Policy Research Working Paper 11070
Effects of a Community-Driven Water, Sanitation,
and Hygiene Intervention on Diarrhea,
Child Growth, and Local Institutions
A Cluster-Randomized Controlled Trial in Rural
Democratic Republic of Congo
John P Quattrochi
Kevin Croke
Caleb Dohou
Luca Stanus Ghib
Yannick Lokaya
Aidan Coville
Eric Mvukiyehe
Development Economics A verified reproducibility package for this paper is
Development Impact Group available at http://reproducibility.worldbank.org,
February 2025 click here for direct access.
Policy Research Working Paper 11070
Abstract
Diarrhea and growth faltering in early childhood reduce The percentage of villages in the intervention group with
survival and impair neurodevelopment. This paper assesses an active water, sanitation, and hygiene (or just water)
whether a national program in the Democratic Republic committee was 21 percentage points higher than the con-
of Congo reduced diarrhea and stunting and strength- trol group. Households in the intervention group were 24
ened local water and sanitation institutions. The program percentage points more likely to report using an improved
combined (i) funds for latrine and water upgrades, (ii) water source, 18 percentage points more likely to report
institutional strengthening activities, and (iii) behavior using an improved sanitation facility, and reported more
change campaigns. In 2018, the program was randomly positive perceptions of water governance. The Democratic
assigned, after stratifying by province and cluster size, with Republic of Congo’s national rural water, sanitation, and
50 intervention and 71 control clusters. In 2022–23, 3,283 hygiene program increased access to improved water and
households were interviewed, at a median of 3.6 years sanitation infrastructure, and created new water, sanitation,
post-intervention. The intervention had no effect on diar- and hygiene institutions, all of which persisted for more
rhea and no effect on length-for-age Z-scores in children. than three years. However, these effects were not sufficient
Villages in the intervention group had a 0.40 higher score to reduce diarrhea or growth faltering.
on the water, sanitation, and hygiene institutions index.
This paper is a product of the Development Impact Group, Development Economics. It is part of a larger effort by the
World Bank to provide open access to its research and make a contribution to development policy discussions around the
world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be
contacted at acoville@worldbank.org. A verified reproducibility package for this paper is available at http://reproducibility.
worldbank.org, click here for direct access.
RESEA
CY
LI R
CH
PO
TRANSPARENT
ANALYSIS
S
W
R
R E
O
KI P
NG PA
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.
Produced by the Research Support Team
Effects of a Community-Driven Water, Sanitation, and Hygiene Intervention on Diarrhea,
Child Growth, and Local Institutions: A Cluster-Randomized Controlled Trial in Rural
Democratic Republic of Congo
John P Quattrochi 1, Kevin Croke 2, Caleb Dohou 3, Luca Stanus Ghib3, Yannick Lokaya3, Aidan
Coville3, Eric Mvukiyehe4
1 Graduate School of Arts & Sciences, Georgetown University
2 Department of Global Health & Population, Harvard TH Chan School of Public Health
3 Development Impact (DIME) Department, World Bank
4 Department of Political Science, Duke University
Corresponding author is Aidan Coville (acoville@worldbank.org). This paper was made possible through
collaboration between the Foreign, Commonwealth, and Development Office (FCDO-DRC), FCDO’s Evaluation Unit,
the United Nations Children's Fund (UNICEF-DRC), the VEA Coordination Team from DRC’s ministries of Health and
Education, and the World Bank's Development Impact Department (DIME). We thank the continued leadership,
and persistent efforts of several members from these organizations. This study was funded by FCDO - DRC under
the World Bank-administered i2i Trust Fund. DIME Analytics conducted a reproducibility review of the data and
results in this paper. The study received ethical clearance from IRB Solutions (Protocol \#2019/10/20) and from the
Research Center for Health Promotion (CRPS) Institutional Ethical Committee (CEI) of the Institut Supérieur des
Techniques Médicales de Bukavu (ISTM-Bukavu) in the DRC, and a Health Research Approval from the DRC
Ministry of Health. The findings, interpretations, and conclusions expressed in this paper are entirely those of the
authors. They do not necessarily represent the views of the World Bank and its affiliated organizations, or those of
the Executive Directors of the World Bank or the governments they represent.
2
Introduction
The most recent estimates of the global burden of morbidity and mortality attributable to unsafe
water, sanitation, and hygiene (WASH) are that 1.4 million deaths and 74 million disability-
adjusted life years lost could have been prevented in 2019 [1]. People living with unsafe WASH
have higher exposure to fecal-oral pathogens, resulting in enteric dysfunction, diarrheal illnesses,
and, in children, growth faltering. Growth faltering, in turn, has long-term negative impacts on
health, cognition, and human capital [2,3]. In 2020, 2.0 billion people did not have access to safely
managed drinking water services, 3.6 billion did not have access to safely managed sanitation
services, and 2.3 billion did not have access to handwashing facilities with soap and water at
home [4,5]. While access has been increasing, progress will need to accelerate by three-to-six-
fold to meet the Sustainable Development Goals for 2030 [6]. The challenge of increasing access
is particularly acute for people living in or near armed conflict – one in six people worldwide – both
through the direct effects of conflict and because violence and insecurity impede collective action
to provide public goods like WASH [7,8]. In the Democratic Republic of Congo (DRC), as of 2020,
48 million people still lacked basic drinking water services, 11 million people still practiced open
defecation, and 72 million people still lacked basic hygiene services [9].
To increase access to safe WASH, governments and donors have increasingly turned to
community-led approaches. While WASH experts called for greater community participation for
over 30 years [10], the sector did not fully embrace this approach until the late 1990s. Beginning
with community-led total sanitation in Bangladesh in 1999, community-led WASH programs have
been implemented in at least 60 countries, and 15 countries have incorporated them into national
policy [11,12].
Despite this broad adoption, the health effects of community-led WASH interventions – and of
many WASH interventions in general – remain poorly understood. The accumulation of evidence
has accelerated in recent years, but the length of the causal chain from the intervention to the
outcome, and the vast design space for WASH interventions (which can incorporate behavior
change campaigns, infrastructure, institutions, and/or new technologies), means that many
fundamental questions remain unanswered. A meta-analysis of 13 randomized WASH trials found
no effect on child length-for-age, but a meta-analysis of 124 WASH studies (randomized and
observational) found a protective effect against diarrhea [11,13–24]. There is a great deal of
heterogeneity in effect size across studies, likely due to variation in intervention components and
intensity, and to the influence of contextual factors such as baseline exposure to fecal matter.
To our knowledge, this is the first trial of an intervention that combines the creation of new
institutions, funding for new or improved infrastructure, and a behavior change campaign, all
within a locally-led process of targeting and implementation. We study this complex intervention
as it is implemented at scale, providing a realistic estimate of effectiveness for policy makers in
similar contexts. Our follow-up period is unusually long (3.6 years), enabling us to address the
question of sustainability. We also provide evidence in a conflict-affected setting, where WASH
interventions have rarely been evaluated using experimental designs.
3
The intervention is the ‘healthy villages’ component of the DRC’s Healthy Villages & Schools
program, co-led by the Ministries of Public Health, and of Primary, Secondary, and Professional
Education, with support from the United Nations Children’s Fund (UNICEF). Since 2008, nearly 9
million people in almost 11,000 villages have been reached with WASH services through the
program. Healthy Villages & Schools was the largest WASH program implemented by UNICEF
globally and comprised 90% of total external funding committed to rural WASH in the DRC from
2005 to 2020 [25]. Our goal was to estimate the effect of Healthy Villages & Schools on diarrhea
prevalence, child length-for-age, and WASH institutions.
Methods
Study design and participants
The intervention of interest is a DRC government-run program that began several years before
our study. For our study, the government agreed to randomly assign the next phase of the
program (i.e. the next group of villages to receive the intervention). In 2018, we randomly assigned
groups of villages to intervention or control (details below). Since we did not collect any data prior
to randomization, we did not yet register the study. We collected the first round of data in late
2019, about 5 months after the intervention was implemented (implementation took about one
year in each group of villages) [26].
We pre-registered the analysis of that first round of data collection (5 month follow-up), before
any of the authors saw the data, in the American Economics Association (AEA) Registry
(AEARCTR-0004648) (https://www.socialscienceregistry.org/trials/4648).
In Feb 2021, we registered plans for additional data collection in the Pan-African Clinical Trials
Registry (PACTR202102616421588;
https://pactr.samrc.ac.za/TrialDisplay.aspx?TrialID=14670). In April 2023, before the PIs had
seen any data for the 3.6-year follow-up (i.e., the data for the current manuscript), we updated
the registration with our primary and secondary outcomes. In April 2023, we also posted our pre-
analysis plan for the 3.6 year follow up on the AEA Registry (see "Three Year Follow up Pre-
Analysis Plan" at https://www.socialscienceregistry.org/trials/4648). All planned studies from this
project are now registered and any future work will be registered prospectively.
This study is reported as per CONSORT guideline (S1 Text).
We worked with intervention implementers to design a cluster-randomized trial in rural villages in
five DRC provinces: Kongo Central, Kasai, Kasai Central, North Kivu, and South Kivu. The
implementers identified 403 candidate villages in which the intervention could be launched during
the study period, based on the established criteria for the intervention: that the village was located
in a secure and accessible Health Area that was not already served by the WASH Consortium,
the Health Area staff were dynamic and interested in participating, and there was a problem of
diarrhea, cholera, and/or malnutrition. Among these villages, 34 already had program activities in
process before research activities began, leaving 369 eligible villages.
4
To avoid spillover effects from treatment villages to control villages, we grouped those villages
into clusters. We considered any villages within 2.5 km of each other (using Euclidean distance
between village centroids) to be part of the same cluster. Therefore, all clusters have at least 2.5
km between them. We relax this rule in South Kivu, where density is greater, and use a minimum
distance of 1km. In total, this resulted in 124 clusters. North Kivu had only three clusters (covering
30 villages); as a result, it was not logistically feasible to include these villages in the trial. That
left 121 clusters (339 villages) in four provinces.
Each village in the intervention clusters received the intervention, as described above. Villages in
control clusters did not receive any intervention. Data collection procedures were identical in the
two groups.
