Water Meta-Analysis

Code for Water Treatment and Child Mortality: A Meta-analysis and Cost-effectiveness Analysis by Kremer, Luby, Maertens, Tan, and Więcek.

Computational requirements

All code runs in R (last run on R version 4.2.3 and RStudio version 2023.09). renv.lock lists all necessary packages and their respective versions. main.R last run on a 10-core Intel-based laptop with Windows 10 and 16.0 GB of RAM. Approximate runtime was 95 minutes.

Description of programs/code

To recreate the exhibits in the paper,

1. Open water-ma.RProj
2. To install all the necessary packages, run renv::restore() (this step requires the package renv)
3. Run main.R

The code used to create the exhibits in the paper is in code/. main.R tracks all the inputs and outputs for each script. Below is an overview of what they do.

• Scripts in code/functions create functions that are used across multiple scripts. The are sourced by .Rprofile when water-ma.RProj is launched.
• Scripts in code/wrangling process the raw data and create the final data sets used across different analysis scripts. The first section of main.R runs the scripts in this directory. Running this section is optional, since it will only recreate the data sets in data/final.
• Scripts in code/cea prepare the data used for cost-effectiveness analysis and include auxiliary code used in the cost-effectiveness analysis. The second section of main.R runs the scripts in this directory. Running this section is also optional. It will recreate data/final/mortality_rate.rds and the under-five mortality rates in data/transformed/u5-per-hh.csv and data/transformed/weighted_u5_mr.csv.
• Scripts in code/ma_models fit meta-analysis models.
  • fit_ma_bayes.R uses rstan model in logit_model.stan. It takes a moment to run and will save outputs to output/stan.
  • load_bayes_ma.R is called from other scripts to load the outputs of fit_ma_bayes.R.
  • fit_ma_frequentist.R is called from other scripts as fits frequentist meta-analysis models.
• Scripts in code/generate_outputs recreate the results in the paper.
  • generate_figures.R recreates all the figures.
  • code/generate_outputs/generate_tables.R recreates the meta-analysis tables in the paper.
  • code/generate_outputs/generate_cea_results.R recreates the cost-effectiveness results.
  • code/generate_outputs/generate_summary_of_individual_studies.R recreates the table with summary of individual studies. The data required to run this script cannot be publicly shared.
  • generate_text.R for printing all values that are cited inline in main text and supplement
• Scripts in code/publication_bias perform the tests mentioned in supplement section 5.

Controlled Randomness

License for code

Code is licensed under , with the exception of code/publication_bias/RobustVariance.R and code/publication_bias/metastudiesfunctions.R by Maximilian Kasy, which are licensed under .

List of paper exhibits and programs

Paper	File	Created by
Fig 2A	output/figures/freq-forest.pdf	generate_figures.R
Fig 2B	output/figures/bayes-forest.pdf	generate_figures.R
Tab 2	output/figures/table-cea-estimates	generate_cea_results.R
Fig S1	output/figures/ma-week-plot.pdf	generate_figures.R
Fig S2	output/figures/fig-compliance-diarr-hist.pdf	generate_figures.R
Fig S3	output/figures/funnel.pdf	generate_figures.R
Fig S4	output/figures/diarr-pub-bias-funnel.pdf	generate_figures.R
Fig S5	output/figures/bubble-plot-year.pdf	generate_figures.R
Fig S6	output/figures/mortality-vs-baseline.pdf	generate_figures.R
Fig S7	output/figures/dist-diarrhea-prevalence.pdf	generate_figures.R
Tab S3	output/tables/mortality_all_summary.csv	generate_tables.R
Tab S4	output/tables/table-loo-study.csv	generate_tables.R
Tab S5	output/tables/additional-sa-results.csv	generate_tables.R
Tab S6	output/tables/cea-globalbenefits.csv	generate_cea_results.R
Tab S7	output/tables/table-cea-estimates.csv	generate_cea_results.R

Data Availability and Provenance Statements

Some data used in this study cannot be made publicly available. The raw data files used by the code are listed below.

