Schools that had taken part in a previous scientific engagement project (I’m a Scientist, Get me out of here! or IAS Debate Kits) were invited to participate in Decipher my Data.(5) Secondary schools in England with pupils aged 11-18 were eligible. We collected data for two influenza seasons in 2011/12 and 2012/13. Schools were all located in England, and, to make the results comparable to Public Health England’s (PHE) national surveillance data, South, Central and North geographical regions were used to classify the region of each one taking part.(6,7)
After consent had been received from the Head teacher, schools uploaded basic data including the number of pupils in each year group, the percentage of students in different ethnic groups, the percentage of children on free school meals and the full time equivalent number of teachers. Each week, schools were asked to submit the total number of half-day absences for medical reasons in each year group via the project website – https://flu.deciphermydata.org.uk/. Data submission was encouraged from week 38 (September) until week 13 and 12 (March) in the 2011/12 and 2012/13 seasons respectively. Data were collected for these periods as it was likely to capture the peak period of influenza circulation. A password protected website was built that enabled school teachers and pupils to submit data each week and take part in the learning activities specifically designed to engage students with data analysis. Detailed instructions were provided on how the data should be collected and uploaded, and telephone advice was provided in case difficulties were encountered. No face-to-face training was provided. Schools were asked to submit data in as timely a manner as possible, and email reminders about the submission of data were sent on an ad hoc basis.
No details were collected about the medical reason for each absence as schools do not routinely collect these. In the second year of the project, weekly school absence data were automatically time stamped when uploaded, allowing analysis of the time lag between the end of a school week and the time to submission on the project website.
The primary outcome was prevalence of daily school absence prevalence. This was calculated using the number of school absences per day as the numerator, and the number of pupils in years 7-11 as the denominator. Consistent with previous studies, we analysed data for years 7-11 only, as children in years 12 and 13 tend to have higher levels of scheduled absence, due to the more variable nature of their school timetable and provision for personal study time, making the denominator data less reliable.(3) Schools submitted the aggregate number of absences for each year group on a weekly basis, along with the number of half-day sessions at the school (e.g. 10 represented a full 5 day week). Poisson regression was used to calculate daily school absence prevalence weighted by region. Weights were generated by using the proportion of children in years 7 to 11 for each region in England using school census data for 2010, and the proportion of children for each region taking part in this project and submitting their data each week.(8) Weighting was performed to account for differences in the proportion of children sampled in each region and proportion of all children attending schools that region. In weeks where there were no data submitted by schools for a particular region, weighting was calculated using the two remaining areas.
We anticipated the recruitment of around 100 schools. This would enable a 2% daily prevalence of school absence to be calculated with 95% CIs of 1.9-2.1% (off-peak influenza) and a 7% prevalence with 95% CIs of 6.8-7.2% (peak influenza).
Descriptive analysis was performed to examine the number of schools submitting data each week, the time to upload data (i.e. lag between end of school week and receipt of data, second year of data only), the total number of pupils in the sample population and their geographical distribution.
To examine the association between weekly weighted school absence prevalence and levels of influenza circulating in the community, results were triangulated graphically and through univariable linear regression against Royal College of General Practitioners (RCGP) influenza like illness (ILI) episode incidence rate per 100,000 population for all ages, and rates among individuals aged 5-14.(9) School absence data were also plotted against microbiological surveillance data (Datamart) on the proportion of samples positive for influenza (A+B). DataMart is based on laboratory results collated from a network of 16 PHE and NHS laboratories in England and includes respiratory swabs from primary and secondary care. These swabs are tested for a variety of viruses using real time polymerase chain reaction assays.(6) Analyses were conducted for weeks 38 to 13, the periods when schools were being actively encouraged to submit their data. All RCGP and Datamart data were taken from the weekly reports and therefore may differ slightly to what is written in the final annual reports produced by HPA/PHE.
School absence prevalence data were plotted against Datamart data for respiratory syncytial virus (RSV) and Rhinovirus and laboratory confirmed cases of Norovirus reported in Public Health England’s weekly health protection report.(10) These infections are common in children and might potentially explain trends in school absence prevalence as the medical reason for each school absence is not routinely collected and therefore we did not ask schools to submit these data. This counterfactual analysis was therefore conducted to examine this possible alternative explanation for any associations found by triangulating results graphically and through univariable linear regression. All analyses were conducted in Stata version 12.
This work was conducted as a public engagement in science project and several interactive lesson plans were developed for schools taking part. Topics covered in these sessions included an introduction to the data, how to analyse the results, and how to write up the results. The project team wrote regular blogs about the project that were posted on the study website and emailed to students and teachers. During the first year of the project, students were able to post questions to scientists taking part, and in both years of the project, students were able to write ‘Lab logs’ about their observations and analysis, which the authors responded to.
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.
5. I’m a Scientist, Get me out of here! | A science outreach education and engagement activity [Internet]. [cited 2014 Mar 13]. Available from: http://imascientist.org.uk/
6. Agency HP. Surveillance of influenza and other respiratory viruses in the UK: 2011-2012 report [Internet]. [cited 2014 Mar 13]. Available from: http://www.hpa.org.uk/webw/HPAweb&HPAwebStandard/HPAweb_C/1317134576275
7. Agency HP. Surveillance of influenza and other respiratory viruses, including novel respiratory viruses, in the UK: Winter 2012-13 [Internet]. [cited 2014 Mar 13]. Available from: http://www.hpa.org.uk/webw/HPAweb&HPAwebStandard/HPAweb_C/1317139320524
8. Schools, pupils and their characteristics: January 2010 – Publications – GOV.UK [Internet]. [cited 2014 Mar 13]. Available from: https://www.gov.uk/government/publications/schools-pupils-and-their-characteristics-january-2010
9. Fleming DM, Crombie DL. The incidence of common infectious diseases: the weekly returns service of the Royal College of General Practitioners. Health Trends. 1985 Feb;17(1):13–6.
10. Public Health England. Health Protection Report [Internet]. [cited 2014 Mar 14]. Available from: http://www.hpa.org.uk/hpr/