The study protocol was approved by the Institutional Ethics Committee of the Institut Superieur
des Techniques Médicales de Bukavu (DRC) (#001/2019 & #008/2022) and by Solutions IRB
(USA) (#2019/10/20). With direction from the study investigators, Innovative Hub for Research in
Africa (IHfRA) was responsible for data collection. No study data was collected by intervention
implementers.
All residents of the targeted villages who had lived in the village for at least four years (i.e., moved
to the village prior to the intervention) were eligible to participate in the study. Two groups of
households were interviewed: households that were interviewed at 5 months post-intervention (4
per village) and households that had not been previously interviewed (6 per village). Both groups
were randomly selected (at their respective entries into the study) as follows: from the center of
the village, interviewers went in opposite directions to the nth household, where n was a randomly
selected number between 1 and 20. We interviewed the head woman of the household. We also
asked to measure the height and weight of the youngest child aged 2-5 years, or, if none, the
oldest child aged 0-2 years. Adult participants provided informed consent verbally, which was
recorded electronically.
Randomization and masking
A total of 339 villages in 121 clusters were eligible for randomization. In all provinces except Kasai
Central, each cluster was given equal probability of being selected for treatment or control. In
Kasai Central, due to budget constraints, we increased the probability of being selected into the
control group to 75%, to reflect the fact that only 16 out of 81 villages could receive the
intervention. Thus the allocation probability for the intervention group was 25%.
We block randomized by province and number-of-villages-per-cluster (12 blocks total; see S1
Table). Since randomization was based on clusters but the implementing organization’s
operational targets were based on villages, it was not possible to force the randomization to select
the exact number of villages targeted without introducing potential bias. Instead, we compared
the number of target villages per province to the number of treatment villages selected after
randomization. In cases where the number of intervention villages was larger than the operational
target, we randomly dropped an equal number of intervention villages from the largest control and
intervention clusters until operational targets were met. We dropped 2 villages in Kongo Central
and 4 villages in Kasai. We also dropped one control village in Kasai due to a coding error. This
left 146 villages in 50 clusters in the intervention group, and 186 villages in 71 clusters in the
5
control group. S1 Table shows how intervention and control villages and clusters are distributed
across provinces. Randomization was done by the research team in Stata.
Due to the participatory and visible nature of the intervention, neither participants nor data
collectors were masked to treatment status. However, data collectors did not participate in
intervention implementation and were employed by a separate, independent organization.
Procedures
The intervention, “Healthy Villages & Schools”, was developed by the DRC government and
UNICEF. We focused on the village rather than the school component. This program mobilizes
communities to become a “Healthy Village” with 3-6 months of support from government health
officials and local NGOs, including approximately $2,000 of financing for new or improved water
infrastructure, $2,000 for new or improved sanitation infrastructure, and $3,000 for personnel
costs, per village. The mean village size in the intervention group was 456 people (median 400;
IQR 502). The seven norms to become a Healthy Village are:
1. There is a dynamic village WASH committee.
2. At least 80% of the population has access to safe drinking water.
3. At least 80% of households use a hygienic latrine.
4. At least 80% of households dispose of their household waste hygienically.
5. At least 60% of the population washes their hands before eating and after going to the
latrine.
6. At least 70% of the population is aware of fecal-oral disease transmission and how to
prevent this.
7. The village is cleaned at least once a month.
The program is implemented in nine steps (Table 1) [27].
Table 1. The Healthy Village and Schools Program’s Nine Steps
Step Description
0 The community learns about the program and collectively decides to adopt it before submitting a formal
request to the relevant Health Zone. (A Health Zone is a geographic unit of the Congolese health system
that contains roughly ten Health Areas and 100,000 residents, run by a Chief Medical Officer (CMO)).
Program protocols state that the entire community should be involved in the decision to participate.
1 A statement of agreement between the community and the Health Zone is signed.
2 Health Zone officials survey 19 households on knowledge, attitudes, and practices (KAP). The community
self-evaluates on eight practices, including handwashing, water use, and sanitation.
3 The community spends about 11 hours over five days creating calendars and maps, visiting water points,
classifying hygienic practices as healthy or unhealthy, discussing fecal-oral disease transmission,
calculating medical costs, and assessing which individuals and organizations influence sanitation and
hygiene in the community. This includes 1.5-2 hours in a facilitated activity around the question, “What are
the hygiene practices that we want to change in our village?”
6
4 The Health Zone provides training for 20 volunteers on maintenance of latrines, water supply systems, and
sanitation, conflict management, and petty cash management. The community elects a village WASH
committee.
5 The community spends ten hours over three days describing a community vision, analyzing the barriers to
reducing diarrheal diseases, choosing improvements to drinking water, sanitation, and hygiene, and
formulating an action plan. The community is asked to identify practical, low-cost solutions with a minimum
of outside assistance. New infrastructure is evaluated in terms of accessibility, technical feasibility, and
technical capacity.
6 The community builds new infrastructure over 90-180 days, supported by project funds. Key messages
about sanitation and hygiene are discussed during sensibilization meetings or during visits to families by
the WASH committee, community health workers, or other volunteers. Health Zone staff are expected to
visit the community monthly during this time; Health Area staff weekly.
7 The community self-evaluates again, to measure progress since Step 2. The Health Zone conducts
additional KAP surveys and hosts three hours of meetings to assess the findings and make a plan to
maintain progress.
8 The CMO spends one day in the community to assess whether or not the community has completed its
action plan and achieved the seven norms. If they have, a certification ceremony is held. The CMO and the
village WASH committee develop a Community Action Plan for Maintenance so that the changes achieved
through the program can be sustained over time.
The IHfRA data collection team used electronic tablets and transmitted data to a cloud-based
server, allowing the research team to conduct quality control measures in real-time, checking for
consistency and errors. Additionally, IHfRA randomly selected 15% of villages for a second round
of interviews, by different interviewers, with a shorter questionnaire, to check consistency across
key variables. Separately, children from two households per village had their height and weight
re-measured by an IHfRA supervisor, as a quality check.
Outcomes
Primary outcomes were caregiver-reported diarrhea in the last seven days among all children
who were under 5 years old at the time of the survey, length-for-age Z score for a randomly
selected child in each household, and a WASH governance index. If the household had one child
between age 2 and 5, we measured the length and weight of that child. If the household had more
than one child between age 2 and 5, we randomly selected one child. If the household had no
children between ages 2 and 5, but one or more children between ages 0 and 2, we measured
7
length and weight of one of those children (randomly selected). Salter scale (Model 235 6S) and
wall-mounted measuring rods (portable baby/child length/height measuring system) were used.
The WASH governance index combined questions about the presence of a water committee,
frequency of committee meetings, WASH expenses (excluding maintenance), presence of a
maintenance plan, whether the committee tracks health conditions in the community, and whether
it tracks hygiene and sanitation.
Secondary outcomes were access to improved water and sanitation facilities, water quality at
water points and in homes, hygiene knowledge and behaviors, observed handwashing,
perceptions of WASH governance, children’s school absenteeism, child weight-for-age Z score,
and child weight-for-height Z score.
Structured observation of handwashing was done in four households per village. We first
attempted to observe the four households that were interviewed at the five-month follow-up; if any
were unavailable or unwilling, we randomly selected from the six new households to replace them.
A research assistant spent two hours in each of these households, recording if handwashing
occurred at critical junctures such as before preparing food (see S8 Table for full list of junctures).
This took place before the interview, to minimize Hawthorne effects.
Access to improved water and sanitation was self-reported, i.e. respondents reported whether
their main water source is improved or not according to the Joint Monitoring Program standard
definitions (e.g. boreholes are considered improved, while unprotected springs or surface water
sources are not). Cost paid by households for water and time spent collecting water were also
self-reported.
To measure water quality, we tested samples collected (i) at each of the water points used by
members of each village, and (ii) at household water storage containers in six randomly selected
households per village, on average. Testing was done concurrently with the household interviews.
We used the Aquagenx Compartment Bag Test E. Coli +Total Coliform Most Probable Number
(MPN) Kit. This measured the MPN of fecal indicator bacteria [28].
To measure subjective performance of local WASH institutions, we used survey questions about:
fairness of selection of water governance entity, perception of fair treatment, confidence in the
entity’s management of money, confidence in the entity’s response to infrastructure breakdowns,
confidence in management, and overall satisfaction. The data collection team also directly
observed water point functionality. Length of breakdowns was reported via a water committee
and village leader survey.
Statistical analyses
The number of village clusters in the study was determined by the program budget and proximity
of villages to one another (details above). Sample size calculations were used to determine how
many households in each village should be interviewed. Based on the primary outcome of
diarrhea prevalence, with a minimum detectable effect of 8 percentage points (pp), 32%
8
prevalence in the control group (based on the 5-month follow-up results), intracluster correlation
of 0.09 (based on 5-month follow-up), and 1.3 children under 5 years per household (based on 5-
month follow-up), we required 10 households per village.
We estimated intervention effects according to random assignment (intention to treat),
irrespective of adherence to the intervention.
For both primary and secondary outcomes, whether binary or continuous, we fitted linear models
with a binary variable indicating whether the participant was in a treatment or control cluster [29].
Because randomization was stratified by province and number-of-villages-per-cluster, we
included binary variables (fixed effects) for each stratum (n=12) in the model. We also included
gender and age (month) indicator variables for all child health outcomes. We clustered standard
errors at the cluster (i.e. group of villages) level.
For the primary outcome of WASH institutions, and secondary outcomes consisting of multiple
measures, we calculated a summary index to avoid over-rejection of the null hypothesis due to
multiple inferences. We rescaled each outcome so that higher values implied better outcomes,
and averaged standardized values relative to the control group. Treatment effects were estimated
as the difference in the summary index between treatment and control groups, such that treatment
effects are expressed in standard deviation units relative to the control group.
We pre-registered two analyses restricted by subgroup: by province (for all three primary
outcomes), and by gender (for diarrhea and length-for-age). We test for interaction on the additive
scale, using interaction terms in linear models.
Statistical analyses were conducted in Stata version 16.0.
