1. Objectives

To evaluate the effects of interventions to promote hand washing on diarrhoeal episodes in children and adults

2. How studies were identified

The following databases were searched to May 2015:

• Cochrane Infectious Diseases Group Specialized Register
• CENTRAL (The Cochrane Library 2015, Issue 5)
• MEDLINE
• EMBASE
• LILACS
• PsycINFO
• Science Citation Index and Social Sciences Citation Index
• ERIC
• SPECTR
• Bibliomap
• RoRe, The Grey Literature
• WHO ICTRP
• mRCT

Reference lists of indentified articles were also searched, and researchers and relevant organizations were contacted

3. Criteria for including studies in the review

3.1 Study type

Randomized controlled trials, including cluster-randomized controlled trials

3.2 Study participants

Adults or children in day-care centres or schools, communities or households, or patients in hospital

3.3 Interventions

Interventions, such as meetings, multimedia communications, and radio/TV campaigns, which promote hand washing after defecation, after disposal of children’s faeces, or before the preparation or handling of foods, compared to no hand washing promotion

3.4 Primary outcomes

• Episodes of diarrhoea (self-reported or from hospital/clinic records)

(Acute diarrhoea was defined as three or more loose stools in a 24-hour period; persistent diarrhoea was defined as diarrhoea lasting more than 14 days; dysentery was defined as stool with blood)

Secondary outcomes included: diarrhoea-related death; behavioural changes, such as changes in the proportion of people who reported/were observed washing their hands after defecation or disposal of children’s faeces or before food handling; changes in knowledge, attitudes and beliefs about hand washing; all-cause mortality in under fives; and cost-effectiveness

4. Main results

4.1 Included studies

Twenty-two randomized controlled trials, enrolling 69,457 participants, were included in this review:

• Twenty-one trials were cluster-randomized, and one was individually randomized
• Most trials reported data on the incidence or episodes of diarrhoea in treatment and control groups
• The age of participants ranged from children under three years to adults
• Large group hygiene education sessions that promoted hand washing through demonstrations and other multi-media materials was the main method of intervention, and soap was also provided in six community-based trials

4.2 Study settings

• Twelve trials were conducted in day-care centres or schools, using school, day-care centre, or classroom as the unit of randomization. These trials were predominantly conducted in high-income countries: Australia, Canada, China, Denmark, the Netherlands, and the United States of America (5 trials), and two trials were conducted in low- and middle-income countries: Egypt, Kenya
• Nine trials were community-based, using villages, squatter settlements, neighbourhoods, or households as units of randomization. These trials were conducted in low- and middle-income countries: Bangladesh, Democratic Republic of the Congo, India, Myanmar, Nepal, Pakistan (3 trials), and Peru
• One hospital-based trial was conducted in a group at high risk of diarrhoea (people with AIDS) in the United States of America and was individually randomized

4.3 Study settings

How the data were analysed
The effect of hand washing promotion interventions on the incidence of diarrhoea was compared to no hand washing promotion. The analyses were stratified into three categories: i) child day-care centres or schools, ii) community-based trials, and iii) hospital-based trials (high-risk groups). Where trials had not appropriately adjusted for clustering, standard methods were used to make adjustments using intra-class correlation coefficients from other trials. Random effects meta-analysis was used to calculate incidence rate ratios (IRR) for count outcomes, with corresponding 95% confidence intervals (CI). Subgroup analyses to explore heterogeneity were performed on the basis of whether hand washing material (soap) was provided as part of the intervention, trial setting, the type of promotional activity employed, and quality of the trial (blinding of outcome assessors). For changes in hand washing behaviour, outcomes or methods of measuring outcomes were too variable to make meta-analysis meaningful and therefore results were tabulated

Results
Child day-care centres or schools
Incidence of diarrhoea
In pooled analysis of 11 trials involving 50,044 children, a 31% reduction in the incidence of diarrhoea was observed in the group who received a hand washing promotion intervention (IRR 0.69, 95% CI [0.59 to 0.81], p<0.00001). A further trial did not report data in a way that was possible to calculate an IRR, but found no statistically significant difference between treatment and control groups. The pooled effect remained statistically significant in subgroup analysis by trial setting (high-income countries: IRR 0.70, 95% CI [0.58 to 0.85], 9 trials/4664 participants; low- and middle-income countries: IRR 0.66, 95% CI [0.43 to 0.99], 2 trials/45,380 participants). Two trials including 1045 participants focused the intervention entirely on hand washing, and no significant reduction in the incidence of diarrhoea was found (IRR 0.69, 95% CI [0.43 to 1.09]), while in the nine trials using multiple hygiene interventions, the effect was statistically significant (IRR 0.69, 95% CI [0.57 to 0.84]). Pooled analysis of data from the trials in which outcome assessors were blinded showed the benefit of hand washing to be somewhat reduced (IRR 0.74, 95% CI [0.56 to 0.98], 3 trials/1303 participants).

