Exploring behavioral determinants of handwashing with soap after defecation in an urban setting in Bangladesh: findings from a barrier analysis Open Access

Handwashing with soap (HWWS) at critical times (e.g., after defecation) is known to be an effective strategy to prevent diarrheal and infectious diseases. Evidence from randomized controlled trials suggests that HWWS may decrease diarrhea, pneumonia and other respiratory infections, and parasitic infections (Curtis & Cairncross 2003; Freeman et al. 2014; Wolf et al. 2018). The recent pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) also supports the importance of HWWS to prevent the risk of infection. Among other non-clinical measures, HWWS was found to be one of the most effective approaches in preventing SARS-CoV-2 infection (Alzyood et al. 2020). While HWWS has long been considered as one of the most cost-effective approaches to reduce the burden of disease (Bartram & Cairncross 2010), it has been estimated that only 19% of the global population wash their hands with soap after defecation or management of children's excreta (Hirai et al. 2016) and Bangladesh is not an exception. Despite the increased risk of diarrheal disease morbidity and mortality among the Bangladeshi population, increasing HWWS has been a major challenge in the country (Parveen et al. 2018).

Empirical evidence has identified a range of determinants that are associated with the practice of HWWS. These determinants include the availability of handwashing facilities, access to soap and water, social and cultural norms and supportive environment, motivations, knowledge on positive outcomes and risks, skills, and broader social and structural factors (White et al. 2020). Using a nationally representative sample, the 2014 Bangladesh National Hygiene Survey showed that 70% of urban households had a handwashing facility for use postdefecation with soap and water while poorer, urban households were less likely to have one (icddrb 2014). Nevertheless, other factors such as the desire to feel clean have also been identified as behavioral determinants of HWWS. A systematic review on formative studies from 11 countries in Asia, Africa, and South America showed that emotional drivers (e.g., disgust, comfort, and the desire for social status) are often key determinants of handwashing behavior (Curtis et al. 2009). Little information is available on which behavioral determinants are important to consider in promoting HWWS in the Bangladeshi context. Most previous studies in Bangladesh focused on identifying the gap between handwashing knowledge and practices and proxy indicators such as access to a handwashing facility (Hulland et al. 2013), observing cleanliness of handwashing (Halder et al. 2010), and the association of household characteristics with HWWS (Luby et al. 2009). Therefore, this study aimed at studying a broader array of behavioral determinants and their association with behavioral adoption to inform behavior change communication strategies and to better design contextualized activities to promote HWWS after defecation in Bangladesh. This study also contributes to the development of a comprehensive social and behavioral change (SBC) strategy and culturally inclusive program activities.

This barrier analysis (BA) study was conducted in Gazipur City Corporation (GCC) which was formed in 2013, merging then Tongi and Gazipur Municipalities, along with large semi-urban and rural areas. In 2017, an initiative entitled ‘Development of Integrated Wastewater Management System for Gazipur City Corporation (GCC)’ was started under the guidance of the Greater Dhaka Watershed Restoration (GDWR) Work-Stream of the Bangladesh Water Multi-Stakeholder Partnership (MSP), supported by the 2030 Water Resources Group (2030WRG) under the World Bank. SNV Netherlands Development Organisation is providing advisory technical assistance to the city council as part of its WASH SDG Program. The aim of this program is to enhance the capacity of the city authority so it can provide citywide, safely managed hygiene, and sanitation. A baseline population-based study found that 54.7% of the households have a handwashing station within 10 m of the toilet with access to water, while 40% of them did not have soap. The study also found that around 39% of the study population shared a toilet in different forms. However, the majority of people who had a handwashing station with soap did not have a water source that provided clean water for handwashing. In order to address this, the city authority agreed to implement an SBC intervention to influence household demand for – and adoption of – improved hygiene and sanitation services. As part of designing the SBC strategy, this BA study aimed to identify important beliefs and other important responses regarding the behavioral determinants of HWWS in order to create contextualized messages and activities.

