water, sanitation and hygiene: interventions and diarrhoea – a review lorna fewtrell & jack...
TRANSCRIPT
Water, sanitation and hygiene: interventions and diarrhoea – a
review
Lorna Fewtrell & Jack Colford
Introduction (1) Diarrhoeal disease continues to be one of
the leading causes of morbidity and mortality in developing countries
Introduction (2) DIARRHOEA KILLS PEOPLE
Introduction (3) The important role of sanitation and safe
water in maintaining health has been recognised for centuries
1980s – International Drinking Water Supply and Sanitation Decade
Reviews of the effectiveness of various levels of water supply and sanitation published
Introduction (3)
30
36
17
15
20
33
0 10 20 30 40 50
Water & Sanitation
Sanitation
Water Quality & Quantity
Water Quality
Water Quantity
Hygiene
% reduction in diarrhoeal illness
Objectives
These are now quite dated so the objective of this review was to update the previous work in the area with a view to informing interested parties on the relative effectiveness of possible interventions addressing water, sanitation and hygiene.
Search strategy (1)
Medline and Embase databases searched using key words pairing, diarrhoea or intervention against: Sanitation Water quality Water quantity Hygiene Drinking water
Search strategy (2) Database searches were restricted to papers
relating to humans dated prior June 26, 2003 The Esrey reviews were used to identify studies
published prior to 1985 Abstracts, where available, were examined –
and papers which appeared to be relevant were obtained for further review
Initial selection criteria The article reported diarrhoea morbidity
as a health outcome under endemic (or non-outbreak) conditions; and
The article reported specific water, sanitation and/or hygiene intervention(s), or some combination thereof
Intervention classification (1)
Hygiene – includes hygiene and health education and the encouragement of specific behaviours (such as handwashing)
Sanitation – those interventions that provided some means of excreta disposal, usually the provision of latrines (at public or private level)
Intervention classification (2)
Water supply – included the provision of a new water source and/or improved distribution (such as installation of a handpump or a household connection)
Water quality – these were related to the provision of water treatment, either at source or household level
Intervention classification (3)
Multiple – those which introduced water, sanitation and hygiene (or health education) elements to the study population
Data extraction (1) Study location Study design Study length Study period Sample size Data collection method Participant age band Confounders examined
Study design Range of epidemiological study designs
that can be (and in many cases, have been) applied to study the impact of improvements to water, sanitation and health: Intervention Case-control Ecological
Data extraction (2) Illness definition Recall period Type & level of water supply and sanitation
(pre-intervention) Water source Intervention Relative risk and 95% CI
Data extraction (3) Relative risk included:
odds ratios, incidence density ratios, cumulative incidence ratios
When both adjusted and unadjusted (for other covariates) measures were reported – the most adjusted estimate was used
Data extraction (4) RR and 95% CI expressed such that a
RR of less than unity means that the intervention group has a reduced frequency of diarrhoea in comparison to the control group
Meta-analysis (intro) Meta-analysis is a tool that allows the
statistical pooling of data across studies to generate a summary estimate of effects Where ‘effect’ is any measure of association
between exposure and outcome (e.g. odds ratio)
It is not always appropriate to conduct a meta-analysis
Meta-analysis(1) Risk estimates from the selected studies were
pooled in meta-analysis using STATA software (STATA Corporation, College Station, TX, USA, version 8)
STATA commands for meta-analysis are not an integral part of the original software but are additional, user-written, add-on programs that can be freely downloaded
Meta-analysis(2) Studies were stratified, prior to data
analysis, into groups of related interventions
Studies were divided according to the level of country development and then analysed by intervention type
Meta-analysis (3)Developing Countries
Multiple(i.e. water, sanitation and
hygiene [or health] education)
SanitationHygiene
Handwashing Education
Source Pt-of-use
Water quality
Community improvements
Household connection
Water supply
Meta-analysis(4) Where sufficient studies were available within
each intervention they were further examined in sub-group analysis defined by: Health outcome Age groups Pre-intervention water and sanitation situation
Pre-intervention scenarios F – basic water and basic sanitation Eb – improved water and basic sanitation Ea – basic water and improved sanitation D – improved water and improved san
Where sufficient studies were available within each intervention they were further examined in sub-group analysis defined by: Health outcome Age groups Pre-intervention water and sanitation situation Design Location Study quality
Meta-analysis(4)
Meta-analysis(5) Forest plots and pooled estimates of risk
were generated Both fixed and random effects estimates
were prepared for all analyses If the heterogeneity is less than 0.2 - a
random effects model was used
Forest plot
Effect.01 .1 1 10
Combined
Ref 4
Ref 3
Ref 2
Ref 1
Random 0.757 (0.425 – 1.349)Fixed 0.582 (0.530 – 0.638)Heterogeneity p = 0.000
ResultsKey word search Initial number of references
Diarrhoea AND sanitation 636
Diarrhoea AND water quality 128
Diarrhoea AND water quantity
26
Diarrhoea AND hygiene 423
Drinking water AND intervention
111
Sanitation AND intervention 263
Hygiene AND intervention 459
Results
1
11
9
3 32 2
1 10
5
14
4 4
0
2
4
6
8
10
12
14
16
Afr D Afr E Amr A Amr B AmrD Emr B Emr D Eur A Eur B Eur C Sear B Sear D Wpr A Wpr B
Results
5
12
6
13
4
9
15
7
0
2
4
6
8
10
12
14
16
EME -Hygiene
EME -Sanitation
EME -Watersupply
EME -Waterquality
Dev -Hygiene
Dev -Sanitation
Dev -Watersupply
Dev -Waterquality
Dev -Multiple
ResultsRef Intervention Country Location Health
outcomeAge group
Result 95% CI
Black et al., 1981
Handwashing with soap
USA Suburban (child care centres)
Diarrhoea 0 – 36 months
0.52 0.36-0.76
Bartlett et al., 1988
Hygiene education
USA Urban (child care centres)
Diarrhoea 0 – 35 months
1.09
Kotch et al., 1994
Handwashing + hygiene education
USA Urban (child care centres)
Diarrhoea 0 – 36 months
0.84 0.50-2.08
Carabin et al., 1999
Hygiene education
Canada Unstated (child care centres)
Diarrhoea 18 – 36 months
0.77 0.51-1.18
Roberts et al., 2000
Handwashing Aus Urban (child care centres)
Diarrhoea 0 – 36 months
0.5 0.36-0.68
ResultsRef Adeq.