Behavioural changes
Four trials investigated behavioural changes, with all studies reporting an improvement in hand washing (2 trials) or hygiene behaviour (2 trials) in the intervention group. In one trial of 23 day-care centres, the improvement in hygiene practices was associated with a reduced risk of illness in children under three years of age (RR 0.34, 95% CI [0.17 to 0.65]), reflecting a reduction in the incidence of diarrhoea by two-thirds.

Community-based trials
Incidence of diarrhoea
In eight community-based trials including 14,762 participants, hand washing promotion reduced the incidence of diarrhoea by 28% (IRR 0.72, 95% CI [0.62 to 0.83], p=0.000017). One trial reported on mean longitudinal prevalence of diarrhoea rather than incidence, and found no difference between treatment groups. For community-based trials, the reduction in the risk of diarrhoea was greater in subgroup analysis of studies promoting hand washing only (IRR 0.63, 95% CI [0.52 to 0.78], 5 trials/10,888 participants) in comparison to studies promoting multiple hygiene interventions (IRR 0.81, 95% CI [0.69 to 0.95], 3 trials/3838 participants). In addition, the effect was greater in trials supplying soap (IRR 0.66, 95% CI [0.56 to 0.78], 6 trials/11,422 participants) versus those that did not (IRR 0.84, 95% CI [0.67 to 1.05], 2 trials/3304 participants), and was lower in trials blinding outcome assessors (IRR 0.80, 95% CI [0.67 to 0.94], 4 trials/3070 participants) compared with those that did not (IRR 0.63, 95% CI [0.48 to 0.83], 4 trials/11,656 participants). In trials reporting on the incidence of dysentery (4 trials) and persistent diarrhoea (2 trials), no evidence of a difference between groups was observed (data not pooled).

Behavioural changes
In three trials reporting on behavioural changes, all found improvement in hygiene practices (2 trials) or household soap consumption (1 trial) among the intervention group in comparison to the controls.

Hospital-based trial (high-risk group)
Episodes of diarrhoea
The mean number of diarrhoeal episodes over one year was lower in the hand washing intervention group compared to the control group, with 2.92 episodes in the control group and 1.24 in the intervention group, a reduction of 1.68 episodes (95% CI [-1.93 to -1.43], p<0.00001; 1 trial/148 individuals).

Behavioural changes
At the trial conclusion, the intervention group reported hand washing seven times per day compared to four times per day in the control group (p<0.05).

5. Additional author observations*

Using the GRADE approach, the evidence that hand washing promotion reduces the incidence of diarrhoea in child day-care centres and schools was considered high quality, although this finding may not be generalizable to low- and middle-income countries. For community-based trials, the evidence that hand washing promotion reduces the incidence of diarrhoea was considered to be of moderate quality, and it is unlikely that these results are generalizable to high-income country settings. Few trials reported adequate allocation concealment methods, and while few trials attempted to blind outcome assessors, when analyses were restricted to blinded trials only, the effect remained statistically significant, albeit slightly smaller. One trial attempted to investigate the possibility that effect sizes may have been attenuated by the intensive monitoring of outcomes, with control groups being monitored typically on a two-weekly basis. The investigators reported that the effect of monitoring on reducing diarrhoea episodes was statistically significant, while no overall treatment effect was found.

Provision of soap may also further reduce the incidence of diarrhoea over the effect of hand washing promotion alone, although too few trials provided data on this intervention to make any firm conclusions. Hand hygiene behavioural outcomes were improved in the intervention group in all settings, with an increase in the proportion of participants hand washing or complying with hand hygiene practices before eating or cooking and after visiting the toilet or cleaning the infant’s bottom.

Further trials conducted in day-care centres or schools in low- and middle-income countries are needed, as are further trials conducted in hospital-based settings. Extended follow-up durations should be examined, outcomes should be assessed in a manner that is structured and reduces monitoring, and outcome assessors should be blinded where possible in future trials.