As part of the Designing for Behavior Change (DBC) approach, BA is a formative research tool that was developed in 1990 by Davis, based on an earlier doer/non-doer analysis tool, and popularized by Kittle and Davis. BA is similar in some respects to the RANAS formative research tool (Mosler 2012), but requires less time and less statistical knowledge to implement by field practitioners. Cross-sectional surveys, key informant interviews, and focus groups have often been used to explore different beliefs and possible drivers of HWWS. However, commonly held beliefs about a behavior and other characteristics of adopters are not necessarily linked with adoption of a behavior and can lead an implementer astray when planning what to change through an intervention (Kalam et al. 2021). BA is meant to make up for that shortcoming by identifying beliefs and other responses associated with behavioral determinants that are correlated with adoption of a behavior. BA draws mostly from the health belief model (HBM), the perceived behavioral control (PBC), and the theory of reasoned action (TRA) model (Kittle 2017), and has been used mostly to study health, nutrition, and WASH behaviors (e.g., HWWS). The beliefs and other responses regarding behavioral determinants assessed during BA are identified so that more effective behavior change communication messages, strategies, and supporting activities can be developed, with a focus on the most actionable findings.

This study adapted the standardized BA questionnaire from the DBC training manual (The TOPS Program 2016) for exploring key beliefs and other responses regarding the behavioral determinants of handwashing behavior. As recommended by the BA developers, the research design began with defining the behavior, the details of when and how the behavior needs to be practiced, and – using information from the earlier baseline study – the priority groups for the behavior (see Box 1).

The BA questionnaire is divided into two parts. The first part includes a set of screening questions to identify the participant as either a ‘doer’ (a person who claims to wash their hands after the last time they defecated using soap and water, and has soap that has been used at a handwashing station) or a ‘non-doer’ (a person who does not regularly practice HWWS after defecation, or does not have soap [used or unused] at the handwashing station). In the screening section of the questionnaire, both self-reported handwashing behavior and the observed presence of used soap at the handwashing facility were used. The second part of the questionnaire consists of open- and closed-ended questions relating to each of the 12 determinants assessed with BA (described in Box 1). While not part of the traditional BA approach, along with the determinant on access, structured observations were done to record the presence of a handwashing facility within 10 feet of a latrine or toilet, the presence of a permanent handwashing facility in the home, and the presence of covered bucket of water used only for handwashing purposes. The questionnaire (available in Supplementary Material, file 1) was developed in English and then was translated into Bengali and back-translated into English to check for accuracy.

Adult men and women were the Priority Group for this BA study and were selected through random sampling from both slum and non-slum areas. The BA approach recommends a sample size of 45 doers and 45 non-doers in order to detect statistically significant odds ratios of 3.0 or higher at a confidence level of 90% (Kittle 2017). For this study behavior, 45 doers and 45 non-doers were selected from different areas of GCC in order to make the results more applicable to both contexts.

Data were collected through individual interviews with responses recorded on paper questionnaires by two teams with one female and one male in each team. The data collection team and the lead author coded the open-ended responses thematically, using both an inductive and deductive coding process. At the end of this process, the team quantified the responses in each category for doers and non-doers separately. These categories and the number of responses registered for each were then entered into a standardized BA tabulation spreadsheet that revealed which response were significant (at the p < 0.05 level) and should be addressed by the behavior change strategy. For each question and category of responses, the BA tabulation calculates the percentage of responses for both doers and non-doers; the odds ratio, the standard error, and its confidence interval; the estimated relative risk (ERR) (Zhang & Yu 1998); and p-values. This allows the user to identify those differences between doers and non-doers that are statistically significant (at p < 0.05) and to see the strength of those associations between each response and the behavior (based on the ERR).

We performed all procedures in studies in accordance with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The study was approved by the senior management of GCC in consultation with higher leadership. The respondents were informed of the study objectives prior to taking part in the interviews, and all respondents were asked to give their written consent before they were interviewed. All respondents had the right to withdraw from the interview at any moment without jeopardizing their access to any services. No identifying data such as name, date of birth, or address was collected from the respondents.

The characteristics of the study respondents are shown in Figure 1. The majority of the respondents were male (71 and 58% of doers and non-doers, respectively), and most of them belonged to the 18–25 years of age group (29 and 29%, respectively). In terms of educational attainment, the majority of the respondents had completed education until 10th grade (36 and 49% of doers and non-doers, respectively), while the dominant profession was service jobs followed by small business.

The results are presented based on the highest, statistically significant differences in responses and beliefs between doers and non-doers of HWWS after defecation. All quantification of associations mentioned below are based on the ERR. The detailed results are included in Supplementary Material, file 2. These beliefs and determinants are discussed below.

When responding to the open-ended question on what makes (or would make) the behavior easier (see Table 1), respondents mentioned access issues. Doers were 22.9 times more likely to mention availability of soap (p < 0.001) than doers. Non-doers were 26 times more likely to mention lower price of soap (p < 0.001). When responding to the open-ended question on what makes (or would make) the behavior difficult, non-doers were 7.1 times more likely to mention (than doers) an inability to buy soap (p = 0.008), an access issue. Doers were 25.2 times more likely to mention availability of water as something that makes the behavior easier. However, when asked directly (using a closed-ended question) how difficult it was to get water for HW, the responses of doers and non-doers were not statistically different (with 11% of doers and 7% of non-doers saying that it was difficult or very difficult). Also, based on household observations, doers were found to be 2.9 times more likely to have a permanent HW facility in the home (p < 0.001) and 2.1 times more likely to have a HW facility within 10 feet of a latrine or toilet (p = 0.01) than non-doers, and non-doers were 2.4 times more likely to not have a covered bucket for HW purposes (p = 0.01) than doers.

When asked specifically about access (see Table 2), non-doers were 8.9 times more likely to say that it was ‘very difficult’ to get soap to wash hands (p = 0.002), and 5.1 times more likely to say that it was ‘somewhat difficult’ (p < 0.001). Conversely, doers were 9.7 times more likely to say that it was ‘not difficult’ to get soap (p < 0.001). When asked what makes it difficult to get soap, non-doers were 18.7 times more likely to mention a financial crisis in their family (p < 0.001), and 15.1 times more likely to mention unavailability of soap at nearby markets (p < 0.001) than non-doers.

When responding to the closed-ended question on whether it was easy to remember to do the behavior (see Table 3), doers were 8.5 times more likely to say that it was ‘not difficult at all’ to remember (p < 0.001) than non-doers. Non-doers were 15.2 times more likely to say that it was ‘very difficult’ (p < 0.001) and 3.4 times more likely to say that it was ‘somewhat difficult’ to remember (p < 0.000) than doers.

When asked if most people the respondent knew approved (or would approve) of their washing their hands with soap after defecation (see Table 3), doers were 12.9 times more likely to say yes (p < 0.001) than non-doers. When asked who the people are (or would be) that approve (or would approve), doers were 7.7 times more likely to say household members (p < 0.001), 3.2 times more likely to say teachers (p < 0.001), and 3.9 times more likely to say NGO agents (p < 0.001).

When asking the respondents how serious it would be if he or she got diarrhea in the coming months (see Table 4), doers were found to be 2.7 times more likely to say that it would be ‘very serious’ (p = 0.001) and non-doers were found to be 5.9 times more likely to say that it would be ‘not serious at all’ (p < 0.001). When asked if the respondent believed that it was Allah's, God's, or the gods’ will that people get diarrhea, non-doers were 3.7 times more likely to believe that it is Allah's, God's, or the gods’ will (p = 0.002).

