Rosen 2006.

Methods	A cluster randomised controlled trial. Initially developed in the spring of 1999, and piloted in the 1999–2000 school year, the main study was run during the 2000–2001 school year, with the intervention delivered January to March 2001 and follow up two to three times to June 2001. In a concurrent subtrial, the families of children in the intervention preschools were individually randomised to home intervention group or home control group to test the impact of a home intervention. Collaborating partners Lead agency: Public health Strategic involvement (policy making and service planning): Local public health department, Ministry of Health, Hadassah Medical Organization, Ministry of Education, Effrata Teachers’ College, Preschool Department of the Municipality of Jerusalem. Commissioning (implementing strategy taking account of resources available): public health, teachers, preschools  Operational (providing services directly): public health, teachers, preschools, doctors, educational experts, school nurses Set in Israel.
Participants	40 preschools (20 intervention, 20 control), stratified by sector (religious and secular) in the state‐run public system of the Jerusalem region, including 1029 toilet‐trained children, aged 3and 4. 73,779 child days were yielded from observations on 6 baseline and 66 study days. In a concurrent subtrial, the 469 families of children in the intervention preschools were randomised  to receive a home component  (intervention group 237 families, controls 232 families). Preschool teachers likely to comply with the trial protocol  were recommended by Ministry of Education officials (60% of eligible teachers). Nearly 90% of those invited agreed to join.
Interventions	The preschool educators were given lectures, printed materials and experiential learning. The children were encouraged to wash their hands for at least 10s by singing a handwashing song. They also had puppet theatre, a self‐reward system, games, posters, puzzles, a video and presentations by school nurses. Environmental interventions included providing each classroom with liquid soap dispensers, paper towel dispensers, cup racks, liquid soap, paper towels ( instead of cloth towels)  and individual cups (instead of communal cups) over a three month period. Equipment was provided and fitted to ensure all intervention schools had the same facilities at the start of the intervention. Control preschools had no input until the close of the study period, when the full intervention was delivered on site and they were followed up once after the intervention. The home component consisted of a video, card and magnet sent home with the children in individually labelled packages about one month after the launch of the intervention. The home component control families received materials related to toothbrushing.
Outcomes	The primary outcome measure was illness absence from preschool. Absences were recorded via telephone using a structured questionnaire. They were classified as due to illness, for unknown reason or for reason unrelated to illness. Where the reason for absence was unknown parents were contacted to clarify the reason. Secondary outcomes were the overall percentage of children washing hands with soap before eating lunch and after bathroom use. Handwashing was measured from 3 post‐intervention visits to the 20 intervention preschools between January and June 2001 and 1 post‐intervention visit to the 20 control preschools in June 2001. In total there were 6 days collecting baseline measurements and 66 study days, yielding 73,779 child days of observation. Preschools were grouped into religious and secular subgroups within the intervention and control groups for comparative analysis. 88% of teachers and 95% of parents agreed to participate. No preschools dropped out from the study. Drop‐out of children was 0.9%, and 0.7% were lost to follow‐up.
Notes	The teachers invited to deliver the programme were identified by Ministry of Education officials. 60% of all eligible teachers were suggested, of which nearly 90% agreed to take part.  Additional resources included lectures for the teachers, environmental equipment for the preschools, educational resource packs and puppet theatre visits. For the home intervention there were information packages. Teachers needed to spend time reinforcing hand hygiene messages with the children. Fidelity of the programme was assessed as good but imperfect. Contamination of the control group was minimal. Two intervention preschools were unexpectedly exposed to raw sewage during the study but results were unchanged following sensitivity analysis excluding these preschools. Overall risk of bias was high.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Parents and field research staff were not aware
Blinding (performance bias and detection bias) All outcomes	Low risk	Educators, parents and field research staff were not informed of the study design but sometimes became aware that the programme was being run in a certain school.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All data appears to be presented
Selective reporting (reporting bias)	High risk	The educators consistently under‐reported absenteeism. The research team conducted surveys which showed that 28% of absences reported by parents were not reported by educators. (ref)
Other bias	High risk	Potential selection bias as the teachers' were put forward by Ministry of Education officials rather than volunteering or all eligible teachers being directly approached.
Randomisation adequately described/protected?	Low risk	Used computer generated random numbers
Protection against contamination?	Low risk	Contamination assessed as minimal. Intervention delivered through discrete pre‐schools
Follow‐up rate adequate?	Low risk	No schools dropped out.  0.9% children dropped out and 0.7% were lost to follow‐up.
Reliable primary outcome measure?	High risk	It was hard to assess which absences were due to illness and educators consistently under‐reported absences when compared to parent's reports.
Groups measured at baseline?	Low risk	Groups approximately balanced.
Appropriate choice of controls (CBA studies only)?	Unclear risk	N/A
Contemporaneous data collection (CBA studies only)?	Unclear risk	N/A
IS THE STUDY AT LOW RISK OF BIAS?	High risk	OVERALL RISK OF BIAS WAS HIGH