the writing of this abstract.