Background
Water, sanitation, and hygiene (WASH) interventions may include home water treatment, latrine construction, and handwashing promotion, typically promoted by community health workers (CHWs). Due to demonstrated effects of WASH interventions on child health outcomes, such as the effect of handwashing on diarrhoea [
1], they have received considerable funding over the years from governments in and donor organizations for low- and middle-income countries [
2‐
5]. However, a considerable research agenda remains regarding (1) the selection of specific WASH technologies and behaviours for inclusion in large-scale, routine programs and (2) how to achieve levels of use and maintenance of technologies that affect sustained adoption of behaviours under routine programmatic conditions sufficient to produce health impact.
Intervention efficacy, effectiveness research and implementation research studies that involve CHWs, particularly to impart behaviour change, differ. There are different uptake goals, CHW-to-population ratios and systems for CHW management and supervision dependent on the size and scope of the study (Table
1). WASH efficacy studies typically examine the effects of specific WASH technologies and behaviors on outcomes of interest in order to inform their subsequent inclusion in routine programs (Table
1). To conduct efficacy studies, CHWs deliver different combinations of WASH interventions to the target population under optimal or ideal conditions (Table
1). In efficacy studies CHWs take great efforts to ensure that the behaviour-enabling technology is present and functional at the household level, and that they are being used (Table
2). This may include free distribution of the technology, and frequent visits by the CHWs to ensure adoption of the behaviours and address difficulties encountered with either the technologies or the behaviours. In contrast, to examine how to achieve high uptake of WASH interventions under near real-world and real-world conditions, researchers build on outcomes from efficacy research and conduct effectiveness research and implementation research (Tables
1 and
2).
Table 1
Comparison of uptake goals and implementation procedures for three categories of research on behaviour change interventions delivered by community health workers (CHWs)
Efficacy research | Focus on internal validity: can we draw causal inferences between interventions received and outcomes observed? | Highly controlled, farther from real-world conditions | Technology uptake: optimal Behavioural uptake: optimal | Less than 1/100, not a real-world ratio | Continuous oversight, typically 2–3 times per month. CHWs replaced within 1 month of attrition or critically low performance |
Effectiveness research | Focus on external validity: can the results be generalized to programmatic settings with near-real world conditions? | Less controlled, near real-world conditions | Technology uptake: near optimal Behavioural uptake: routine | Ratio based on national Ministry of Health (MOH) policy. Ongoing technical support from NGO staff3 | Periodic oversight, typical monthly or less. Facilitation of problem resolution by non-governmental organization (NGO) |
Implementation research | Focus on external validity: can the results be generalized to programmatic settings with real-world conditions? | Not controlled, real-world conditions | Technology uptake: routine Behavioural uptake: routine | Ratio based on national MOH policy. Limited technical support from NGO staff4 | Oversight depends on research question Limited external facilitation of problem resolution |
Table 2
Content of behaviour-change interventions delivered by community health workers (CHWs) to achieve different levels of uptake
Optimal / ideal | • Technology distributed by CHWs/ project • CHWs provide users with information on use and maintenance of the technology • CHWs conduct frequent home visits to verify that technology is functional and in good working order at all times, and intervene if technology ceases to function | • Behavioural recommendations explained by CHWs during one or more initial home visits • CHWs conduct frequent home visits to verify that people follow behavioural recommendations, and intervene/problem solve if they do not follow the recommendations |
Routine | • Technology distributed by CHWs/ project • CHWs provide users with information on use and maintenance of the technology • CHWs do not intervene directly to ensure functionality | • Behavioural recommendations explained by CHWs during one or more initial home visits • CHWs do not intervene directly to ensure that people follow the behavioural recommendations |
This article describes the intervention delivery system adopted in the WASH Benefits study in rural Bangladesh, a large-scale efficacy trial of different combinations of WASH and child nutrition behaviour-change interventions delivered by CHWs [
6]. The efficacy trial found that nutrition only and nutrition plus WASH interventions improved linear growth and decreased diarrhoea among recipient children [
7]. The WASH Benefits interventions were selected from 2 years of pilot studies and the delivery methods aimed at sustained uptake are described in this paper, part of a three-paper series on WASH Benefits Intervention Delivery and Performance [
8,
9]; the paper by Rahman et al. [
8] describes fidelity measurement methods and monitoring; the paper by Parvez et al. [
9] describes methods and findings for intervention uptake. In this paper we describe both the intervention delivery system necessary to achieve the high level of uptake, and the difficulties encountered during intervention delivery and how they were addressed.
