Water Treatment and Child Survival: A Meta-Analysis

Last registered on July 13, 2020


Trial Information

General Information

Water Treatment and Child Survival: A Meta-Analysis
Initial registration date
July 05, 2020
Last updated
July 13, 2020, 3:58 PM EDT



Primary Investigator

Harvard University

Other Primary Investigator(s)

PI Affiliation
Harvard University

Additional Trial Information

Start date
End date
Secondary IDs
Measuring program impacts on child mortality requires large sample sizes. Randomized controlled trials (RCTs) on water treatment are therefore typically designed and powered only to detect effects on the more common intermediate outcome of caregiver reported diarrhea, which may or may not be a good proxy for water-related mortality. In this paper, we increase statistical power for examining the child survival effect of improved water quality by conducting a meta-analysis. To address the possibility that researchers are more likely to report child survival findings if estimates are positive, we systematically contact the authors of all RCTs on water interventions in the developing world to find studies with data on mortality outcomes.
External Link(s)

Registration Citation

Kremer, Michael and Brandon Tan. 2020. "Water Treatment and Child Survival: A Meta-Analysis." AEA RCT Registry. July 13. https://doi.org/10.1257/rct.5977-1.0
Experimental Details


We conduct a meta-analysis combining all existing RCT evidence on water interventions in the developing world.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
All-cause under-5 mortality
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We contact the authors of all RCTs on water interventions in the developing world to identify all studies from 1970 - 2020 with data on mortality outcomes even when not reported in the original publication.
Experimental Design Details
The selection of papers to be included in our analysis is as follows. We first review all papers identified by Clasen et al. (2015) and Wolf et al. (2018) for their meta-analyses of the impact of interventions to improve water quality on diarrhea which covers studies from 1970 up to February 2016. Next, we replicate their selection criteria and search procedure for the period from February 2016 to May 2020 to add more recent studies. We systematically search Pubmed, Embase, Scopus and Cochrane Library using both keywords and MeSH terms to identify all water intervention studies following Wolf et al. (2018). We then search the reference sections of all papers for additional papers for inclusion. We also gather further references from subject matter experts. We impose additional restrictions for inclusion conditional on studying a water intervention. First, we require that the study is a randomized controlled trial. Interventions needed to be tested against a control group that did not receive the respective intervention(s) or that received a control or placebo intervention. We exclude studies where the control group was contaminated by treatment from the main sample, but include them in robustness checks. Second, we require that the study is in a low- or middle-income country, according to the World Bank classification. Following Clasen et. al (2015) and Wolf et. al (2018) we exclude studies that mainly targeted institutions (e.g. schools or the work place), or cases where the study population is considered to be non-representative (e.g. interventions targeting HIV+ populations). We systematically contact the authors of all studies identified to obtain all-cause child mortality data even when not reported in the original publication. We exclude studies where the control group is contaminated from the main analysis, but consider them for robustness checks.

We follow standard practice in the health literature and use odds ratios (ORs) for our main analysis. We use both frequentist and Bayesian methods for meta-analysis. Our main frequentist specification is the Peto one-step odds ratio method, but we also report results using inverse-variance random effects odds ratio estimation for completeness. Our Bayesian approach uses a hierarchical logistic regression model. Additionally, we show that the results are robust to: 1) Different definitions of control groups and treatment groups, 2)The inclusion of studies in which the control group is contaminated, and 3) The exclusion of any single particular study.

We classify our studies by whether it pre-specifies mortality as an outcome and whether it reported results on mortality in the original publication. We report estimates of the mean effect of interventions to improve water quality on child mortality studies which pre-specified mortality as an outcome and were the mortality results were published even though they found no significant results. We also report estimates for the excluded studies to assess whether they are significantly different. We use methods from Andrews and Kasy (2019), Begg and Mazumdar (1994) and Egger et al., (1997) to test for publication bias.
Randomization Method
Varies by study.
Randomization Unit
Varies by study.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
Varies by study.
Sample size: planned number of observations
Varies by study.
Sample size (or number of clusters) by treatment arms
Varies by study.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number


Post Trial Information

Study Withdrawal

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information


Is the intervention completed?
Intervention Completion Date
June 01, 2020, 12:00 +00:00
Data Collection Complete
Data Collection Completion Date
June 01, 2020, 12:00 +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
13 studies
Was attrition correlated with treatment status?
Final Sample Size: Total Number of Observations
13 studies
Final Sample Size (or Number of Clusters) by Treatment Arms
13 studies
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials