In this cluster-randomized trial [1], the authors tested whether a scalable village-level intervention based on emotional drivers of behavior, rather than knowledge, could improve handwashing behaviour in rural India. Fourteen villages were randomly assigned (1:1) to intervention (community and school-based events incorporating an animated film, skits, and public pledging ceremonies) or control (no intervention). Outcomes were measured by direct observation in 20-25 households per village at

more »

... eline and at three follow-up visits (6 weeks, 6 months, and 12 months after the intervention). At 6 weeks, hand washing with soap at key events was more common in the intervention group than in the control group (19% vs 4%; P=0·005). At the 6-month follow-up visit, the proportion washing hands with soap was 37% in the intervention group versus 6% in the control group. At the 12-month follow-up visit, after the control villages had received the shortened intervention, the proportion washing hands with soap was 29% each in the intervention and control group. The authors concluded that substantial increase in hand washing with soap can be achieved using a scalable intervention based on emotional drivers. COMMENTARIES Evidence-based-medicine Viewpoint This study has considerable relevance to the Indian setting for three distinct reasons. First, although it is well known that household hand hygiene practices are associated with reduction in infection-related morbidity [2-6], it is a challenge to convey this information to the target audience in an appealing fashion [7] . Second, merely conveying the message is inadequate unless it is backed by actions to promote (and measure) the appropriate behavior [8] [9] [10] . Third, the partial success observed in this trial suggests that such strategies could be scaled-up to include a wider population, and could also be extended to other health-care needs wherein a combination of education with behavior change are required (for example, routine immunization, infant nutrition, newborn care and rational antibiotic use). This trial has included the methodological refinements associated with high-quality randomized trials; this is commendable considering the practical difficulties of stationing an observer in the household for three hours each day. Sample size calculation and statistical treatment of data appear appropriate. However, there is no mention of an intention-to-treat analysis, and it is unclear what was done for households where data could not be collected. The steep rise in handwashing observed between 6 weeks and 6 months in the intervention villages (19% to 37%) was because 4 villages that initially showed no improvement appeared to improve later in the trial. The reasons for the difference in behavior at 6 weeks, and the delayed response are unclear, especially as 3 'better' villages showed no dramatic increase beyond 6 weeks. Village-wise data analysis also suggests that there was a decline in handwashing in one village. Interestingly, at the 6-month observation point, 3 of the 7 control villages showed greater hand-washing practices than the paired intervention villages. This has also not been satisfactorily explained. These observations suggest that pooling the data together masks the differences at the ground level (and the reasons thereof). This has important implications because it suggests that there are behavioral differences at the individual household level, rather than village level. Therefore a campaign targeting households rather than villages could have greater efficacy. The fact that even the small-scale intervention provided to the control villages (after six months) resulted in remarkable improvements, suggests that improvement (in both sets of villages) may be influenced both by population contact with program providers as well as the program content. This trial was conducted in a typical rural setting within India, using local resources, tools and language. Therefore, theoretically it should be easy to extend such a strategy all over the country, and obtain similar impressive results. However the key