Healthcare-seeking behavior and hidden influenza-like-illness across the COVID-19 pandemic: a multi-country participatory surveillance study

Assessing healthcare seeking behavior is critical to estimate Influenza-like-illness (ILI) burden, i.e., the fraction of population infected, and inform vaccination campaigns for Influenza, SARS-CoV-2 and RSV, aimed at reducing disease circulation and severe illness. The Covid-19 pandemic put national healthcare systems under stress, affecting surveillance and healthcare seeking behavior of patients across demographic groups and countries. The under-reporting of influenza sentinel surveillance due to patients not consulting healthcare centers hinders consistent influenza burden estimates and hampers the understanding of influenza transmission at the international level. We leveraged an online participatory surveillance platform, InfluenzaNet, collecting data in six European countries, to study how the pandemic impacted individual healthcare seeking behavior in ILI patients and to estimate age-specific burden of ILI cases in terms of seasonal attack rates in the digital cohorts, hence including individuals who did not visit healthcare services. We studied drivers of healthcare seeking behavior such as demographics, socio-economic traits, comorbidities, duration and severity of symptoms, education and work status via a logistic regression model adjusted by age and sex of national populations, correcting for sampling biases. The presence of comorbidities, the duration and severity of the symptoms, and the self-assessed cause of symptoms, reflecting patients self-diagnosis, were drivers for health-seeking behavior consistently found across most countries, before and during the COVID-19 pandemic. The pandemic instead had a country-specific impact on age, gender, education and labor as determinants of healthcare seeking behavior throughout the crisis. We estimated ILI burden across countries using a standardized case definition, and assessed the burden of ILI that went undetected due to individuals not seeking healthcare from the general practitioner (GP, family physician) or other services. Our work highlights the advantage of integrating traditional systems with online participatory surveillance, which captures healthcare seeking behavior and helps to better assess the impact of pandemics on respiratory illnesses circulation. Further adoption of these tools has the potential to favor EU-level respiratory diseases surveillance, vaccination guidelines for Influenza, RSV and SARS-CoV-2 and a deeper understanding and management of ILI epidemiology and impact. Research in contextO_ST_ABSEvidence before this studyC_ST_ABSBefore the SARS-CoV-2 pandemic, age, gender, education, the self-assessed cause of symptoms, their severity and duration were identified as known predictors of healthcare seeking behavior due to influenza-like-illness. However, most studies on population attendance to medical service focused on single countries, without assessing the robustness of these predictors across countries. Furthermore, it is not known what the impact of the pandemic was on their significance. Assessing healthcare seeking behavior is critical to estimate the burden of influenza-like-illness and inform decisions on the roll out of vaccination campaigns. We performed a literature search on Google Scholar using the keywords " healthcare seeking behavior", "ILI burden estimate", "influenza-like-illness burden", "burden estimate", "ILI burden", with no time limits, in the English language. Added value of this studyOur study highlights the value of cross-country analyses by showing that not all known predictors of healthcare seeking behavior were robust across countries. Self-assessed cause of symptoms, their severity and duration proved robust across countries and pre-pandemic and pandemic periods except where small sample sizes prevented detection of significant effects. On the other hand, the pandemic had country-specific impact on socio-demographic predictors like age, gender, education and labor. Our work highlights substantial underdetection of ILI cases with respect to those estimated by national sentinel surveillance in all countries. Implications of all the available evidenceThis work suggests that the pandemic did not alter healthcare seeking behavior known predictors related to symptoms duration, severity and self-assessed cause, while we observed country-specific effects on age, gender and education factors. Digital cohorts may play a decisive role in integrating surveillance systems across countries for the estimation of ILI burden including medically unattended patients.