Role of the funding source
This study was funded by the UK Foreign, Commonwealth, and Development Office (FCDO)
(https://www.gov.uk/government/organisations/foreign-commonwealth-development-office), via
Amendment No. 3 to the Supplemental Arrangement with the World Bank regarding Multi-Donor
Trust Fund for Impact Evaluation to Development Impact (TF072617, parallel to TF072161). AC,
KC, and EM were staff at Development Impact at that time. The Healthy Villages & Schools
program was a DRC government national program funded by UK’s FCDO and implemented with
UNICEF’s support. The funder and implementing partners provided inputs at the design stage to
ensure the study addressed policy and program priorities of importance to them. The funders had
no role in data collection and analysis, decision to publish, or preparation of the manuscript.
Results
Of the 1,312 respondents in 328 villages interviewed for the 5-month follow-up, 1,133 (86%) in
328 villages were re-interviewed at the 3.6-year follow-up, between November 24, 2022 and
February 10, 2023 (Figure 1). We also reached two villages that were not accessible during 5-
month follow-up and lost one village that was surveyed at 3.6-year follow-up due to conflicts. In
39 households with a 5-month follow-up interview, a new respondent was interviewed at 3.6
9
years, and 140 households (11%) were replaced between 5-month and 3.6-year follow-ups.
Additionally, in each village at the 3.6-year follow-up, six never-previously-interviewed households
were randomly selected, conditional on having lived in the village for at least four years, yielding
1,970 interviews (in four villages, only five households were reached). Thus, at 3.6 years, we
interviewed a total of 3,283 households. Of those households, 75% (2,466 out of 3,283) had at
least one child eligible for a caregiver’s reports of diarrhea, and 72% (2,374 out of 3,283) had at
least one child eligible for length and weight measurement. The primary outcome of WASH
institutions was measured in 329 villages. In the intervention group, the median time since the
completion of Healthy Villages Step 6 (construction of infrastructure) was 3.6 years (IQR = 3.4 to
3.7).
At 3.6 years, respondents in intervention and control groups were similar with regard to
characteristics unlikely to be affected by the intervention, such as marital status, educational
attainment, age, religion, household size, and home construction materials (Table 2).
Table 2. Household and respondent characteristics by intervention group, at 3.6-year follow-up
Control Intervention
Outcomes n Mean SD n Mean SD Adj. Diff. p-value
Household has improved roof 1843 0.42 0.49 1436 0.47 0.50 -0.01 0.81
Household has improved wall 1845 0.01 0.09 1438 0.01 0.12 0.01 0.18
Household has improved floor 1816 0.04 0.19 1430 0.08 0.27 0.03 0.11
Household size 1845 7.20 2.86 1438 7.18 2.93 0.03 0.80
Respondent identifies as Catholic 1845 0.18 0.38 1438 0.19 0.40 0.02 0.54
Respondent identifies as Protestant 1845 0.31 0.46 1438 0.32 0.47 -0.06 0.05
Respondent identifies with other religion 1845 0.02 0.15 1438 0.03 0.18 0.01 0.26
Respondent age 1845 40 13.39 1438 40 13.07 0.72 0.25
Respondent has completed primary school 1845 0.31 0.46 1438 0.34 0.48 -0.02 0.50
Respondent has completed secondary school 1845 0.06 0.23 1438 0.07 0.25 0.00 0.78
Respondent is married or cohabitating 1845 0.83 0.37 1438 0.82 0.38 -0.01 0.49
Adj. diff. = adjusted difference between intervention group and control group, estimated with models that include controls for randomization
blocks based on province and number of villages per cluster, and standard errors clustered by cluster. There were 121 clusters in total. Improved
roof=1 if roof is finished roofing (i.e., metal, wood, calamine/cement fiber ceramic tiles, cement or roofing shingles); improved walls=1 if walls are
‘finished walls’; improved floor=1 if floor is ‘finished floor’. All variables are binary except ‘HH size’ and ‘respondent age’; for these binary
variables, the mean represents the proportion of respondents who are in the listed category.
In the intervention group, 96% of villages reported that they created a community action plan and
prioritized actions to improve WASH (as instructed by the program); 86% reported that they had
implemented that plan.
The intervention had no effect on diarrhea (adjusted mean difference -0.01 [95% -0.05– 0.03])
(Table 3). Diarrhea prevalence was high overall, at 38% in the treatment group and 42% in the
control group. The intra-cluster coefficient (ICC) for diarrhea in the control group was 0.05; in the
intervention group, 0.07.
10
Table 3. Intervention effects on primary outcomes: diarrhea, length-for-age, and WASH institutions
Control Intervention
Outcomes n Prevalence/Mean SD ICC n Prevalence/Mean SD ICC ITT CI 95%
Diarrhea prevalence 2310 42% 0.05 1762 38% 0.07 -0.01 -0.05 0.03
Length-for-age Z-score 1223 -2.18 1.60 0.04 919 -2.20 1.59 0.06 -0.01 -0.15 0.12
WASH institutions index 185 0.00 1.00 0.47 144 0.46 0.75 0.11 0.40 0.16 0.65
ITT = intention-to-treat effect estimate; ICC =intracluster correlation; HH = household. Effects are estimated with models that
include controls for randomization blocks based on province and number of villages per cluster. There were 121 clusters in total.
The WASH institutions index was calculated by rescaling each variable in the index (eg, presence of WASH committee) so that
higher values imply better outcomes, then standardizing relative to the control group, following Kling et al. Effects are in standard
deviation units.
The intervention had no effect on length-for-age Z-scores in children (adjusted mean difference -
0.01 [95% CI -0.15–0.12]). In the control group, the mean length-for-age Z-score was -2.18 (1.60
SD) (Figure 2). The ICC for length-for-age Z-score in the control group was 0.04; in the
intervention group, 0.06.
Villages in the intervention group had a 0.40 higher score on the WASH institutions index (95%
CI 0.16–0.65). The percentage of villages in the intervention group with an active water,
sanitation, and hygiene (or just water) committee was 21 pp higher than the control group. The
ICC for the WASH institutions index in the control group was 0.47; in the intervention group, 0.11.
Households in the intervention group were 24 pp (95% CI 12–36) more likely to report using an
improved water source, 18 pp (95% CI 10–25) more likely to report using an improved sanitation
facility, and reported more positive perceptions of water governance (adjusted difference 0.19 SD
[95% CI 0.04–0.34]) (Table 3). The more positive perceptions of water governance were driven
by higher reported satisfaction with water access (0.56 points higher (95% CI 0.27–0.85) on a 1-
5 scale). Intervention group households were also 9 pp more likely to report paying for water (95%
CI 0-19). Conditional on paying for water, there was no difference in the amount paid between
the intervention and control groups. The intervention had no effect on time spent collecting water.
Intervention group water points were 11 pp less likely to be currently functional (95% CI -0.18– -
0.05); 97% of water points in control villages and 85% of water points in intervention villages were
functioning.
Intervention group respondents scored 0.23 higher (95% CI 0.12–34) on the index of self-reported
hygiene & behavior index. This was driven by several measures. The intervention group was 3
pp more likely to report treating their water (95% CI 1–5), and 9 pp more likely to report
handwashing with soap or ash at least once in the previous day (95% CI 6-13). The intervention
group also scored 0.43 points higher (95% CI 0.19–0.67) on the handwashing knowledge scale
(range 0-10). However, in structured observations of handwashing behavior, there was no
difference between intervention and control households in measures of any observed
handwashing or handwashing with soap or ash.
Water samples from intervention village water points showed a small but statistically significant
improvement in thermotolerant coliforms per 100mL compared to control village samples (-0.17
adjusted mean difference in log10(MPN); 95% CI -0.32– -0.02). Overall water quality was low,
11
even from improved sources. Among unimproved sources, 86% of samples had coliform levels
over 100 per 100mL, ‘very high risk’ according to WHO standards [30]; among improved sources,
60% of samples had coliform levels over 100 per 100mL. Water samples from intervention
household water containers showed no difference in thermotolerant coliforms per 100mL
compared to control household samples.
The intervention had no effect on the psychological well-being index, the life satisfaction & self-
esteem index, or school attendance.
In the prespecified subgroup analysis of primary outcomes by province, we find that the
intervention reduced diarrhea in one of four provinces (Kongo Central), reduced length-for-age in
one province (Kasai Central), and increased the WASH institutions index in two provinces (Kasai
and Kasai Central) (see S3 Table). We also used interaction terms in linear models to test for
effect modification on the additive scale (S5 Table). Of the nine coefficients (three provinces,
leaving out a reference, and three primary outcomes), two were statistically significant: in Kasai
Central province, the results suggest that the intervention increased diarrhea prevalence relative
to the treatment effect in the reference province (Kongo Central); in Kasai province, the results
suggest that the intervention increased length-for-age. For both diarrhea and length-for-age, the
Wald test rejects the null hypothesis that all province-by-intervention coefficients are zero at the
0.05 level. For the WASH institutions index, the Wald test fails to reject the null.
In the prespecified subgroup analysis of diarrhea and length-for-age by child sex, we find no
difference in intervention effects by sex (see S4 Table). In linear models with interaction terms for
intervention-by-sex, the coefficients are not statistically significant (S7 Table).
Discussion
We tested the effects of the national community-led rural WASH program in the DRC on child
length-for-age, diarrhea, and WASH institutions. The program improved community WASH
institutions, with intervention villages more likely to have a WASH committee, and for this
institution to actively monitor community health conditions. Intervention villages also had greater
access to improved water and sanitation infrastructure as a result of the program. However, we
cannot reject the null hypothesis that the intervention did not have any effect on child length-for-
age or diarrhea.
The finding of no effect on length-for-age is consistent with a recent meta-analysis of 11 WASH
trials with length-for-age as a primary outcome that found an adjusted mean difference in Z-score
between intervention and control of 0.00 (95% CI -0.03–0.04) [13]. These trial results stand in
contrast to many observational studies finding that WASH protects against growth faltering [31].
This suggests that the observational results may be confounded by other household or community
characteristics.
We measured fecal indicator bacteria in water sources and household water containers. We found
no difference between the intervention and control group household water quality, despite the fact
that intervention households were 24 pp more likely to use an improved water source. This is
likely due to the fact that improved water sources in our study had low-quality water, consistent
with evidence from DRC [25] and elsewhere [33]. It may also be linked to recontamination of water
12
between collection and its ultimate use in the household. Overall, intervention water points had
only slightly lower levels of fecal indicator bacteria than control water points; log10MPN/100mL
was 0.17 lower in intervention water points (95% CI -0.32– -0.02), with a mean of 1.71 in the
control group. This is consistent with a meta-analysis of five WASH trials that found only a 6%
reduction in prevalence of enteropathogens in environmental samples [32].
In addition, two-thirds of our respondents reported spending at least one day per week working
on an agricultural plot (modal response = 4 days). Of those who did, 95% report open defecation
while in the field, and 91% report drinking from surface water or unprotected springs. This
highlights the challenge of delivering safe and comprehensive WASH services in some
agricultural settings.
Our finding of no effect on diarrhea contrasts with a meta-regression (conducted as part of a
systematic review) which found use of an off-site improved water source reduces the relative risk
of diarrhea by 19% compared to unimproved water [14]. The same analysis found that basic
sanitation without sewer connection lowered the relative risk of diarrhea by 21% relative to
unimproved or limited sanitation, which also contrasts with our results, while hygiene interventions
reduced the relative risk of diarrhea by 30%. However, another review found that effective
handwashing promotion typically requires daily to fortnightly contact between the promoter and
participant [31]. It is possible that our intervention achieved that frequency of contact during the
most intensive 90-180 days of implementation, but also likely that effects would have faded out
by our measurements over three years later.
We found no effect of the intervention on handwashing with soap or ash during structured
observations of study participants by our research team. However, participants in the intervention
group were 9 pp more likely to report washing with soap or ash the previous day. This underscores
the limitations of self-reported data, particularly when the socially desirable outcome is likely to
be known and salient.
Sustainability is a widely-recognized challenge for WASH interventions. At five months post-
intervention, Healthy Villages and Schools increased access to improved water sources and
improved sanitation facilities[26]. Notably, these improvements largely persisted to 3.6 years post-
intervention, as did the improvements in WASH institutions. Yet these improvements did not result
in any measurable effect on diarrhea or growth faltering. This highlights that it is crucial to measure
health outcomes directly and not assume that better inputs are sufficient to yield improvements.
This study has several limitations. First, the trial had incomplete adherence: 86% of villages in the
intervention group reported that they had implemented the community action plan to address
WASH challenges (e.g. by building new infrastructure) by the 3.6-year follow-up. However, this is
a realistic level of adherence for a government-implemented program. Indeed, given that many
study villages were conflict-affected, the take-up rate was substantial. Second, we have no
baseline measures. Although the randomized design means that such measures are not required
for unbiased estimates of treatment effects, there are theoretical challenges; for example, if
permanent migration out of study villages was affected by the intervention, then our estimates
may be biased. However, at 3.6 years post-intervention we were able to re-interview 88% of the
13
households interviewed at 5 months post-intervention, suggesting that migration was relatively
rare in our study population. We also restricted households that were newly recruited at 3.6 years
to those who had lived in their current residence for at least four years. Third, the outcomes that
are self-reported may suffer from reactivity or social-desirability bias.
Our results reinforce calls for more ambitious attempts to improve WASH services to reduce
stunting and diarrhea, such as “transformative WASH.” [34] Proponents of transformative WASH
argue that some or all of the following may be necessary to produce significant health gains: high
community coverage of improved sanitation facilities; living environments free from animal feces;
continuous, convenient access to clean water; new approaches to behavior change; or new
technologies to deliver WASH services. [31] Others go further and argue for transformative
housing, with connections to water and sanitation networks. [12] These critiques are relevant in
our setting, given the multiple potential sources of contamination, the low quality of even
“improved sources”, and the high burden of disease. Business as usual is not enough.
Data sharing statement
Individual-level, de-identified data from this study and code to reproduce all results are publicly
available in the World Bank micro-data catalogue here:
https://reproducibility.worldbank.org/index.php/catalog/239.
Figure titles and legends
Fig 1. Trial Profile.
HH = household
Fig 2. Distribution of length-for-age Z-scores, intervention and control groups
Length-for-age Z-scores for children aged 0-5 years, in the intervention group (n=919) and the
control group (n=1223).
Supporting information
S1 Table. Randomization strata
S2 Table. Intervention effects on WASH institutions index and index sub-components
S3 Table. Intervention effects on all secondary outcomes, including index sub-components
S4 Table. Intervention effects on all primary outcomes, separately by province (pre-specified)
S5 Table. Intervention effects on all primary outcomes, province-by-intervention interaction
models
S6 Table. Intervention effects on diarrhea and length-for-age z score, separately by sex (pre-
specified)
S7 Table. Intervention effects on diarrhea and length-for-age z score, sex-by-intervention
interaction models
S8 Table. Variable definitions
S1 Text. Consort checklist
S2 Text. Protocol and pre-analysis plan, 5-month follow-up
S3 Text. Protocol and pre-analysis plan, 3.6 year follow-up
14
References
1. Wolf J, Johnston RB, Ambelu A, Arnold BF, Bain R, Brauer M, et al. Burden of disease
attributable to unsafe drinking water, sanitation, and hygiene in domestic settings: a global
analysis for selected adverse health outcomes. The Lancet. 2023;401: 2060–2071.
2. Grantham-McGregor S, Cheung YB, Cueto S, Glewwe P, Richter L, Strupp B, et al.
Developmental potential in the first 5 years for children in developing countries. The
Lancet. 2007;369: 60–70.
3. Black MM, Walker SP, Fernald LC, Andersen CT, DiGirolamo AM, Lu C, et al. Early
childhood development coming of age: science through the life course. The Lancet.
2017;389: 77–90.
4. Prüss-Ustün A, Wolf J, Bartram J, Clasen T, Cumming O, Freeman MC, et al. Burden of
disease from inadequate water, sanitation and hygiene for selected adverse health
outcomes: An updated analysis with a focus on low- and middle-income countries. Int J
Hyg Environ Health. 2019;222: 765–777. doi:10.1016/j.ijheh.2019.05.004
5. WHO. Progress on household drinking water, sanitation and hygiene 2000-2020: five years
into the SDGs. Progress on household drinking water, sanitation and hygiene 2000-2020:
five years into the SDGs. 2021.
6. WHO/UNICEF JMP. Progress on household drinking-water, sanitation and hygiene 2000-
2022: Special focus on gender. WHO/UNICEF JMP; 2023. Available:
https://www.who.int/publications/m/item/progress-on-household-drinking-water--sanitation-
and-hygiene-2000-2022---special-focus-on-gender
7. Bendavid E, Boerma T, Akseer N, Langer A, Malembaka EB, Okiro EA, et al. The effects
of armed conflict on the health of women and children. The Lancet. 2021;397: 522–532.
8. Armed Conflict Location and Event Data Conflict Index Mid-Year Update. 2023. Available:
https://acleddata.com/acled-conflict-index-mid-year-update-2023
9. UNICEF. Progress on drinking water, sanitation and hygiene in Africa 2000–2020: five
years into the SDGs. 2022.
10. Okun DA. The value of water supply and sanitation in development: an assessment. Am J
Public Health. 1988;78: 1463–1467. doi:10.2105/AJPH.78.11.1463
11. Pickering AJ, Djebbari H, Lopez C, Coulibaly M, Alzua ML. Effect of a community-led
sanitation intervention on child diarrhoea and child growth in rural Mali: a cluster-
randomised controlled trial. The Lancet Global Health. 2015;3: e701–e711.
doi:10.1016/S2214-109X(15)00144-8
12. Whittington D, Radin M, Jeuland M. Evidence-based policy analysis? The strange case of
the randomized controlled trials of community-led total sanitation. Oxford Review of
Economic Policy. 2020;36: 191–221.
13. Bekele T, Rawstorne P, Rahman B. Effect of water, sanitation and hygiene interventions
alone and combined with nutrition on child growth in low and middle income countries: a
systematic review and meta-analysis. BMJ Open. 2020;10: e034812.
doi:10.1136/bmjopen-2019-034812
14. Wolf J, Hubbard S, Brauer M, Ambelu A, Arnold BF, Bain R, et al. Effectiveness of
interventions to improve drinking water, sanitation, and handwashing with soap on risk of
diarrhoeal disease in children in low-income and middle-income settings: a systematic
review and meta-analysis. The Lancet. 2022;400: 48–59. doi:10.1016/S0140-
6736(22)00937-0
15. Null C, Stewart CP, Pickering AJ, Dentz HN, Arnold BF, Arnold CD, et al. Effects of water
quality, sanitation, handwashing, and nutritional interventions on diarrhoea and child
growth in rural Kenya: a cluster-randomised controlled trial. The Lancet Global Health.
15
2018;6: e316–e329. doi:10.1016/S2214-109X(18)30005-6
16. Luby SP, Rahman M, Arnold BF, Unicomb L, Ashraf S, Winch PJ, et al. Effects of water
quality, sanitation, handwashing, and nutritional interventions on diarrhoea and child
growth in rural Bangladesh: a cluster randomised controlled trial. The Lancet Global
Health. 2018;6: e302–e315. doi:10.1016/S2214-109X(17)30490-4
17. Humphrey JH, Mbuya MN, Ntozini R, Moulton LH, Stoltzfus RJ, Tavengwa NV, et al.
Independent and combined effects of improved water, sanitation, and hygiene, and
improved complementary feeding, on child stunting and anaemia in rural Zimbabwe: a
cluster-randomised trial. The Lancet Global Health. 2019;7: 132–147.
18. Patil SR, Arnold BF, Salvatore AL, Briceno B, Ganguly S, Jr JMC, et al. The Effect of
India’s Total Sanitation Campaign on Defecation Behaviors and Child Health in Rural
Madhya Pradesh: A Cluster Randomized Controlled Trial. PLOS Medicine. 2014;11:
e1001709. doi:10.1371/journal.pmed.1001709
19. Bowen A, Agboatwalla M, Luby S, Tobery T, Ayers T, Hoekstra RM. Association between
intensive handwashing promotion and child development in Karachi, Pakistan: a cluster
randomized controlled trial. Archives of pediatrics & adolescent medicine. 2012;166: 1037–
1044.
