Water and sanitation interventions for better child health: Evidence from a
synthetic review Hugh Waddington Birte Snilstveit Howard White Lorna Fewtrell
3ie
2
Contents
n Objectives
n Methodology
n Results: effectiveness
n Results: sustainability
n Conclusions
3
1. Objectives of the synthetic review
Conduct review to Cochrane/Campbell standards
Mixed methods: analysis of quantitative and qualitative information
Synthesise quantitative information using meta-analysis: assess whether existing ‘consensus’ stands up to inclusion of new studies, and internal and external validity criteria
Theory-based: draw out behavioural and contextual factors shaping success/failure and (likely) sustainability
4
2. Methodology Extensive search (published and unpublished sources)
Inclusion criteria: impact of WSH on diarrhoea; experimental and quasi-experimental methods
Coding of studies: Internal validity: study design; data quality External validity: contextual information Quantitative and qualitative information relating to process, context, behaviour,
confounding, moderation
Effectiveness of interventions assessed using: Meta-analysis (fixed/random effects models; impact heterogeneity, publication
bias) Meta-regression
Sustainability of interventions assessed using quantitative and qualitative information relating to process (outputs), study length and size, replicability
5
6
3. Results: about the interventionsTotal num
Num RCTs
Total sample
Ave sample
Ave length (months)
Water supply 14 0 64,651 5,877 19
Water treatment 40 27 16,400 586 10
Point of use water treatment (POU)
35 24 11,457 327 7
Source water treatment 5 2 4,943 989 15
Sanitation 13 2 33,682 3,368 16
Hygiene 38 14 29,254 1,083 12
Hand-washing with soap 9 4 12,472 1,386 6
Education 29 10 16,782 579 13
7
Effectiveness: Water supply interventions
NOTE: Weights are from random effects analysis
Overall (I-squared = 75.8%, p = 0.000)
Gasana et al (2002)
Jalan & Ravallion (2003)
Study
Galiani et al (2007)
Pradhan & Rawlings (2002)
Alam et al (1989)
Tonglet et al (1992)
ID
Khanna (2008)
Ryder et al (1985)
0.92 (0.83, 1.01)
1.08 (0.92, 1.27)
0.76 (0.69, 0.84)
0.50 (0.19, 1.31)
0.74 (0.37, 1.49)
0.83 (0.72, 0.96)
0.94 (0.89, 1.00)
ES (95% CI)
0.92 (0.85, 1.00)
1.34 (1.00, 1.80)
0.92 (0.83, 1.01)
1.08 (0.92, 1.27)
0.76 (0.69, 0.84)
0.50 (0.19, 1.31)
0.74 (0.37, 1.49)
0.83 (0.72, 0.96)
0.94 (0.89, 1.00)
ES (95% CI)
0.92 (0.85, 1.00)
1.34 (1.00, 1.80)
Favours intervention 1.25 .5 .75 1 2
8
Effectiveness: Water treatment interventions
NOTE: Weights are from random effects analysis
Overall (I-squared = 87.7%, p = 0.000)
Stauber et al (2009)
Kirchhoff et al (1985)
Chiller et al (2006)
Mahfouz et al (1995)
de Wilde et al (2008)
Jensen et al (2003)
Gasana et al (2002)
Lule et al (2005)
Conroy et al (1999)
Semenza et al (1998)
Reller et al (2003)
Austin (1993)
Clasen et al (2005)
du Preez et al (2008)
Sobsey et al (2003)
Rose et al (2006)
Crump et al (2005)
Brown et al (2008)
Kremer et al (2008)
Roberts et al (2001)
Clasen et al (2006)
Brown et al (2007)
Blanton et al (2008)
Luby et al (2004)
Tiwari et al (2009)