Overall, in response to an open-ended question about the advantages of performing this behavior (see Table 5), doers were 3.3 times more likely to say ‘feeling clean’ (p = 0.045) and 2.6 times more likely to say being ‘free from illness’ were advantages of the behavior. When asked about the disadvantages of performing this behavior, non-doers were 2.8 times more likely to mention ‘increasing costs’.

Non-doers were 2.5 times more likely to say that it was ‘not likely’ that they would get diarrhea (p = 0.004) if they did not wash hands, and doers were 2.3 times more likely to say that was ‘somewhat likely’ (p = 0.005) (Table 6).

This BA study on HWWS after defecation revealed important differences in responses and beliefs between doers and non-doers regarding ten different (but sometimes linked) behavioral determinants of HWWS in this urban setting of Bangladesh. The sheer number of important determinants for this behavior may help to explain why HWWS has been a difficult behavior to scale in many countries. Even if one or two determinants or barriers are addressed, there are a multitude of important determinants and barriers that may make it difficult for a person to adopt the behavior.

A study like this that examines associations with behavioral adoption cannot definitively identify the true drivers of the behavior, but can provide clues as to the possible drivers of a behavior and the interventions that may be effective in promoting adoption. The factors most highly associated with HWWS in this population – based on responses to questions on the 12 determinants studied – were around beliefs and responses regarding self-efficacy and access. Randomized controlled trials and population-based research have shown that integrating handwashing facility into WASH programs may increase HWWS (Ram et al. 2017) and access to soap and water is a vital component in assuring HWWS in many contexts (Hoque 2003; Jumaa 2005). In terms of access to water, our baseline study found that almost 86% of the households had a permanent water facility, but access to soap was found to be a major barrier in this population. Doers were much more likely to say that they could do the behavior with their current knowledge, skills, and resources than non-doers. Key enablers included availability of soap (especially low-priced soap) and money (presumably for purchasing soap), and key barriers included unavailability of soap in local markets and difficulty in remembering to buy soap for handwashing. When asked the open-ended question, ‘what makes it easier?’, doers were more likely to mention that having water made it easier (than non-doers), but when asked specifically ‘how difficult is it to get water that you need to wash your hands every time after defecation?,’ there were no significant differences between doers and non-doers. Non-doers were more likely to say that getting soap was very difficult when asked the similar closed-ended question, so soap availability appears to be a more important barrier than water access in this population. Findings from the observation on handwashing facilities show that having a permanent HW facility in the home that is less than 10 feet from a latrine or toilet, and having a covered bucket of water for HW purposes are important, as well. In this urban context, poorer people are living in the slum setting where access to a handwashing station with soap has been found to be a major barrier in this and other studies (Hoque 2003). Also, it is widely accepted that access to a handwashing facility is an important determinant of HWWS in lower-income settings globally (Zangana et al. 2020) and having a handwashing facility is a critical determinant of HWWS in critical times including after defecation (White et al. 2020). These findings are similar to our study.

Cues for action – specifically, being able to remember to do the behavior – is another important behavioral determinant of HWWS in this population. Results of our study showed that non-doers were more likely to say that it is very difficult to remember to perform the behavior and represents a barrier for them. This result is aligned with other previous WASH studies of Bangladesh (Rahman et al. 2017; Parveen et al. 2018).

Perceived social norms were found to be a powerful determinant of HWWS in this study. In addition to doers being much more likely to say that ‘most people approve’, they were more likely to mention key groups who they believe approve: household members, NGO agents, and teachers. These groups should be reached with messaging and used as key influencers of the behavior in this context. Other studies have found that perceived social norms and self-efficacy can be powerful factors in adoption of WASH behaviors (Parveen et al. 2018) and determinants associated with the HBM are often found to be driving adoption of HW behavior (Zeigheimat et al. 2016). Perceived severity of diarrhea was very important, with non-doers being much more likely to say that diarrhea is ‘not serious at all’. Using testimonials of people who have had diarrhea and caregivers who have lost children to diarrhea may be effective in changing this erroneous perception.