Data	File	Provided	Citation	License	Notes
List of diarrhea studies	data/raw/diarrhea_studies.xlsx	Yes			Contains data provided in the supplementary material Wolf et al., 2018. The original dataset extracted from the publication is available here. The data was provided for 80 studies included in the meta-analysis. We added the relevant data for studies included in the meta-analysis but not included in the Wolf et al., 2018 study (Peletz et al., 2012, Null et al., 2018, Luby et al., 2018, Humphrey et al., 2019, Kirby et al., 2019, Haushofer et al., 2020, Dupas et al., 2021, and ucation vs control) Quick et al., 1999, Conroy et al. 1999, Morris et al. 2018). In addition to the existing data, we added the following information for each of the studies: (1) Compliance rate, and (2) How is compliance defined. Wolf et al., 2018 collects estimates for under-5 diarrhea morbidity in studies with any WaSH intervention.
Studies summary	data/raw/summary_data.csv	Yes			Summary data on each RCT, number of cases in treatment, number of cases in control etc.
UN World Population Prospects, 2019 revision	data/raw/weighted_mr/WPP2019_POP_F07_1_POPULATION_BY_AGE_BOTH_SEXES.xlsx	Yes	United Nations, Department of Economic and Social Affairs, Population Division ([2019]). World Population Prospects [2019], archive.	Copyright © 1992-2022 by United Nations	Downloaded from https://population.un.org/ on on Apr 5, 2023
UNdata M49 Country Codes	data/raw/u5-per-hh/UNSD.csv	Yes		Terms of use available at https://data.un.org/	Copied from https://unstats.un.org/ on on Apr 5, 2023
Studies microdata	Files in data/raw/mortality_counts	No			This data was acquired directly from the authors of each paper included in the study.
DHS data	data/raw/u5-per-hh/idhs_00001	No	Elizabeth Heger Boyle, Miriam King and Matthew Sobek. IPUMS-Demographic and Health Surveys: Version 9 [dataset]. IPUMS and ICF, 2022. https://doi.org/10.18128/D080.V9		Accessed through [https://dhsprogram.com/data/Access-Instructions.cfm] on Apr 5, 2023
IHME Diarrhea Prevalence	data/raw/IHME_GLOBAL_DIARRHEA_2000_2019_PREV_A1_S3_ADMIN_1_Y2020M08D31.CSV	No	Institute for Health Metrics and Evaluation (IHME). Global Under-5 Diarrhea Incidence, Prevalence, and Mortality Geospatial Estimates 2000-2019. Seattle, United States of America: Institute for Health Metrics and Evaluation (IHME), 2020.	Terms of use can be found here	Downloaded from https://ghdx.healthdata.org/) on on Jul 28, 2023
World Bank region and income groups	data/raw/weighted_mr/CLASS.xlsx	Yes	World Bank: World Bank Country and Lending Groups. Downloaded on April 5, 2023.	Detailed terms of use for data is available here	Downloaded from https://datahelpdesk.worldbank.org on Apr 5, 2023
	data/raw/weighted_mr/all_data.csv
WHO/UNICEF Joint Monitoring Programme for Water Supply, Sanitation and Hygiene (JMP) - Household data	data/raw/weighted_mr/JMP_2021_WLD.xlsx	Yes	Progress on household drinking water, sanitation and hygiene 2000–2022: special focus on gender. New York: United Nations Children’s Fund (UNICEF) and World Health Organization (WHO), 2023.		Downloaded from https://washdata.org on on Apr 5, 2023
WHO/UNICEF Joint Monitoring Programme for Water Supply, Sanitation and Hygiene (JMP) - Access to drinking water	data/raw/weighted_mr/washdash-download.csv	Yes	Progress on household drinking water, sanitation and hygiene 2000–2022: special focus on gender. New York: United Nations Children’s Fund (UNICEF) and World Health Organization (WHO), 2023.		Downloaded from https://washdata.org/ on Apr 5, 2023

Intermediate data sets created during data processing are stored in data/transformed. The final data sets used for analysis or to create exhibits in the paper are stored in data/final and listed below

File	Created by	Provided	Citation
data/final/diarrhea_studies.rds	code/wrangling/clean_studies_list.R	Yes
data/final/ma_datasets.Rdata	code/wrangling/prep_adjusted_data.R	Yes
data/final/mortality_rate.rds	code/cea/weighted-mr.R	Yes
data/final/individual_data_anonymised.rds	code/individual_data_anonymised/pre_individual_data.R	Yes