20. Briceño B, Coville A, Gertler P, Martinez S. Are there synergies from combining hygiene
and sanitation promotion campaigns: Evidence from a large-scale cluster-randomized trial
in rural Tanzania. PLOS ONE. 2017;12: e0186228. doi:10.1371/journal.pone.0186228
21. Cameron LA, Shah M, Olivia S. Impact evaluation of a large-scale rural sanitation project
in Indonesia. World Bank policy research working paper. 2013 [cited 30 May 2024].
Available: https://www.jstor.org/stable/pdf/resrep26254.pdf
22. Clasen T, Boisson S, Routray P, Torondel B, Bell M, Cumming O, et al. Effectiveness of a
rural sanitation programme on diarrhoea, soil-transmitted helminth infection, and child
malnutrition in Odisha, India: a cluster-randomised trial. The Lancet Global Health. 2014;2:
e645–e653. doi:10.1016/S2214-109X(14)70307-9
23. Du Preez M, Conroy RM, Ligondo S, Hennessy J, Elmore-Meegan M, Soita A, et al.
Randomized Intervention Study of Solar Disinfection of Drinking Water in the Prevention of
Dysentery in Kenyan Children Aged under 5 Years. Environ Sci Technol. 2011;45: 9315–
9323. doi:10.1021/es2018835
24. Galiani SG, Orsola-Vidal P, Alexandra. Promoting Handwashing Behavior in Peru: The
Effect of Large-Scale Mass-Media and Community Level Interventions. Policy Research
Working Papers The World Bank. 2012. doi:10.1596/1813-9450-6257.
25. World Bank. WASH Poor in a Water-Rich Country: A Diagnostic of Water, Sanitation,
Hygiene, and Poverty in the Democratic Republic of Congo. Washington, DC: World Bank;
2017.
26. Quattrochi JP, Coville A, Mvukiyehe E, Dohou CJ, Esu F, Cohen B, et al. Effects of a
community-driven water, sanitation and hygiene intervention on water and sanitation
infrastructure, access, behaviour, and governance: a cluster-randomised controlled trial in
rural Democratic Republic of Congo. BMJ global health. 2021;6: e005030.
27. Programme National Ecole et Village Assaini (PNEVA). Atlas 2018 : Acces a L’eu potable,
a l’hygeine et a l’assainaissement pour les commautes rurales et periurbaines de la
republique democratique du Congo. 2018. Available:
https://www.unicef.org/drcongo/media/2806/file/COD-Atlas2018.pdf
28. Gronewold AD, Sobsey MD, McMahan L. The compartment bag test (CBT) for
enumerating fecal indicator bacteria: basis for design and interpretation of results. Science
of the Total Environment. 2017;587: 102–107.
29. Gomila R. Logistic or linear? Estimating causal effects of experimental treatments on
binary outcomes using regression analysis. J Exp Psychol Gen. 2021;150: 700–709.
doi:10.1037/xge0000920
16
30. WHO. Guidelines for Drinking-water Quality. World Health Organization; 2004.
31. Pickering AJ, Null C, Winch PJ, Mangwadu G, Arnold BF, Prendergast AJ, et al. The
WASH Benefits and SHINE trials: interpretation of WASH intervention effects on linear
growth and diarrhoea. Lancet Glob Health. 2019;7: 1139–46.
32. Mertens A, Arnold BF, Benjamin-Chung J, Boehm AB, Brown J, Capone D, et al. Effects of
water, sanitation, and hygiene interventions on detection of enteropathogens and host-
specific faecal markers in the environment: a systematic review and individual participant
data meta-analysis. Lancet Planet Health. 2023;7: e197–e208. doi:10.1016/S2542-
5196(23)00028-1
33. Bain R, Johnston R, Khan S, Hancioglu A, Slaymaker T. Monitoring Drinking Water Quality
in Nationally Representative Household Surveys in Low- and Middle-Income Countries:
Cross-Sectional Analysis of 27 Multiple Indicator Cluster Surveys 2014–2020. Environ
Health Perspect. 2021;129: 097010. doi:10.1289/EHP8459
34. Cumming O, Arnold BF, Ban R, Clasen T, Esteves Mills J, Freeman MC, et al. The
implications of three major new trials for the effect of water, sanitation and hygiene on
childhood diarrhea and stunting: a consensus statement. BMC Medicine. 2019;17: 173.
doi:10.1186/s12916-019-1410-x
1
17
Figures
Figure 1. Trial profile
403 villages identified by
intervention implementer
34 villages in Kasai province
excluded because they received
the intervention prior to
369 villages eligible, grouped
into 124 clusters
30 villages (3 clusters) in
North Kivu dropped due to
339 villages in 121 clusters randomly
assigned to intervention or control
50 clusters randomly 71 clusters randomly
assigned to the intervention assigned to the control
group group
( )
2 villages randomly dropped
from largest cluster in in Kasai, 3 villages randomly dropped
due to intervention budget from largest cluster in in Kasai,
constraint due to intervention budget
constraint
1 village randomly dropped
from largest cluster in in Kongo 1 village randomly dropped
Central, due to intervention from largest cluster in in Kongo
budget constraint Central, due to intervention
budget constraint
1 village in Kasai was not able
to be reached 1 village in South Kivu did not
exist
1 village in Kasai did not exist
50 clusters 71 clusters
144 villages 185 villages
Four midline panel HHs per village = Four midline panel HHs per village =
489 644
Replacement HHs for midline HHs = Replacement HHs to midline HHs = 92
87
Six new randomly selected HHs per
Six new randomly selected HHs per village = 1,109 HHs*
village = 862 HHs* *In one village, only five new HHs
*In three villages, only five new HHs were interviewed
were interviewed
Total = 1,845 HHs interviewed 3.5
18
Figure 2. Distribution of length-for-age Z-scores, intervention and control groups
19
Supporting Information
S1 Table. Randomization strata
Total Control Total
Villages- villages Intervention villages clusters Intervention Control
per- in villages in in in clusters in clusters in
Province Stratum cluster stratum stratum stratum stratum stratum stratum
Kongo
Central 1 1-2 15 8 7 12 6 6
2 3-4 13 6 7 4 2 2
3 5,7 12 7 5 2 1 1
Kasai 1 1-2 37 19 18 26 13 13
2 3-5 39 21 18 11 6 5
3 10,12 22 12 10 2 1 1
Kasai
Central 1 1-2 34 8 26 25 6 19
2 3-4 34 8 26 10 2 8
3 6-7 13 0 13 2 0 2
South Kivu 1 1-2 20 9 11 13 6 7
2 4-8,10 74 39 35 12 6 6
3 12,14 26 12 14 2 1 1
Total 339 149 190 121 50 71
20
S2 Table. Intervention effects on WASH institutions index and index sub-components
Control Intervention CI 95%
Lower Upper
Outcomes n Mean SD n Mean SD ITT Bound Bound
WASH institutions index 185 0.00 1.00 144 0.46 0.75 0.40 0.16 0.65
Committee (y/n) 185 0.70 0.46 144 0.97 0.16 0.21 0.10 0.32
Committee mtg freq* 88 2.91 1.49 104 2.62 1.61 -0.29 -0.79 0.20
WASH expenditures (CDF
per month, IHS) 130 2.12 4.07 140 3.28 4.58 1.14 0.07 2.20
Track health (y/n) 185 0.69 0.46 144 0.84 0.37 0.16 0.07 0.25
Track sanitation (y/n) 185 0.74 0.44 144 0.78 0.41 0.05 -0.06 0.15
ITT = intention-to-treat effect estimate. CDF = Congolese francs. IHS = inverse hyperbolic spline. Effects are estimated with
models that include controls for randomization blocks based on province and number of villages per cluster. There were 121
clusters in total. The WASH institutions index was calculated by rescaling each variable in the index (e.g., presence of WASH
committee) so that higher values imply better outcomes, then standardizing relative to the control group, following Kling et al.
Effects are in standard deviation units. The index values range from -2.35 to 2.12.