Quick et al (1999)
ID
Luby et al (2006)
Clasen et al (2004)
Universidad Rafael Landivar (1995)
Study
0.64 (0.57, 0.72)
0.74 (0.51, 1.07)
0.97 (0.78, 1.21)
0.61 (0.45, 0.83)
0.55 (0.31, 0.99)
1.02 (0.88, 1.19)
0.94 (0.89, 1.00)
0.67 (0.55, 0.81)
0.90 (0.51, 1.59)
0.69 (0.63, 0.75)
0.33 (0.19, 0.57)
0.74 (0.58, 0.94)
1.01 (0.19, 5.38)
0.40 (0.21, 0.76)
0.22 (0.13, 0.38)
0.79 (0.74, 0.84)
0.63 (0.47, 0.84)
0.75 (0.60, 0.94)
0.58 (0.41, 0.82)
0.64 (0.44, 0.93)
0.68 (0.45, 1.02)
0.48 (0.22, 1.06)
0.52 (0.32, 0.85)
1.10 (0.96, 1.26)
0.30 (0.17, 0.54)
0.49 (0.24, 1.01)
0.46 (0.25, 0.86)
ES (95% CI)
0.68 (0.42, 1.08)
0.29 (0.23, 0.36)
0.47 (0.20, 1.11)
0.64 (0.57, 0.72)
0.74 (0.51, 1.07)
0.97 (0.78, 1.21)
0.61 (0.45, 0.83)
0.55 (0.31, 0.99)
1.02 (0.88, 1.19)
0.94 (0.89, 1.00)
0.67 (0.55, 0.81)
0.90 (0.51, 1.59)
0.69 (0.63, 0.75)
0.33 (0.19, 0.57)
0.74 (0.58, 0.94)
1.01 (0.19, 5.38)
0.40 (0.21, 0.76)
0.22 (0.13, 0.38)
0.79 (0.74, 0.84)
0.63 (0.47, 0.84)
0.75 (0.60, 0.94)
0.58 (0.41, 0.82)
0.64 (0.44, 0.93)
0.68 (0.45, 1.02)
0.48 (0.22, 1.06)
0.52 (0.32, 0.85)
1.10 (0.96, 1.26)
0.30 (0.17, 0.54)
0.49 (0.24, 1.01)
0.46 (0.25, 0.86)
ES (95% CI)
0.68 (0.42, 1.08)
0.29 (0.23, 0.36)
0.47 (0.20, 1.11)
Favours intervention
1.1 .5 .75 1 2
9
Effectiveness: Sanitation interventions
NOTE: Weights are from random effects analysis
Overall (I-squared = 97.9%, p = 0.000)
Pradhan & Rawlings (2002)
Moraes et al (2003)
ID
Chavasse et al (1999)
Kolahi et al (2008)
Root (2001)
Study
0.52 (0.29, 0.96)
0.43 (0.07, 2.70)
0.31 (0.28, 0.34)
ES (95% CI)
0.78 (0.70, 0.87)
0.96 (0.75, 1.24)
0.32 (0.18, 0.57)
0.52 (0.29, 0.96)
0.43 (0.07, 2.70)
0.31 (0.28, 0.34)
ES (95% CI)
0.78 (0.70, 0.87)
0.96 (0.75, 1.24)
0.32 (0.18, 0.57)
Favours intervention 1.1 .5 .75 1 2
10
Effectiveness: Hygiene interventions
NOTE: Weights are from random effects analysis
Overall (I-squared = 91.3%, p = 0.000)
Luby et al (2005)
Ahmed et al (1993)
Han & Hlaing (1989)
Haggerty et al (1994)
Bateman et al (1995)
Luby et al (2008)
Sircar et al (1987)
Pattanayak et al (2007)
Shahid et al (1996)
Wilson et al (1991)
Luby et al (2004)
Torun (1982)
Pinfold & Horan (1996)
Luby et al (2006)
Stanton et al (1988)
ID
Lee et al (1991)
Study
0.68 (0.60, 0.76)
0.47 (0.36, 0.61)
0.64 (0.61, 0.66)
0.75 (0.60, 0.93)
0.94 (0.87, 1.03)
0.27 (0.19, 0.38)
0.95 (0.69, 1.32)
0.92 (0.72, 1.17)
0.77 (0.68, 0.87)
0.44 (0.37, 0.53)
0.14 (0.03, 0.62)
0.71 (0.49, 1.04)
0.81 (0.75, 0.87)
0.61 (0.37, 1.00)
0.72 (0.47, 1.11)
0.78 (0.74, 0.83)
ES (95% CI)
0.67 (0.58, 0.77)
0.68 (0.60, 0.76)
0.47 (0.36, 0.61)
0.64 (0.61, 0.66)
0.75 (0.60, 0.93)
0.94 (0.87, 1.03)
0.27 (0.19, 0.38)
0.95 (0.69, 1.32)
0.92 (0.72, 1.17)
0.77 (0.68, 0.87)
0.44 (0.37, 0.53)
0.14 (0.03, 0.62)
0.71 (0.49, 1.04)
0.81 (0.75, 0.87)
0.61 (0.37, 1.00)
0.72 (0.47, 1.11)
0.78 (0.74, 0.83)
ES (95% CI)
0.67 (0.58, 0.77)
Favours intervention 1.1 .5 .75 1 2
11
Heterogeneity – ‘high quality’
0
0.25
0.5
0.75
1
1.25
All studies(7 obs)
'Quality'(2 obs)
POU (3obs)
All studies(29 obs)
'Quality'(23 obs)
All studies(5 obs)
‘Quality’(3 obs)
All studies(16 obs)
‘Quality’(12 obs)
Soap(7 obs)