Results
Implementation monitoring highlighted both successes and shortcomings. Fidelity benchmarks were attained by the fourth intervention assessment month demonstrating moderate to high technology and behavioural uptake, described in the companion paper on monitoring WASH Benefits intervention coverage and quality [
8]. This was sustained throughout the program period, described in the companion paper on WASH Benefits intervention uptake [
9] for single and combined intervention recipient households. Similar technology and behavioural uptake indicators were detected by the external monitoring agency [
22‐
24]. Periodic internal CHW monitoring resulted in discontinuation of a small number of low performers (
n = 33). During the intervention period, a total 156 CHWs discontinued service, the most common reason being migration of some type (moved with family, moved abroad, moved after marriage;
n = 50).
We attempted to address problems as they were detected by the Intervention Delivery Team. The first and most important difficulty was developing a way to manage an intensive training timetable. Considering the complexity of the intervention design and distribution of CHW by intervention type and arm, we initially engaged a professional trainer group to lead the CHW training. They developed the training guideline and communication package and we assigned them to conduct the first batch of training. Despite their expertise in training activities, we observed that the trainer group was not effectively engaging trainees. We found that their highly structured conventional methods likely did not fit with the varied trainee background, particularly for an inexperienced group of selected CHWs. We also understood that CHW motivation would be key and the professional trainers seemed lacking. Thus, we developed an internal training resource group from among our team, thoroughly analyzed the gaps and revised the training methods and materials. Moreover, supervisors were found to be initially minimally engaged with the training and their minimal involvement in training was a lost opportunity for becoming acquainted with intervention detail and developing rapport with the CHWs. Additionally, based on the intervention roll-out timeline, we anticipated that CHW training would take approximately 12–13 months so we used the train the trainer method (described earlier). During plans for training we detected that it was important to further define supervisory roles, hence the development of the Training Manual (described earlier).
Another difficulty was the need to deliver six different interventions, and the potential distance between randomly assigned study clusters. If a CHW had to deliver an intervention in more than one cluster, it would have to be a geographically proximate one; there would have been the potential for CHWs to become confused about message delivery.
Early in the roll-out period, the Intervention Delivery Team detected issues with the pace of hardware installation which could impact the intervention timeline. Latrines were installed for 4533 households and involved numerous construction and quality assurance steps and were thereby delivered independently from the other sanitation technologies and behavioural promotion. Thus, latrine roll out began earlier than other sanitation and other interventions due to an anticipated longer installation period.
To maximize uptake throughout the intervention period, new behaviour-change activities were developed, to address sub-optimal practices detected from structured observations, and to address intervention fatigue reported by CHWs during monthly meetings. Initiatives included further technology use and behaviours that related to the index child age, increasing self-efficacy, increasing roles for men, and decreased emphasis on behaviors that were commonly being practiced. During a 6-month period of political instability, mid-intervention (January to June 2014), we ensured field staff safety through close oversight by the Intervention Delivery Team members at field offices with staff accommodation facilities.
Discussion
WASH Benefits Bangladesh study intervention delivery built on lessons from prior implementation research [
18] and effectiveness research studies [
16,
18], to guide development of the intervention delivery system for this efficacy study. As reported in the companion papers in this issue of the journal, the implementation fidelity and uptake data demonstrated that WASH behaviour change met targets considered sufficient for a large-scale efficacy trial [
8,
9]. Thus, when interventions were assessed for impact on linear growth and childhood diarrhoea, the WASH Benefits study demonstrated that children from households that received nutrition alone or combined with WASH interventions had significantly greater linear growth. Additionally, children from households that received any of the interventions, with the exception of water treatment, had significantly lower prevalence of diarrhoea. Intervention effects in an efficacy trial cannot be determined when intervention uptake is low.
The high intervention uptake achieved in the WASH Benefits Bangladesh study may be attributed to several factors. Prior formative research and pilot trials provided researchers with some clear choices on attractive, durable enabling technologies and supplies that were provided to enrolled households and would likely encourage behavioural uptake; choices and behaviour-change promotion were underpinned by a theory- and evidence-based behaviour-change strategy. We had conducted community pilots for the ICVB effectiveness study, but behavioural uptake was considerably lower [
16]. However, the ICVB pilot period was shortened by an intervention delivery deadline, in line with vaccine delivery, which meant that we were less able to troubleshoot community and household-level problems, particularly with the liquid chlorine-based water treatment intervention. Moreover, in the ICVB intervention, several adjoining unrelated households were expected to share study-provided technologies and supplies [
17]. Lower uptake was possibly attributed to additional constraints in urban, low-income environments [
13].