*Committee meeting frequency is coded 1-6, where 1=weekly, 2=Fortnightly, 3=Monthly, 4=Every 3 months, 5=Every 6 or more
months, 6=No regular schedule, based on needs
21
S3 Table. Intervention effects on all secondary outcomes, including index sub-
components
Control Intervention CI 95%
Prevalence/ Prevalence/
Outcomes n Mean SD N Mean SD ITT Lower Upper
Improved water source 1845 43% 1437 74% 0.24 0.12 0.36
Improved sanitation 1843 14% 1437 34% 0.18 0.10 0.25
Water governance perception index 1845 0.00 1.00 1438 0.22 0.88 0.19 0.04 0.34
Committee selected fairly 600 1.76 0.81 1119 1.82 0.76 0.06 -0.04 0.17
Committee treats community fairly 621 4.15 1.31 1158 4.07 1.31 -0.09 -0.27 0.09
Committee manages money well 635 3.25 1.12 721 3.31 1.14 0.05 -0.13 0.23
Comm responds breakdown well 972 3.69 1.18 1227 3.67 1.23 0.01 -0.13 0.15
Confidence committee will solve reported issue 1009 3.10 1.05 1227 3.12 1.03 0.02 -0.13 0.17
Satisfaction with water access (1-5) 1845 2.76 1.49 1438 3.38 1.43 0.56 0.27 0.85
Time spent collecting water (IHS) 1845 4.17 2.07 1438 4.04 2.05 0.03 -0.20 0.27
HHs pay for water (Dummy) 1845 10% 1438 23% 0.09 0.00 0.19
Water use expenditure (weekly) 137 153 217 303 254 351 114 -16 243
Hygiene and behaviour index 1845 0.00 1.00 1438 0.30 1.07 0.23 0.12 0.34
Handwashing score (0-10) 1845 2.79 2.22 1438 3.41 2.16 0.43 0.19 0.67
Open defecation (%) 1843 32% 1437 27% -0.03 -0.08 0.01
Self-reported handwashing with soap/ash (%) 1843 67% 1437 82% 0.09 0.06 0.13
Frequency of latrine cleaning over past 2 weeks 761 4.89 4.26 648 5.00 4.06 0.25 -0.32 0.82
Water in pot is clean and covered (%) 1669 57% 1273 57% 0.02 -0.03 0.07
Water treated for consumption, any method (%) 1669 3% 1273 7% 0.03 0.01 0.05
Handwashing action 616 15% 489 17% 0.00 -0.04 0.03
Handwashing with soap/ash 616 2% 489 4% 0.01 -0.01 0.03
Water point has water 380 97% 266 85% -0.11 -0.18 -0.05
WP FIB MPN/100mL 366 79.31 37.92 221 72.41 42.56 -9.46 -18.85 -0.06
WP FIB log10 MPN/100mL 366 1.71 0.60 221 1.58 0.70 -0.17 -0.32 -0.02
HH FIB MPN/100mL 1098 82.27 35.05 851 79.44 37.71 -3.87 -9.59 1.86
HH FIB log10 MPN/100mL 1098 1.76 0.53 851 1.72 0.58 -0.06 -0.15 0.04
Weight for age 1223 -1.17 1.16 919 -1.14 1.16 0.06 -0.04 0.15
Weight for length 1224 0.14 1.07 918 0.22 1.03 0.10 -0.02 0.22
Life satisfaction & self-esteem index 1843 0.00 1.00 1435 0.03 0.98 0.01 -0.12 0.14
Life satisfaction (WVS) 1843 4.87 3.20 1435 5.04 2.89 0.20 -0.10 0.49
Feel I am person of worth (Rosenberg) 1843 2.25 0.90 1435 2.19 0.90 -0.03 -0.13 0.07
Feel that I have good qualities (Rosenberg) 1843 2.42 0.74 1435 2.36 0.75 -0.02 -0.10 0.05
Inclined to feel I am a failure (Rosenberg) 1843 1.43 1.11 1435 1.49 1.05 -0.04 -0.16 0.08
Able to do things as well as other people (Rosenberg) 1843 2.44 0.75 1435 2.40 0.76 0.01 -0.06 0.08
Feel have not much to be proud of (Rosenberg) 1843 0.98 0.98 1435 0.99 0.92 -0.05 -0.14 0.04
Take a positive attitude towards self (Rosenberg) 1843 2.35 0.78 1435 2.37 0.74 0.08 -0.01 0.17
I am satisfied with myself (Rosenberg) 1843 2.03 1.02 1435 2.12 0.96 0.16 0.07 0.26
22
Wish could have more respect for myself (Rosenberg) 1843 0.42 0.60 1435 0.40 0.57 -0.07 -0.13 0.00
Certainly feel useless at times (Rosenberg) 1843 1.33 1.10 1435 1.40 1.07 -0.03 -0.16 0.10
At times think I am no good at all (Rosenberg) 1843 1.32 1.09 1435 1.43 1.07 0.00 -0.12 0.12
Psychological well-being index 1843 0.00 1.00 1435 0.05 0.99 0.06 -0.06 0.17
Felt cheerful last 2 weeks (WHO) 1843 2.59 1.72 1435 2.64 1.64 0.05 -0.12 0.22
Felt calm & relaxed last 2 weeks (WHO) 1843 2.62 1.68 1435 2.68 1.59 0.06 -0.11 0.22
Felt active & vigorous last 2 weeks (WHO) 1843 2.57 1.64 1435 2.61 1.54 0.04 -0.13 0.21
Woke up fresh & rested last 2 weeks (WHO) 1843 2.58 1.62 1435 2.56 1.53 0.02 -0.14 0.18
Daily life filled with things that interest last 2 weeks (WHO) 1843 2.25 1.76 1435 2.30 1.70 0.12 -0.06 0.29
Felt unable to control important things last month (Cohen) 1843 2.95 1.26 1435 2.92 1.12 -0.06 -0.17 0.05
Felt confident about ability to handle personal problems last month (Cohen) 1843 3.09 1.27 1435 3.05 1.17 0.02 -0.08 0.13
Felt confident things were going your way last month (Cohen) 1843 2.81 1.33 1435 2.89 1.25 0.10 -0.02 0.23
Felt difficulties were piling up could not overcome them last month (Cohen) 1843 2.39 1.28 1435 2.56 1.20 0.10 -0.03 0.23
School attendance (days past week) 299 3.21 2.72 226 2.74 2.72 -0.38 -0.82 0.07
ITT = intention-to-treat effect estimate; HH = household; FIB = fecal indicator bacteria; MPN = most probable number; IHS =
inverse hyperbolic spline; WHO = World Health Organization; WVS = World Values Survey. Effects are estimated with models that
include controls for randomization blocks based on province and number of villages per cluster. There were 121 clusters in total.
The WASH governance perceptions index and the hygiene and behavior index were calculated by rescaling each variable in the
index (e.g., satisfaction with water access) so that higher values imply better outcomes, then standardizing relative to the control
group, following Kling et al. Effects are in standard deviation units.
23
S4 Table. Intervention effects on all primary outcomes, by province (pre-specified)
Control Intervention CI 95%
Outcomes Province n Prevalence/Mean SD n Prevalence/Mean SD ITT Lower Bound Upper Bound
Diarrhea prevalence Kongo Central 146 17% 163 10% -0.07 -0.15 0.00
Kasai 599 40% 679 40% 0.00 -0.09 0.09
Kasai Central 781 49% 196 59% 0.10 0.00 0.19
South Kivu 784 41% 724 37% -0.05 -0.10 0.01
Length-for-age Z-score Kongo Central 96 -1.59 1.32 93 -1.75 1.67 -0.18 -0.47 0.11
Kasai 319 -2.05 1.60 328 -1.86 1.60 0.20 -0.05 0.44
Kasai Central 409 -2.12 1.71 103 -2.58 1.49 -0.42 -0.80 -0.04
South Kivu 399 -2.49 1.48 395 -2.50 1.50 0.02 -0.16 0.19
WASH institutions index Kongo Central 18 0.27 0.66 20 0.59 0.57 0.31 -0.06 0.68
Kasai 43 -0.48 1.10 48 0.21 0.75 0.70 0.21 1.19
Kasai Central 65 -0.18 0.85 16 0.28 0.61 0.35 0.03 0.68
South Kivu 59 0.46 0.96 60 0.67 0.78 0.23 -0.25 0.71
ITT = intention-to-treat effect estimate. Effects are estimated with models that include controls for randomization blocks based on
province and number of villages per cluster. There were 121 clusters in total. The WASH institutions index was calculated by rescaling
each variable in the index (e.g., presence of WASH committee) so that higher values imply better outcomes, then standardizing
relative to the control group, following Kling et al. Effects are in standard deviation units.
24
S5 Table. Intervention effects on all primary outcomes, province-by-intervention
interaction models
(1) (2) (3) (4) (5) (6)
WASH WASH Diarrhea - Diarrhea - Height for Height for
Intitutions Intitutions Simple Interaction Age - Age -
Index - Index - Simple Interaction
Simple Interaction
Intervention 0.404*** 0.308* -0.010 -0.070* -0.014 -0.172
(0.125) (0.167) (0.022) (0.036) (0.070) (0.134)
Kasai Province # cluster_strata=2 -0.644** -0.024 0.267*** 0.003 -0.085 0.090
(0.251) (0.256) (0.050) (0.044) (0.200) (0.149)
Kasai Province # cluster_strata=3 -0.432* 0.196 0.232*** -0.031 0.222 0.396***
(0.221) (0.315) (0.078) (0.077) (0.172) (0.091)
Kasai Central Province #
cluster_strata=2 -0.263 0.185 0.372*** -0.024 -0.347 -0.005
(0.201) (0.199) (0.059) (0.055) (0.261) (0.222)
Kasai Central Province #
cluster_strata=3 -0.866** -0.430 0.335*** -0.036 -0.143 0.106
(0.410) (0.411) (0.063) (0.059) (0.168) (0.140)
Sud-Kivu Province # cluster_strata=2 -0.064 0.042 0.273*** 0.035 -0.729*** 0.096
(0.231) (0.237) (0.041) (0.032) (0.161) (0.141)
Sud-Kivu Province # cluster_strata=3 0.670** 0.763*** 0.268*** 0.030 -0.240 0.586***
(0.262) (0.223) (0.042) (0.028) (0.164) (0.142)
Kasai Province -0.835*** 0.226*** -0.359*
(0.314) (0.054) (0.212)
Kasai Central Province -0.488* 0.339*** -0.323
(0.272) (0.050) (0.223)
Sud-Kivu Province -0.065 0.224*** -0.912***
(0.293) (0.050) (0.227)
Intervention # Kasai Province 0.391 0.069 0.365**
(0.293) (0.058) (0.182)
Intervention # Kasai Central Province 0.044 0.168*** -0.242
(0.230) (0.059) (0.226)
25
Intervention # Sud-Kivu Province -0.076 0.022 0.184
(0.286) (0.044) (0.158)
Child's sex 0.010 0.010 0.134** 0.135**
(0.014) (0.014) (0.062) (0.062)
Child's age (years) -0.043*** -0.043*** -0.233*** -0.234***
(0.004) (0.004) (0.033) (0.033)
Constant 0.272 0.324 0.235*** 0.268*** -1.323*** -1.248***
(0.168) (0.214) (0.039) (0.044) (0.182) (0.204)
Observations 329 329 4072 4072 2142 2142
Wald_F 0.790 3.169 3.022
Wald_p 0.502 0.027 0.032
Coefficients and standard errors from linear models of the primary outcomes on intervention group, fixed effects for randomization
strata, fixed effects for province, and intervention-by-province interaction terms. Standard errors are clustered by cluster (group of
villages). There were 121 clusters in total. The WASH institutions index was calculated by rescaling each variable in the index (e.g.,
presence of WASH committee) so that higher values imply better outcomes, then standardizing relative to the control group, following
Kling et al. Effects are in standard deviation units. The Wald F is a test statistic for the null hypothesis that all of the interaction term
coefficients are zero. * p<.1, ** p<.05, *** p<.01"
26
S6 Table. Intervention effects on diarrhea and length-for-age z score, separately by sex
(pre-specified)
Control Intervention CI 95%
Prevalence Prevalence
/ / Lower Upper
Outcomes Province n Mean SD n Mean SD ITT Bound Bound
Diarrhea Female 1151 43% 880 38% -0.01 -0.06 0.04
Male 1159 41% 882 38% 0.00 -0.06 0.05
Length for Female
age 605 -2.08 1.59 475 -2.19 1.64 -0.07 -0.24 0.10
Male 618 -2.28 1.61 444 -2.22 1.53 0.06 -0.14 0.26
ITT = intention-to-treat effect estimate. Effects are estimated with models that include controls for randomization blocks based on
province and number of villages per cluster. There were 121 clusters in total. The WASH institutions index was calculated by
rescaling each variable in the index (e.g., presence of WASH committee) so that higher values imply better outcomes, then
standardizing relative to the control group, following Kling et al. Effects are in standard deviation units.