Water supply Water treatment Sanitation Hygiene
12
A closer look at water treatment
Effect 95% CI Num studies
Point of use 0.60 0.52 0.68 21
Source 0.85 0.63 1.15 2
No placebo/blinding 0.60 0.52 0.70 22
Placebo/blinded trials 0.75 0.56 1.01 6
Placebo/blinded trials excl. Luby et al. (2004, 2006) 0.90 0.65 1.25 4
Possible conflict of interest 0.51 0.36 0.73 7
No conflict of interest declared 0.69 0.60 0.78 19
13
4. Sustainability Water supply and sanitation: sustainability/scalability assessed
Water treatment and hygiene: replicability demonstrated; sustainability and scalability not=> only 5 follow-up studies assessing compliance + health impact over one
year after intervention completed
Most studies assess adoption/compliance with intervention => success / failure
Contextual factors identified, often through use of moderator (interaction variable) in analysis: e.g. age, carer’s education level, income, time.
Behavioural factors identified: individual preferences (e.g. taste) and agency (intra-household effects), intra-community ownership + interpersonal networks
14
Sustainability: quantitative
0
0.5
1
1.5
> 12 months(5 obs)
≤ 12 months(1 obs)
> 12 months(1 obs)
≤ 12 months(26 obs)
> 12 months(4 obs)
≤ 12 months(4 obs)
> 12 months(11 obs)
≤ 12 months(31 obs)
Water supply POU w ater treatment Source w ater treatment Total interventions
15
Sustainability: qualitative (water treatment) 3 follow-up studies of successful trials were conducted over one
year after the initial intervention ended:
Brown et al (2007): ceramic filter provision in Cambodia; 3 years later only 31% compliance
Iijima et al (2001): pasteurisation in Kenya; 4 years later only 30% compliance
Luby et al (2008): water disinfectant in Guatemala; compliance (repeated use) only 5%.
One study evaluated the reasons for low compliance of unsuccessful intervention: source water treatment (UV filtration) in rural Mexico (de Wilde et al 2008):
Community capacity to manage, physical faults or under-valuing of safe water by users were NOT found to be limiting the intervention’s effectiveness
Constraints (money & time) and availability of other sources, meant households chose more convenient water sources
16
No one single intervention for improving global access to water and sanitation for reducing diarrhoeal disease:
the ‘right’ solution is the one that fits the (social, economic, political) context
emphasise behavioural factors, particularly where these are of overriding importance to adoption (water quality + hygiene interventions)
5. Conclusions
17
5. Conclusions (cont’d)
Effectiveness: Water supply interventions least effective, excl household
connection Water treatment at point-of-use very effective, but concerns
about study quality (blinding) and conflict of interest Water treatment at source less effective but few high quality
studies Sanitation effective – more studies needed Hygiene interventions are effective (at least in short term) but
resource intensive Interventions substitutes (results not reported)
Evidence on sustainability + scalability of water treatment and hygiene interventions limited