controlMeasure of confounders
Random. Health indicator definition
Health indicator recall
Analysis by age
Intervention /compliance assessed
Blinding Placebo
1 Yes Yes Yes Non standard
Daily Yes Yes No No
2 Yes Not clear Yes Non standard
Daily or twice weekly
NA No Some Some
3 No Yes No Non standard
2 weeks Yes Yes No No
4 Yes Yes Yes None Daily NA Yes Not clear No
5 Yes Yes Yes Standard 3 weeks Yes Yes Some No
Hypothetical example
Results All the data are outlined in the report Following is a summary of the
intervention studies reported from developing countries on an intervention-by intervention basis
Hygiene
Hygiene (1) 15 papers 13 studies 11 included in the meta-analysis
Hygiene (2)
Effect.01 .1 1 10
Combined
Shahid et al., 1996
Pinfold and Horan, 1996
Haggerty et al., 1994a/b
Wilson et al., 1991
Lee et al., 1991
Han + Hlaing 1989
Alam et al., 1989
Stanton et al., 1988/ Stanton + Clemens 1987
Sircar et al., 1987
Torun, 1982
Khan, 1982Random - 0.63 (0.52 – 0.76)
Fixed - 0.75 (0.72 – 0.78)
Heterogeneity - p = 0.000
Hygiene (3) Overall summary measure
0.633 (0.524 – 0.765)
Removing poor quality studies 0.547 (0.400 – 0.749)
Hygiene (4) Handwashing seemed to be more effective than
hygiene education There seemed to be a greater impact on
diarrhoea than dysentery (but only 2 dysentery data points)
Intervention was effective whatever the baseline scenario, but more so where there was poorer water and/or sanitation facilities
Hygiene (summary)
37
45
37
42
28
44
44
26
0 10 20 30 40 50
Overall (11)
Excluding poor studies (8)
Scenario D (6)
Scenario E + F (4)
Education (6)
Handwashing (5)
Hw + diarrhoea (5)
Hw + dysentery (2)
% reduction in diarrhoea
Sanitation
Sanitation (1) 4 studies 2 included in the meta-analysis (1 of
which examined cholera) Pooled estimate 0.678 (0.529 – 0.868) Adding an additional study (1957 – USA)
– pooled estimate 0.642 (0.514 – 0.802) 1/5 not considered to be poor quality
Water supply
Water supply (1) These included the provision of new or
improved water supply and/or improved distribution
Complex – could include public OR private water supply
Water supply (2) 9 studies, 6 could be included in meta-
analysis Initial results suggested a significant
impact – 0.749 (0.618 – 0.907) BUT that included an ecological study and one examining cholera
Water supply (3)
Effect.01 .1 1 10
Combined
Tonglet et al., 1992
Wang et al., 1989
Esrey et al., 1988
Ryder et al., 1985
Bahl, 1976
Azurin and Alvero, 1974 Random – 0.75 (0.62 – 0.91)Fixed – 0.63 (0.63 – 0.64)Heterogeneity - p < 0.2
Water supply (4) Excluding the ecological study:
Pooled RR 0.869 (0.632 – 1.195) Excluding the ecological study and
restricting analysis to ‘standard’ diarrhoea Pooled RR 1.031 (0.730 -1.457)
Water supply (5) Standpipe versus household on diarrhoea
- suggests a small but not stat significant effect BUT….