27
S7 Table. Intervention effects on diarrhea and length-for-age z score, sex-by-intervention
interaction models
(1) (2) (3) (4)
Diarrhea - Diarrhea - Height for Height for
Simple Interaction Age - Age -
Simple Interaction
Treatment -0.010 -0.004 -0.014 0.071
(0.022) (0.027) (0.070) (0.098)
Child's sex 0.010 0.014 0.134** 0.206**
(0.014) (0.018) (0.062) (0.084)
Child's age (years) -0.043*** -0.043*** -0.233*** -0.233***
(0.004) (0.004) (0.033) (0.033)
Kasai Province # cluster_strata=2 0.267*** 0.267*** -0.085 -0.089
(0.050) (0.050) (0.200) (0.199)
Kasai Province # cluster_strata=3 0.232*** 0.232*** 0.222 0.213
(0.078) (0.078) (0.172) (0.174)
Kasai Central Province # cluster_strata=2 0.372*** 0.372*** -0.347 -0.351
(0.059) (0.058) (0.261) (0.260)
Kasai Central Province # cluster_strata=3 0.335*** 0.335*** -0.143 -0.150
(0.063) (0.063) (0.168) (0.167)
Sud-Kivu Province # cluster_strata=2 0.273*** 0.273*** -0.729*** -0.737***
(0.041) (0.041) (0.161) (0.159)
Sud-Kivu Province # cluster_strata=3 0.268*** 0.268*** -0.240 -0.242
(0.042) (0.042) (0.164) (0.163)
Treatment # Female -0.011 -0.168
(0.028) (0.122)
Constant 0.235*** 0.233*** -1.323*** -1.354***
(0.039) (0.040) (0.182) (0.185)
Observations 4072 4072 2142 2142
Wald_F 0.159 1.883
Wald_p 0.691 0.173
Coefficients and standard errors from linear models of the primary outcomes on intervention group, fixed effects for randomization
strata, fixed effects for child sex, and intervention-by-sex interaction terms. Standard errors are clustered by cluster (group of
villages). There were 121 clusters in total. The WASH institutions index was calculated by rescaling each variable in the index (e.g.,
presence of WASH committee) so that higher values imply better outcomes, then standardizing relative to the control group, following
Kling et al. Effects are in standard deviation units. The Wald F is a test statistic for the null hypothesis that all of the interaction term
coefficients are zero. * p<.1, ** p<.05, *** p<.01"
28
S8 Table. Variable definitions
Descriptions of primary and secondary outcomes, including index
subcomponents
Category Outcomes Definition
Primary Diarrhea Prevalence of diarrhea in last 7 days of
Outcome children under 5 (caregiver reported):
1. Yes to ‘diarrhea’ , or
2. Yes to ‘three or more stools” AND
‘watery or soft stool’, or
3. Yes to blood in the stool
Primary Length for age Length for age z score for children under
Outcome 5.
The sample does not include
observations with implausible z-scores.
Primary WASH institutions index The index is built from: (1) Presence of
Outcome water committee; (2) Frequency of
meeting; (3) Average amount spent per
month for water activities excluding
maintenance (inverse hyperbolic sine); (4)
Tracks health conditions; (5) Tracks
hygiene and sanitation
Subcomponent Presence of water committee 1 if the village has a water committee
0 Otherwise
Subcomponent Frequency of meeting Frequency of committee meeting. The
variable ranges from 1 to 6:
1 Weekly
2 Fortnightly
3 Monthly
4 Every 3 months
5 Every 6 or more months
6 No regular schedule, based on
needs
Subcomponent Average amount spent per month for Average amount spent per month for
water activities excluding maintenance. water activities excluding maintenance.
An inverse hyperbolic sine transformation
has been applied to the variable.
Subcomponent Tracks health conditions 1 if village keeps track of community
health conditions
0 Otherwise
Subcomponent Tracks hygiene and sanitation 1 if village keeps track of WASH practices
0 Otherwise
29
Secondary Water governance perception index Water governance perception index is
Outcome constructed based on the following
variables: fairness of selection of water
governance entity; perception of fair
treatment; confidence in managing money
of committee; confidence in committee
response to breakdowns; confidence in
committee management; Satisfaction with
water access
Subcomponent Committee selected fairly The extent to which the process of
choosing committee members was fair
The answer choices are:
Not fair at all;
Somewhat fair;
Fair;
Fully fair
Subcomponent Committee treats community fairly The extent to which the water committee
treats you fairly
The answer choices are:
Very unfair
Somewhat unfair
Neither fair nor unfair
Somewhat fair
Very fair
Subcomponent Committee manages money well How well the water committee manages
money
The answer choices are the following:
Very badly
Badly
Neither bad nor good
Good
Very good
Subcomponent Comm responds breakdown well How well the water committee responds
to breakdowns
The answer choices are the following:
Very badly
Badly
Neither bad nor good
Good
Very good
Subcomponent Confidence committee will solve Level of confidence that water committee
reported issue will solve issue respondent brings up
The answer choices are the following:
Not confident at all
Not very confident
30
Somewhat confident
Very confident
Subcomponent Satisfaction with water access (1-5) Level of satisfaction of respondent with
your access to water
The answer choices are the following:
Not satisfied at all
Not satisfied
Indifferent
Satisfied
Very satisfied
Secondary Improved water source 1 if Household’s Primary source of
Outcome drinking water is improved source (JMP
definition) and 0 otherwise.
● Improved main drinking source
includes Piped into dwelling,
Piped into plot, Piped/public tap,
Tube well or borehole, Protected
dug well, Protected spring,
Rainwater, Tanker truck.
● Unimproved main drinking source
includes Unprotected dug well,
Unprotected spring, Surface
water, No other main source
Secondary Time spent collecting water (IHS) Each household was asked to list
Outcome household members who participated on
previous day, and the number of trips
undertaken by the household. For each
trip, the collector was identified and the
time spent for the round trip was
collected. For each household, we sum
time spent for each trip in minutes to get
the total time spent collecting water per
household. To estimate treatment effects,
an inverse hyperbolic sine transformation
was applied due to the large number of
zeroes.
Secondary HHs pay for water (Dummy) 1 if household pay for water
Outcome 0 Otherwise
Secondary Water use expenditure (weekly) Estimated household’s weekly
Outcome expenditure for water use (in CDF,
Congolese Franc)
31
Secondary Improved sanitation Household uses an improved latrine (JMP
Outcome definition)
● Improved sanitation includes
options such as Flush / Pour
Flush: to Piped Sewer System,
Flush / Pour Flush: to Septic
Tank, Flush / Pour Flush: to Pit
Latrine, Ventilated Improved Pit
Latrine (VIP), Pit Latrine with
Slab, Composting Latrine.
● Unimproved sanitation includes
Flush / Pour Flush: to Elsewhere,
Flush / Pour Flush: to Don’t Know
Where, Pit Latrine without Slab /
Open Pit, Bucket Latrine,
Hanging Toilet, No Facilities
(Bush, Open Field, River), Other
(specify)
Secondary Hygiene and behaviour index Self-reports of: Knowledge: Caregiver
Outcome knows how and when to wash hands;
what causes diarrhea. Sanitation
practices: cleanliness of household area
and latrine (presence of flies and fecal
matter); open defecation; observed
indicators of toilet use –worn pathway,
presence of water; improvements to
latrine; disposes of child feces safely
(JMP definition). Water storage practices:
has a clean pot for water that is covered.
Subcomponent Handwashing score (0-10) Self-report of: Knowledge: Caregiver
knows how and when to wash hands. The
variables considered are the following:
count 1 if respondent mentioned
unprompted that they wash hands with
soap/ash in critical juncture X (= 0 if they
say no or they say it prompted). The
counts are then added up to create a
score.
The considered junctures are the
following: After toilet; after washing baby’s
bottom/changing; after eating; before
preparing food; before eating; before
feeding/breastfeeding baby; before or
after handling children; after taking care
of pets or farm animals; after
coughing/sneezing; after coming back
from the fields.
32
Subcomponent Open defecation (%) 1 if no defecation facility used and 0
otherwise
Subcomponent Self-reported handwashing with 1 if respondent washed their hands with
soap/ash (%) soap/ash at least once since previous day
0 Otherwise
Subcomponent Frequency of latrine cleaning over past Number of times the latrine has been
2 weeks cleaned in the past 2 weeks
Subcomponent Water pot is clean and covered (%) 1 if the pot has clean water and is
covered
0 Otherwise.
This indicator is measured via actual
observations
Subcomponent Water treated for consumption, any 1 if drinking water stored in household is
method (%) treated with any product/method for safe
consumption
0 otherwise
Secondary Life satisfaction & self-esteem index Summary index of 11 questions
Outcome The 11 questions include the (1) life
satisfaction question as defined by the
World Values Survey and (2) the 10
questions as defined in Rosenberg's Self-
Esteem Scale.
Subcomponent Life satisfaction (WVS) All things considered, on a scale of 1 to
10, how satisfied are you with your life as
a whole?
1 means completely dissatisfied
10 means completely satisfied
Subcomponent Feel I am person of worth (Rosenberg) I feel I am a person of worth, at least on
an equal plane with others.
Tell me to what extent you: Strongly
agree, Agree, Disagree, or Strongly
disagree with this statement about you.
Subcomponent Feel that I have good qualities I feel that I have a number of good
(Rosenberg) qualities.
Tell me to what extent you: Strongly
agree, Agree, Disagree, or Strongly
disagree with this statement about you.
Subcomponent Inclined to feel I am a failure All in all, I am inclined to feel that I am a
(Rosenberg) failure.
Tell me to what extent you: Strongly
agree, Agree, Disagree, or Strongly
disagree with this statement about you.
33
Subcomponent Able to do things as well as oth people I am able to do things as well as most
(Rosenberg) other people.
Tell me to what extent you: Strongly
agree, Agree, Disagree, or Strongly
disagree with this statement about you.
