At the beginning of the COVID-19 pandemic, over a quarter of people around the world didn’t have access to handwashing facilities at home. The Hygiene Behaviour Change Coalition (HBCC), set up by the UK government and Unilever, sought to change this. As part of the coalition, BIT supported BRAC to pilot, adapt and roll out 1,000 foot-operated public handwashing stations across 20 sub-districts in Bangladesh, providing an opportunity to practice hand hygiene for thousands of people. Many development programmes would stop here. But access alone is often not enough.
So how do you test different approaches to see what works in an emergency setting? This is the challenge we took on with BRAC in 2020, early on in the pandemic.
Figure 1: Foot-operated handwashing stations installed by BRAC in Bangladesh
Measuring what people do, not just what they say
People often overstate their adherence to hygiene behaviours, because they know what the “correct” thing to do is, even if they don’t normally do it. Simply asking people how they behave can therefore lead to unrealistic estimates.
So instead of settling for self-reports, we worked with BRAC’s engineers to develop a new, low-cost source of behavioural data: we attached clicker counters to the station foot pedals for soap. As part of our piloting process, we found that clicks were highly correlated with observers’ counts of the number of station users.
From this initial work we found that only 15% of passersby stopped to use the new handwashing stations. It was up to BRAC and BIT to figure out how to change that.
Video 1: The tally counter (‘clicker’) mechanism used to measure station use
What does it take to increase handwashing?
Disappointing uptake of hygiene infrastructure isn’t a new challenge. However, low-cost changes to the design of the handwashing environment have shown promise at boosting hygiene behaviours. For instance, mirrors installed above sinks, designed to draw people to wash their hands, and footsteps directing people to sinks have shown large impacts. We therefore adapted these ideas to the local context and tested the effectiveness of mirrors on the handwashing stations and a series of large signposts pointing to them.
However, the success of these “low intensity” interventions often relies on a clear trigger for handwashing such as using a toilet. That’s why we also tested a higher intensity alternative. Here a promoter visited the station each week, offering free soap and facemasks as incentives to use the station, and updated a community board on how many people were using the station (i.e. highlighting the ‘social norm’). Unlike the low labour intensity interventions, the bundle of high-intensity interventions involved ongoing costs, and were only delivered for three weeks. Nevertheless, we were keen to see whether they had lingering effects in the weeks after the promoter stopped visiting.
Thanks to the monumental efforts of BRAC’s WASH team and the help of our partners D2, these two alternative approaches were tested at scale in a randomised controlled trial. Of the 855 stations and 540 villages and urban areas included, one-third were a control group, one-third received the bundle of low labour intensity interventions, and the remaining third received the higher intensity bundle.
Figure 2: Low labour intensity bundle – mirrors above sinks and signposts leading to the station
Figure 3: High labour intensity bundle – promoters updating a community message board and offering free soap and masks to station users
Active promotion and incentives worked (for a while) but signposts and mirrors didn’t
We collected data at each station for six weeks after the interventions were implemented. As shown in the graph below, station usage was roughly 15% higher in the villages that received the high-intensity bundle in the first three weeks (compared to the control villages). In contrast, we didn’t find evidence of any effect from installing the low-intensity interventions. Perhaps disappointingly, the effect of the high-intensity bundle didn’t persist in the three weeks after the promoters stopped visiting, with usage eventually equalising across the three conditions.
Figure 4: Results from the RCT comparing low and high labour intensity interventions against business-as-usual programming
We cannot know for sure why our low-intensity interventions did not work, but we have some clues thanks to supplementary data collected by BIGD at 100 of the stations. It provided some tentative evidence that our signposts increased footfall, but only for stations that weren’t already in busy locations. So in busy areas (where most stations were placed), this intervention seems to have low potential.
Just as interesting for the high-intensity interventions: active promotion worked well, but only whilst it was taking place – afterwards usage dropped back down. This is an important finding because it may suggest that such interventions need to continue over a longer period to build up a robust habit of use. Even so, these 3 weeks led to an estimated 100,000 additional handwashes with soap during the course of the trial, working out at roughly 50p each.
Even during a crisis, measurement and testing is worthwhile and possible
Our work with BRAC (as well as the excellent trial on mask promotion they helped to implement) shows that even in a crisis requiring urgent action, rigorous evaluations are possible and effects on actual behaviour can and should be measured. By taking an adaptive approach from the beginning of the project — including using online experiments and pilot evaluations — and embedding data collection into programme activities, we made sure to create actionable evidence for BRAC and the HBCC without getting in the way of programme rollout.
That’s not to say that running an RCT in this context was without its challenges – which we’ll explore in more detail in a subsequent blog. But despite challenges, and thanks to the efforts of BRAC, BIGD and FCDO, we were able to offer evidence for the fact that, at least in this context, if you want them to come, you’ll have to do a little more than just build it.