CHW-accessible supervision has been highlighted as critical for successful CHW-delivered interventions [
25‐
27]. Particular attention was given to CHW supervision, management, and performance monitoring, based on limitations detected during the SHEWA-B implementation research study, which detected low uptake [
2,
19]. SHEWA-B involved more than 10,000 CHWs, each expected to cover 450–550 households [
2], who were hired and supervised by 40–60 NGOs during the 5-year intervention period. In an effectiveness research study conducted in India, six different NGOs were contracted to deliver the intervention, and similarly sub-optimal behavioural uptake was found [
28]. In the ICVB effectiveness study, a single NGO hired and supervised CHWs, with monthly input from the study research team; however, CHW supervision problems were not totally avoided. In the WASH Benefits Bangladesh efficacy study, the Intervention Delivery Team directly supervised CHWs ensuring frequent interaction, and greater control over intervention delivery.
Training can impact CHW performance [
25‐
27,
29,
30]. The advantages of the training program in this study were that it was developed in collaboration with the research team which had conducted the formative research and pilot studies, and it comprised a train the trainer method for CHW supervisors, longer duration CHW trainings, monthly sessions with the training officer, and refresher sessions driven by adjustments in the behaviour-change strategy. Monthly interactive training meetings that encouraged and acknowledged problem identification likely assisted job satisfaction. In contrast, the SHEWA-B program CHWs received 15 days of initial training in 2007 and 12 total days of refresher training delivered by third-party NGOs between 2009 and 2015 [
19].
The supervisor-to-CHW and CHW-to-household ratios in this efficacy study were considerably lower than typical programs and thus facilitated open communication channels. CHW workload has been examined with respect to program delivery quality, and studies have demonstrated that including additional tasks may not compromise quality [
31], but heavy workload can impact motivation, performance, and job satisfaction [
27]. These indicators were all high among our CHWs, when assessed in a qualitative study (manuscript in preparation).
Non-monetary rewards can contribute to morale and performance [
25‐
27]. For the majority of our CHWs this position was the first paid employment; we used initiatives that we thought would enhance prestige in the community and general job satisfaction while minimizing CHW turnover. We provided ID badges on lanyards, signifying that they were working women and associated with a known and respected health organization. We ensured timely stipend payments through mobile phone networks; frustration over delayed payments was reported by CHWs during the SHEWA-B impact assessment [
19].
The intervention delivery system brought together a theory- and evidence-based behaviour-change strategy (Leontsini et al., manuscript in preparation), regular staff interaction at each level of the Intervention Delivery Team, continuous quality improvement principles, and learning from other WASH trials in Bangladesh (Table
3). The co-location of intervention delivery and research staff is relatively uncommon, and likely allowed each group to share and understand the intervention delivery constraints and strengths, ensuring continuous communication and a rapid response to problems identified in the field. An alternative strategy could include assembling research and delivery teams with key program stakeholders in the field to develop and revise a Program Impact Pathway, as described for nutrition interventions [
32‐
34].
An important limitation of our intervention delivery was cost; this was an efficacy trial. Provision of enabling technologies and supplies free of cost to households is not feasible for larger-scale implementation. Opportunities for savings include revising technologies to include simpler hardware, providing technology and product subsidies, or encouraging households to provide some of their own components [
35], which may increase feasibility. Intensive behaviour-change activities using frequent one-on-one visits were also expensive and not feasible for larger efficacy trials or implementation programs. However, some aspects of the intervention delivery system, such as communication approaches that focus on developing problem-solving skills rather than didactic transfer of information and lower supervisor-to-CHW ratios could be adopted at a moderate cost.
Conclusions
Achieving sufficient WASH intervention uptake is attainable; efforts to reduce intervention costs need further exploration. Prior studies have demonstrated that lower intensity is cheaper [
36]. However, studies that compare the impact of behavioural and technology uptake at different CHW-visit intensity levels have not been conducted. Regular training that includes office-based interactive sessions with CHWs and their supervisors is costly. Potentially cheaper electronic training (e-training) methods have been employed in high-income countries [
37,
38] and will likely appeal to increasingly technology-savvy members of low-income countries where mobile phone and smartphone penetration is high and increasing.
Access to, and interaction with, supervisors has been described previously as impacting on performance [
25,
26]. The supervisor-to-CHW and CHW-to-household ratios in WASH Benefits Bangladesh that facilitated open communication channels were considerably lower than typical programs. Exploring effectiveness of higher ratios on CHW performance is needed.
Sustainability of a program beyond program staff presence is an important intervention objective [
39], typically addressed in implementation research studies. The extent to which CHW performance impacts uptake beyond the promotion period is an important area for future research.