Subcomponent Feel have not much to be proud of I feel I do not have much to be proud of.
(Rosenberg)
Tell me to what extent you: Strongly
agree, Agree, Disagree, or Strongly
disagree with this statement about you.
Subcomponent Take a positive attitude towards self I take a positive attitude toward myself.
(Rosenberg)
Tell me to what extent you: Strongly
agree, Agree, Disagree, or Strongly
disagree with this statement about you.
Subcomponent I am satisfied with myself (Rosenberg) On the whole, I am satisfied with myself.
Tell me to what extent you: Strongly
agree, Agree, Disagree, or Strongly
disagree with this statement about you.
Subcomponent Wish could have more respect for I wish I could have more respect for
myself (Rosenberg) myself.
Tell me to what extent you: Strongly
agree, Agree, Disagree, or Strongly
disagree with this statement about you.
Subcomponent Certainly feel useless at times I certainly feel useless at times.
(Rosenberg)
Tell me to what extent you: Strongly
agree, Agree, Disagree, or Strongly
disagree with this statement about you.
Subcomponent At times think I am no good at all At times I think I am no good at all.
(Rosenberg)
To what extent do respondents: Strongly
agree, Agree, Disagree, or Strongly
disagree with this statement about them
Secondary Psychological well-being index Summary index of 9 questions on well-
Outcome being in last 2 weeks and stress in last 4
weeks.
The questions used here belong to two
different sets of questions. The first set is
the 5 WHO questions on well-being, and
the second set is the Cohen stress scale
4 questions.
34
Each of the five WHO statements on well-
being refers to how respondents have
been feeling over the last two weeks. The
answer choices are the following: At no
time; Some of the time; Less than half of
the time; More than half of the time; Most
of the time; All of the time.
Questions in the Cohen stress scale, ask
the respondent about their feelings and
thoughts during THE LAST MONTH. The
answer choices are the following: Never;
Almost never; Sometimes; Fairly often;
Very often.
Subcomponent Felt cheerful last 2 weeks (WHO) Over the last two weeks, I have felt
cheerful and in good spirits.
The answer choices are the following: At
no time; Some of the time; Less than half
of the time; More than half of the time;
Most of the time; All of the time.
Subcomponent Felt calm & relaxed last 2 weeks Over the last two weeks, I have felt calm
(WHO) and relaxed.
The answer choices are the following: At
no time; Some of the time; Less than half
of the time; More than half of the time;
Most of the time; All of the time.
Subcomponent Felt active & vigorous last 2 weeks Over the last two weeks, I have felt active
(WHO) and vigorous.
The answer choices are the following: At
no time; Some of the time; Less than half
of the time; More than half of the time;
Most of the time; All of the time.
Subcomponent Woke up fresh & rested last 2 weeks Over the last two weeks, I woke up
(WHO) feeling fresh and rested.
The answer choices are the following: At
no time; Some of the time; Less than half
of the time; More than half of the time;
Most of the time; All of the time.
Subcomponent Daily life filled with things that interest Over the last two weeks, my daily life has
last 2 weeks (WHO) been filled with things that interest me.
The answer choices are the following: At
no time; Some of the time; Less than half
35
of the time; More than half of the time;
Most of the time; All of the time.
Subcomponent Felt unable to control important things In the last month, how often have you felt
last month (Cohen) that you were unable to control the
important things in your life?
The answer choices are the following:
Never; Almost never; Sometimes; Fairly
often; Very often
Subcomponent Felt confident about ability to handle In the last month, how often have you felt
personal problems last month (Cohen) confident about your ability to handle your
personal problems?
The answer choices are the following:
Never; Almost never; Sometimes; Fairly
often; Very often
Subcomponent Felt confident things were going your In the last month, how often have you felt
way last month (Cohen) confident that things were going your
way?
The answer choices are the following:
Never; Almost never; Sometimes; Fairly
often; Very often
Subcomponent Felt difficulties were piling up could not In the last month, how often have you felt
overcome them last month (Cohen) difficulties were piling up so high that you
could not overcome them?
The answer choices are the following:
Never; Almost never; Sometimes; Fairly
often; Very often
Secondary Handwashing action Share of adult household members who
Outcome were observed washing their hands at
any juncture, measured via structured
observations.
The considered junctures are the
following:
Before obtaining water from a wide-
mouthed storage container;
Before cutting or preparing food;
Before serving food;
Before eating;
Before feeding child under 5;
Before breastfeeding child;
After defecation;
After toileting;
After cleaning child post- toileting
36
Secondary Handwashing with soap/ash Share of adult household members who
Outcome were observed washing their hands with
soap/ash at any juncture, measured via
structured observations.
The considered junctures are the
following:
Before obtaining water from a wide-
mouthed storage container;
Before cutting or preparing food;
Before serving food;
Before eating;
Before feeding child under 5;
Before breastfeeding child;
After defecation;
After toileting;
After cleaning child post- toileting
Secondary School attendance Number of days child aged 6 to 18 years
Outcome old attended school in past week, based
on responses to “How many days has this
child attended school in the past week?”
Children who were not enrolled in school
were coded as zero.
Secondary Water point has water 1 if water point provides water
Outcome 0 Otherwise
(Locked water points during survey are
not considered)
Secondary WP Coliforms MPN/100mL Water quality at point of collection (village
Outcome water source) is Most Probable Number
(MPN) in 100 mL as defined for
Aquagenx CBT EC+TC MPN water
quality testing kits and follow WHO
standards. For nondetects, we substitute
half the lower detection limit.
Secondary HH Coliforms MPN/100mL Water quality at point of use (Drinking
Outcome water stored in the household) is Most
Probable Number (MPN) in 100 mL as
defined for Aquagenx CBT EC+TC MPN
water quality testing kits and follows WHO
standards. For nondetects, we substitute
half the lower detection limit.
Secondary Weight for age Weight for age z score for children under
Outcome 5.
37
The sample does not include
observations with implausible z-scores.
Secondary Weight for length Weight for length z score for children
Outcome under 5.
The sample does not include
observations with implausible z-scores.
38
S1 Text. CONSORT 2010 checklist of information to include
when reporting a randomized trial*
Reported in
Item Section/
Section/Topic No Checklist item Paragraph
Title and abstract
1a Identification as a randomized trial in the title Title
1b Structured summary of trial design, methods, Abstract
results, and conclusions (for specific guidance see
CONSORT for abstracts)
Introduction
Background and 2a Scientific background and explanation of Introduction
objectives rationale
2b Specific objectives or hypotheses Last
paragraph,
Introduction
Methods
Trial design 3a Description of trial design (such as parallel, Study design
factorial) including allocation ratio
3b Important changes to methods after trial NA
commencement (such as eligibility criteria), with
reasons
Participants 4a Eligibility criteria for participants Study design,
9th paragraph
4b Settings and locations where the data were Study design,
collected 5th paragraph
Interventions 5 The interventions for each group with sufficient Procedures
details to allow replication, including how and
when they were actually administered
Outcomes 6a Completely defined pre-specified primary and Outcomes
secondary outcome measures, including how
and when they were assessed
6b Any changes to trial outcomes after the trial NA
commenced, with reasons
Sample size 7a How sample size was determined Statistical
analysis
7b When applicable, explanation of any interim NA
analyses and stopping guidelines
Randomization:
Sequence 8a Method used to generate the random allocation Randomization
generation sequence and masking
39
8b Type of randomization; details of any restriction Randomization
(such as blocking and block size) and masking
Allocation 9 Mechanism used to implement the random Randomization
concealment allocation sequence (such as sequentially and masking
mechanism numbered containers), describing any steps
taken to conceal the sequence until
interventions were assigned
Implementation 10 Who generated the random allocation sequence, Randomization
who enrolled participants, and who assigned and masking
participants to interventions
Blinding 11a If done, who was blinded after assignment to NA
interventions (for example, participants, care
providers, those assessing outcomes) and how
11b If relevant, description of the similarity of NA
interventions
Statistical 12a Statistical methods used to compare groups for Statistical
methods primary and secondary outcomes analysis
12b Methods for additional analyses, such as Statistical
subgroup analyses and adjusted analyses analysis
Results
Participant flow (a 13a For each group, the numbers of participants who Study design
diagram is were randomly assigned, received intended
strongly treatment, and were analyzed for the primary
recommended) outcome
13b For each group, losses and exclusions after Study design
randomization, together with reasons
Recruitment 14a Dates defining the periods of recruitment and Results
follow-up
14b Why the trial ended or was stopped NA
Baseline data 15 A table showing baseline demographic and Table 2
clinical characteristics for each group
Numbers 16 For each group, number of participants Tables 3-4
analysed (denominator) included in each analysis and
whether the analysis was by original assigned
groups
Outcomes and 17a For each primary and secondary outcome, Results,
estimation results for each group, and the estimated effect Tables 3-4
size and its precision (such as 95% confidence
interval)
17b For binary outcomes, presentation of both Tables 3-4
absolute and relative effect sizes is
recommended
40
Ancillary analyses 18 Results of any other analyses performed, Results; Supp
including subgroup analyses and adjusted Info
analyses, distinguishing pre-specified from
exploratory
Harms 19 All important harms or unintended effects in NA
each group (for specific guidance see CONSORT for harms)
Discussion
Limitations 20 Trial limitations, addressing sources of potential Discussion, 2nd
bias, imprecision, and, if relevant, multiplicity of to last
analyses paragraph
Generalizability 21 Generalizability (external validity, applicability) of Discussion
the trial findings
Interpretation 22 Interpretation consistent with results, balancing Discussion
benefits and harms, and considering other
relevant evidence
Other information
Registration 23 Registration number and name of trial registry Study design
Protocol 24 Where the full trial protocol can be accessed, if Study design
available
Funding 25 Sources of funding and other support (such as Abstract
supply of drugs), role of funders
*We strongly recommend reading this statement in conjunction with the CONSORT 2010 Explanation and
Elaboration for important clarifications on all the items. If relevant, we also recommend reading CONSORT
extensions for cluster randomized trials, non-inferiority and equivalence trials, non-pharmacological treatments,
herbal interventions, and pragmatic trials. Additional extensions are forthcoming: for those and for up-to-date
references relevant to this checklist, see www.consort-statement.org.