1. Objectives

To determine the effectiveness of interventions to improve water quality for preventing diarrhoea

2. How studies were identified

The following databases were searched to November 2014:

• Cochrane Infectious Diseases Group Specialized Register
• CENTRAL (The Cochrane Library)
• MEDLINE
• EMBASE
• LILACS

In addition, relevant conference proceedings were searched, as were reference lists of all identified studies. Individual researchers working in the field and the following organizations were approached to identify unpublished and ongoing trials: Water, Sanitation and Health Programme of the WHO; Water and Sanitation Program of the World Bank; UNICEF Water, Environment and Sanitation; the IRC International Water and Sanitation Centre; the CDC Foodborne and Diarrhoeal Diseases Branch, Division of Bacterial and Mycotic Diseases; USAID, including its Environmental Health Project; and the UK Department for International Development

3. Criteria for including studies in the review

3.1 Study type

Cluster-randomized trials, quasi-randomized trials, and controlled before-and-after studies

3.2 Study participants

Children and adults

3.3 Interventions

Interventions aimed at improving the microbiological quality of drinking water in comparison to no intervention or a placebo intervention

(Studies combining improvements in water quality with other hygiene or health promotion interventions were included, but studies combining water quality interventions with other water, sanitation, and hygiene (WASH) interventions were excluded)

(Studies focusing the intervention on water quality in places outside the home, such as schools, clinics, markets, or workplaces, were excluded)

3.4 Primary outcomes

• Diarrhoea episodes among individuals (regardless of microbiological confirmation or trial definition of diarrhoea)

Secondary outcomes included death and adverse events

4. Main results

4.1 Included studies

Fifty-five studies involving 84,023 participants, were included in this review:

• Fifty-two trials, corresponding to 65 comparisons due to multiple intervention arms, were included in quantitative analysis
• Forty-five studies were cluster-randomized controlled trials, two were quasi-randomized controlled trials, and eight were controlled before-and-after studies. Units of randomization included households, villages, communities, and neighbourhoods
• Most studies enrolled both children and adults; nine studies included data only for children less than five years of age and three studies included data only for adults; trial duration ranged from eight weeks to four years
• Eight studies evaluated source-based interventions: improved wells/boreholes or improved community sources and distribution to public tap stands
• Forty-seven studies evaluated point-of-use (POU) interventions: chlorination (17 studies), filtration (20 studies), combined flocculation and disinfection sachets (5 trials), SODIS solar disinfection (6 studies), combination UV disinfection and filtration (1 trial), and improved storage (2 trials)
• Many trials also provided hygiene education or improved water storage facilities, and only three multiple-intervention arm studies attempted to isolate the impact of water quality alone
• Control arms included placebo interventions, use of usual water supply, and use of usual water supply plus storage containers
• Definitions and surveillance of diarrhoea varied across trials

4.2 Study settings

• Fifty studies were conducted in lower-middle or low-income countries; five studies were conducted in high-income countries: Australia, Saudi Arabia, and the United States of America (3 studies)
• Most studies were conducted in villages/rural settings; five trials were conducted in urban settings, five in peri-urban settings, two in informal urban/squatter settlements, two in refugee/displaced persons camps, and five in multiple settings
• Prior to the intervention, drinking water was unimproved (unprotected springs, vendor-/tanker-provided, bottled) in 30 studies, improved (household connections, public standpipes, boreholes, protected dug wells/springs, rainwater) in 15 studies, and unclear/unreported in five studies. Sanitation facilities were improved (connection to a public sewer or septic system, pour-flush latrine, simple pit latrine, ventilated improved pit latrine) in 12 studies, unimproved (service/bucket latrines, public latrines, open latrines) in 15 studies, and unclear/unreported in 19 studies. Access to a water source was sufficient in 14 studies, insufficient in four studies, and unclear/unreported in the remaining studies. The quantity of water available to study participants was sufficient (≥15 L/person/day) in eight studies, insufficient (<15 L/person/day) in four studies, and unclear in 43 studies

4.3 Study settings

How the data were analyzed
Eight comparisons were made: i) any water quality intervention versus control; ii) interventions at the water source versus control; iii) POU chlorination versus control; iv) POU combined flocculation and disinfection versus control; v) POU filtration versus control; vi) POU solar disinfection versus control; vii) POU UV disinfection versus control; and viii) POU improved storage versus control. Random effects meta-analysis was use to generate pooled estimates of rate ratios (RR) and corresponding 95% confidence intervals (CI), with odds ratios being transformed to RR using standard Cochrane methods. Where studies included multiple intervention arms, each arm was compared separately to the control, although this double counting of control participants is known to produce artificially precise results. Sensitivity analyses were performed restricting analyses to trials at low risk of bias. To explore potential sources of heterogeneity, the following subgroup analyses were planned:

• By age: all ages, children <5 years
• By intervention type: source water improvement, POU treatment
• By study design: cluster-randomized controlled trials, controlled before-and-after studies
• By adherence with intervention: <50%, 50% to 85%, >85%
• By water source: unimproved, improved, unclear
• By water source access: sufficient, insufficient
• By water quantity: insufficient, sufficient
• By sanitation conditions: unimproved, improved, unclear
• By country income level: high, upper-middle, lower-middle, low
• By length of follow-up

Results
Any water quality intervention versus control
Diarrhoea episodes
Overall, any water quality intervention reduced the risk of diarrhoea by 41% in comparison to the control (RR 0.59, 95% CI [0.51 to 0.69], p<0.00001; 64 comparisons/81,215 participants). This finding remained statistically significant when restricted to children <5 years old (RR 0.61, 95% CI [0.49 to 0.75], 49 comparisons), and for POU interventions (RR 0.58, 95% CI [0.48 to 0.69], 58 comparisons/72,054 individuals), but not for source water improvement interventions (RR 0.76, 95% CI [0.48 to 1.19], 6 comparisons/9181 individuals).

Additional outcomes
Five studies reported the number of deaths in each intervention arm, with no differences between treatment groups. None of the included trials reported on adverse events.

Interventions at the water source
In one cluster-randomized trial including 3266 participants, the risk of diarrhoea was not statistically significantly different between treatment groups (RR 1.24, 95% CI [0.98 to 1.47], p=0.073). In five controlled before-after studies, the risk was also not statistically significantly different between groups (RR 0.68, 95% CI [0.42 to 1.09], p=0.11). Results did not differ meaningfully when subgrouped by age.

POU: water chlorination versus control
Overall, the risk of diarrhoea was reduced by 28% among the intervention group (RR 0.72, 95% CI [0.61 to 0.84], p=0.000053; 19 comparisons/34,694 participants). This finding became non-significant for trials >12 months in duration, settings with unimproved/unclear sanitation, settings with an improved water source, settings with sufficient or insufficient water quantity described, trials using no water storage kits, trials at low risk of bias due to blinding, and trials in which residual chlorine was found in ≤50% of samples or was unreported.

POU: flocculation and disinfection versus control
The risk of diarrhoea was not significantly reduced with POU flocculation and disinfection overall (RR 0.48, 95% CI [0.20 to 1.16], p=0.10; 7 comparisons), but when one trial with a large reduction in risk (88%) was excluded, the results became statistically significant (RR 0.69, 95% CI [0.58 to 0.82], p=0.000037; 6 comparisons/11,788 individuals). Results also became non-significant in trials where a storage container was provided, where water quantity was insufficient, where water source was unimproved, where sanitation was unimproved, and where trial duration was less than six months.

POU: filtration versus control
The risk of diarrhoea was halved with POU filtration among children <5 years of age (RR 0.49, 95% CI [0.38 to 0.62], p<0.00001 19 comparisons), and among all ages (RR 0.48, 95% CI [0.38 to 0.59], p<0.00001; 23 comparisons). This finding was rendered non-significant in analyses restricted to plumbed filtration methods and interventions lasting >12 months.

POU: solar disinfection versus control
Episodes of diarrhoea were reduced with solar disinfection in both cluster-randomized trials (RR 0.62, 95% CI [0.42 to 0.94], p=0.023; 4 comparisons/3460 participants) and quasi-randomized trials (RR 0.82, 95% CI [0.69 to 0.97], p=0.021; 2 comparisons/555 individuals). In subgroup analyses, findings became non-significant for trials with low adherence (≤50%), settings with sufficient or unclear water supply, settings with both improved and unimproved water sources, settings with unimproved sanitation, and for trials lasting <12 months.

POU: UV disinfection versus control
In one trial including 1913 participants, UV disinfection did not reduce the risk of diarrhoea in comparison to the control (RR 0.79, 95% CI 0.49 to 1.27]).

POU: improved storage versus control
Improved water storage did not statistically significantly reduce the risk of diarrhoea for all ages (RR 0.91, 95% CI [0.74 to 1.11], p=0.34; 2 comparisons), or for children <5 years of age (RR 0.69, 95% CI [0.47 to 1.01], p=0.056; 1 comparison).

5. Additional author observations*

The evidence for improvements at the water source was considered to be very low quality; for POU water chlorination was judged as being low quality; and for POU flocculation and disinfection, POU filtration, and POU solar disinfection was considered to be of moderate quality. In addition to unexplained heterogeneity, the other main reasons for downgrading the quality of evidence was the risk of bias due to lack of blinding and the use of a self-reported outcome. Importantly, none of the included studies investigated the impact of boiling, which is the most common type of POU water treatment.

The evidence summarized in this review supports the implementation of interventions to improve water quality by addressing microbial contamination at the POU until households can obtain piped, safe and reliable water connections.

Further studies employing rigorous methodology are needed, including further water source intervention trials in which drinking water quality is maintained through to the POU. Longer-term studies of water quality interventions among vulnerable populations are also required.