Only two studies considered to be of good quality – one of each: HH 0.62 (0.59 – 0.65) Standpipe 0.95 (0.88 – 1.00)
Water supply (6) In one of the hh connection studies,
household storage was still practiced – omitting this study and adding two from developed countries (1976 UK; 1969 USA) – suggests that a household supply can be an effective intervention for reducing diarrhoea 0.557 (0.464 – 0.669)
Water supply (summary)
0
6
10
25
44
0 10 20 30 40 50
Water supply (diarrhoea)
Standpipe (diarrhoea)
HH connection (diarrhoea)
Water supply (overall)
'True' HH connection (diarrhoea)
Water quality
Water quality (1) 15 studies All had data suitable for inclusion in the
meta-analyses 5 papers judged to be poor quality
Water quality (2) Overall intervention effective – pooled
estimate 0.687 (0.534 – 0.885): 31% reduction
This included both source and household treatment
Water quality (3) Source treatment (3) Source only – a reduction in diarrhoea
seen but not stat significant. Some problems with the studies.
Water quality (4) Household treatment/safe storage (12) Household treatment effective 0.645
(0.475 – 0.875): 35% reduction
Impact increased if poor quality studies are removed from the analysis: 39% reduction
Water quality (5) Examining the effect of study location on
the intervention, showed that there seemed to be a greater impact seen on diarrhoea in people from rural communities: 47% reduction
compared to urban/periurban settings 23% reduction
Water quality (summary)
31
11
35
39
47
23
0 10 20 30 40 50
Overall
Source treatment
HH treatment
HH (excluding poor studies)
HH (rural)
HH (urban/periurban)
% reduction in diarrhoea
Multiple
Multiple interventions (1) Nine papers 7 studies, 6 of which had risk estimates
and 5 of which were used in the meta-analysis
5/6 risk measures less than 1
Multiple interventions (2)
Effect.01 .1 1 10
Combined
Nanan et al., 2003
Messou et al., 1997
Hoque et al., 1996
Mertens et al., 1990a,b
Aziz et al., 1990Random - 0.69 (0.64 – 0.77)
Fixed - 0.72 (0.68 – 0.76)
Heterogeneity - p < 0.2
Overall summary
pooled effect
1
HH treatment - rural settings
HH treatment only
Source treatment only
Water quality
Standpipe and diarrhoea
HH connection and diarrhoea Diarrhoea only
Water supply
Sanitation
Education
Handwashing Excluding poor quality studies
Hygiene
Multiple
HH treatment - excl poor quality studies
HH treatment - urban + periurban settings
Discussion (hygiene) Most conducted where water and
sanitation already improved Seem to be effective whatever the starting
conditions Actual interventions vary widely Diarrhoeal reductions improved when
poor papers excluded
Discussion (sanitation) Few studies looked at actual sanitation
interventions Most (75%) were classified as poor
quality Meta-analysis does suggest that the
intervention is effective Scope for much more work here – dry
sanitation study?
Discussion (water supply) Public and private supplies Compliance generally poorly assessed,
with few data on water usage Suggestion that household connection is
effective in reducing diarrhoea levels, especially bringing in 2 studies conducted in developed countries
Discussion (water quality) Source treatment and household
treatment Household treatment particularly
effective (especially when poor quality papers removed from analysis)
Range of household treatment types Source treatment studies hampered by
methodological problems
Discussion (multiple) Complex! All provided water supply,
sanitation and hygiene measures – but final provision varied
None reported final water quality (after storage) and none employed household treatment
Lack of additive effect, when compared to single interventions disappointing
Discussion (study quality) Studies classified as poor quality if:
Lack of adequate control group; No measurement of confounding factors; Undefined health indicator; and/or Health indicator recall of >2 weeks
32% of studies (19 from 60) classed as poor! Results generally improved if these were removed
Discussion (baseline scenario) Reasonable to expect diarrhoea reduction to be
dependent upon starting conditions: F – basic water, basic sanitation Eb – improved water, basic sanitation Ea – basic water, improved sanitation D – improved water, improved sanitation
Not surprisingly, most studies were conducted in areas classified as F – so not possible to examine except for hygiene
Discussion (pre-intervention) Most studies do not ascertain (or report)
pre-intervention diarrhoea level or water, sanitation and hygiene behaviour
Discussion (hh storage) Household storage of water prior to
consumption is common In many intervention studies (except hh
treatment ones), this is often not considered
Contamination of stored water is extremely common
Comparison with Esrey
3336
1915
30
42
23
39
33
24*
05
1015202530354045
Hygiene Sanitation Watersupply
Waterquality
Multiple
Intervention
% r
educ
tion
in d
iarr
hoea
Esrey
Current
Conclusions Some 15 years on from Esrey et al. and
over 20 years from Blum and Feachem diarrhoea is still killing people in developing countries
Loosing data Poor community involvement
Conclusions There is a lot more that we could look at:
Water usage Sustainability of the interventions Sustainability of the health effects Different ways of encouraging intervention
uptake Other health outcomes
Conclusions In study terms in has to be said that
often:
WE COULD DO BETTER!
Conclusions If we do it right we can save lives –
we can make a difference
BUT……….
Thanks to Wayne Enanoria and Jack Colford Rachel Kaufmann Jamie Bartram and Dave Kay NAS, CDC, WELL, WASH, World Bank,
Water Aid, WHO, UNICEF