Perceived divine will is a determinant that is usually studied with BA but has not been systematically studied in many other studies of determinants of HW, despite the fact that the HBM has been used for many years to suggest various influences on health care, including religion (Kirn 1991). This determinant was also found to be an important driver in this study: Non-doers were much more likely to believe that it was Allah's, God's, or the gods’ will when one gets diarrhea. Reminding people who adhere to Islam of key verses that promote handwashing in the Quran and texts from other holy books may be useful for other religions in Bangladesh to address this determinant.

Aligned with another study in Bangladesh (Parveen et al. 2018), the current study found several positive consequences of HWWS after defecation as important enablers of the behavior. Doers of this behavior were much more likely to say that HWWS after defecation enables them to ‘feel clean’ and keeps them ‘free from illness’.

This study has a number of limitations. First, given that this study was only done in a limited urban area with a limited sample size, the results should not be generalized to the rest of the country. Since 67 individual comparisons were made between doers and non-doers response during data analysis, we chose to focus on findings where p < 0.01 rather than p < 0.05 to reduce the chance of identifying spurious findings as important ones. This, of course, does not entirely reduce this risk, so the findings presented here should be considered less than definitive especially for any individual finding. Second, by design, the BA approach does not consider respondents' socio-economic information which may lead to some confounding or interaction of variables. Finally, as another means to make the tool easy to use, the questions on some determinants that are assessed are closed-ended and may lead to some bias that may be avoidable through use of more open-ended questions. However, applying these changes to the BA methodology could make it harder to use with less personnel, time, statistical skill, and funding for formative research.

An effective behavior change strategy for HWWS after defecation is required to address multiple beliefs and behavioral determinants, reducing barriers and leveraging enablers identified in this study. For example, an intervention to improve access to low-cost soap could be educating the population (and doing demonstrations) on how to use soapy water as a means to substantially reduce the cost of handwashing, as has been done in other projects in Bangladesh. To increase cues for action, reminder stickers could be put in key places in the home (e.g., in sight when using the latrine or toilet) and designated handwashing stations could be encouraged. To increase perceived positive social norms, people – especially those found to be important influencers of this behavior (e.g., teachers, NGO agents) – can be videoed modeling and giving testimonials in each neighborhood on why and how they do the behavior and distributed over social media and other means. To increase the perception that diseases linked with not HWWS are serious (to address perceived severity), country-wide data on deaths from diarrhea and other WASH-related diseases could be disseminated, and testimonials by people who have lost or almost lost family members due to WASH-related diseases could be used. To address perceived divine will, religious leaders could be assisted in creating sermon outlines using verses from the Quran or other sacred texts that promote HWWS, and assisted in creating radio spots on this. Essay contests could be used to encourage people to name positive consequences of the behavior. To increase perceived action efficacy, information on the effectiveness of HWWS in decreasing diarrhea, pneumonia, cholera, COVID-19, and other diseases could be disseminated using social and traditional media.

A host of important behavioral determinants were found to be associated with HWWS after defecation in this population. A broad set of activities and messages – based on key beliefs and responses associated with these determinants – will most likely be needed to significantly scale HWWS in this and other populations. Based on the HBM, PBC, and TRA, the BA formative research tool is an important and easy-to-use tool for rapidly identifying beliefs and behavioral determinants of key WASH and other sector behaviors.

We acknowledge Gazipur City Corporation for their approval and support of this study. We also thank the respondents who participated in the study, and to the data collectors who conducted the interviews. Finally, we would like to thank our donor.

The study was funded by the Dutch Government (DGIS) under the WASH SDG program through the initiative, ‘Development of Integrated Wastewater Management System for Gazipur City Corporation (GCC)’. However, the donor did not have any role in the study design, interpretation of the results, or selection of the journal.

All relevant data are included in the paper or its Supplementary Information.