This study aimed to establish an electronic system for influenza surveillance using public engagement with science, and investigate whether school absence prevalence due to illness was correlated with established surveillance measures for influenza. The project ran during the 2011/12 and 2012/13 seasons and recruited 47 schools with data submitted by 27. In both seasons, peaks in school absence prevalence occurred one weak before peaks in national data. Linear regression showed a moderate association between school absence prevalence and RCGP reported influenza like illness, laboratory confirmed cases of influenza A & B, and some evidence for a linear association with Rhinovirus (negatively correlated) and Norovirus.
Strengths and weaknesses of the study
This study has demonstrated the feasibility of establishing a novel mechanism for influenza surveillance in schools using public engagement in science. The study not only resulted in academically interesting findings, but served an important educational purpose. Students taking part were provided with a unique opportunity to analyse data and learn about basic epidemiological concepts and the interpretation of scientific results. The fact that levels of influenza were low during the first year, whilst frustrating from a research perspective, was educationally useful, demonstrating how not everything in science goes exactly according to plan despite much preparation.
There are several limitations of using school absence data for surveillance of influenza activity in the community, particularly the fact that it is not possible to collect any data during school holidays. This was important during the 2012/13 season when traditional surveillance indicated a peak during week 52, which fell during the Christmas holidays. School absence prevalence demonstrated a peak during week 51, but it was not possible to examine levels during week 52. As a result, it was not possible to examine further whether school absence prevalence peaked earlier than other markers of influenza activity. It was not possible to analyse national surveillance data for comparable age categories to those collected by this research project. Clinical reason for an absence was not collected, and despite using only data for illness absences it is likely that other illnesses rather than flu were the cause in some cases..
Schools taking part in Decipher my Data were predominantly based in the South of England, and despite weighting results to make them nationally representative, the convenience sampling and low number of schools in the North may have lead to a bias and greater levels of uncertainty within these estimates. Schools were given detailed instructions on how to collect and upload data, however, measurement bias (due to inconsistencies in the way the data were processed and uploaded may have varied across schools taking part) is likely to have been randomly distributed and will therefore lead to a non-differential bias and a reduction in the study power.
Comparison to existing literature
A previous study conducted in England recruited eight primary schools and three secondary schools during the 2005/6 season.(3) This study was carried out for one season and included self-reported cause of the illness by the parent or guardian at the time of notifying the school about a child’s absence. The RCGP ILI peak occurred one week after school absence data. A similar study was also performed during 2005-2007 using results from six primary schools in a single local authority in East London.(4) This study was able to calculate both the incidence and prevalence of school absences from the data collected. Peaks in the prevalence of school absence showed a greater correlation with laboratory confirmed cases of influenza A & B than incidence rates.
A further study used public engagement in science to investigate the transmission of influenza in young children.(12) The study used children aged 13-15 to capture mixing patterns in children aged 4-11. Data collection questionnaires were designed in association with the school children and administered by the students aged 13-15. The study found evidence of sex-specific assortive mixing within and between classes in the same school, and a marked social structure. The authors concluded that the methods were a helpful way to examine mixing patterns in this difficult to research group.
Interpretation of the findings
The results of this study provide evidence that school absence prevalence could be a useful tool for surveillance of influenza in children aged 11 to 16 and may have utility for providing earlier warning of an outbreak than existing measures. The data are likely to be more representative of the community burden of disease in children compared to existing surveillance data. Schools were actively engaged in the collection and analysis of the data. It was not possible to establish with certainty whether school absence prevalence detected outbreaks of disease earlier than existing data, or whether peaks occurred at an earlier stage, and future work should be carried out to examine these possibilities as such data would enable a more timely public health response to influenza epidemics and pandemics.
References:
3. Mook P, Joseph C, Gates P, Phin N. Pilot scheme for monitoring sickness absence in schools during the 2006/07 winter in England: can these data be used as a proxy for influenza activity? Euro Surveill Bull Eur Sur Mal Transm Eur Commun Dis Bull. 2007 Dec;12(12):E11–12.
4. Schmidt WP, Pebody R, Mangtani P. School absence data for influenza surveillance: a pilot study in the United Kingdom. Euro Surveill. 2010;15(3).
12. Conlan AJK, Eames KTD, Gage JA, von Kirchbach JC, Ross JV, Saenz RA, et al. Measuring social networks in British primary schools through scientific engagement. Proc R Soc B Biol Sci. 2010 Nov;278(1711):1467–75.
