Upload
endah-utami-tri-k
View
2
Download
0
Embed Size (px)
DESCRIPTION
jurnal HSM masyarakat tahun 2010
Citation preview
Water, sanitation and hygiene for theprevention of diarrhoeaSandy Cairncross,1� Caroline Hunt,1 Sophie Boisson,1 Kristof Bostoen,1 Val Curtis,1 Isaac CH Fung2
and Wolf-Peter Schmidt1
1London School of Hygiene & Tropical Medicine, Department of Infectious & Tropical Diseases, London, UK and 2Department ofEpidemiology and Biostatistics, College of Public Health, University of Georgia, Athens, GA, USA.
�Corresponding author. London School of Hygiene & Tropical Medicine, Keppel Street, London WC1E 7HT, UK.E-mail: [email protected]
Background Ever since John Snow’s intervention on the Broad St pump, theeffect of water quality, hygiene and sanitation in preventingdiarrhoea deaths has always been debated. The evidence identifiedin previous reviews is of variable quality, and mostly relates tomorbidity rather than mortality.
Methods We drew on three systematic reviews, two of them for theCochrane Collaboration, focussed on the effect of handwashingwith soap on diarrhoea, of water quality improvement and ofexcreta disposal, respectively. The estimated effect on diarrhoeamortality was determined by applying the rules adopted for thissupplement, where appropriate.
Results The striking effect of handwashing with soap is consistentacross various study designs and pathogens, though it depends onaccess to water. The effect of water treatment appears similarlylarge, but is not found in few blinded studies, suggesting that itmay be partly due to the placebo effect. There is very little rigorousevidence for the health benefit of sanitation; four interventionstudies were eventually identified, though they were allquasi-randomized, had morbidity as the outcome, and were inChinese.
Conclusion We propose diarrhoea risk reductions of 48, 17 and 36%,associated respectively, with handwashing with soap, improvedwater quality and excreta disposal as the estimates of effect for theLiST model. Most of the evidence is of poor quality. More trials areneeded, but the evidence is nonetheless strong enough to supportthe provision of water supply, sanitation and hygiene for all.
Keywords Water, sanitation, hygiene, diarrhoea, mortality
BackgroundIt has been estimated, at least for Africa, that 85% ofthe burden of disease preventable by water supply iscaused by feco-oral, mainly diarrhoeal diseases,largely due to the substantial child mortality which
they cause.1 In 1854, Dr John Snow famouslyincriminated the water from the Broad St pump asthe vehicle of cholera transmission in London’sSoho, but much of the medical establishmentcontinued to uphold the miasma theory for manyyears thereafter. Ever since then, the role of
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/
by-nc/2.5/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Published by Oxford University Press on behalf of the International Epidemiological Association.
� The Author 2010; all rights reserved.
International Journal of Epidemiology 2010;39:i193–i205
doi:10.1093/ije/dyq035
i193
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
water in diarrhoea transmission and preventionhas been hotly debated. More recently, awarenesshas also grown about the importance of excretadisposal in preventing diarrhoeal disease, culminatingin the recent poll of readers of the BritishMedical Journal in which sanitation was voted thegreatest advance in public health in the last century.2
There is strong temptation to conduct evaluationsof the health impact of water supply, sanitation andhygiene interventions, but the challenges also aremany. Often it is difficult or impossible to randomizeor to blind the intervention. In practice, most studiesdo not identify specific aetiologies and so deal with anoutcome (diarrhoea) which is caused by variouspathogens, transmitted by various routes and asso-ciated with various potential confounding factors.The vulnerability of such studies to confounding iscompounded by the use of observational studydesigns and the low relative risks (RRs) involved,which are typically less than two. In developing coun-tries, most episodes of diarrhoea morbidity—evenmuch of the life-threatening morbidity—are notreported to the health system, so that active surveil-lance involving home visits is used to detect them,often with excessive recall periods. In the circum-stances then, it is not surprising that the first meth-odological review of this literature3 located some 50epidemiological studies, but found serious flaws inevery one.
A series of literature reviews conducted by Esreyand others4–7 established a consensus view on theimpacts on health of improved water quality, waterquantity and sanitation, which was summarized inthe relevant chapter of Disease Control in DevelopingCountries.8 A more recent review9 gave prominence toa number of studies of household-based watertreatment, and arrived at a greater estimate ofthe impact of water quality than previous reviews.However, the confidence intervals (CIs) for thisand the other such estimates were very wide, sowide as to show that the new figures were notsignificantly different from the corresponding previousestimates.
In this article, drawing on three systematicreviews of the literature, we present the evidence foran impact on diarrhoea mortality from improvementsin hygiene (specifically, handwashing with soap),10
drinking-water quality,11 and excreta disposal. Thereviews of effectiveness of the interventions areshaped in large part by the needs of the LiST model.In that model, increases in coverage of an interven-tion result in a reduction of deaths due to one ormore causes or in the reduction of a risk factor.Therefore, the reviews and the grade process usedwere designed to develop estimates of the effectof an intervention in reducing either a risk factor ordeaths due to a specific cause. For more details ofthe review methods, the adapted grade approach orthe LiST model, see other articles in this supplement.
MethodsHandwashing with soapThe original review10 aimed to identify all studiespublished in English up to the end of 2002 relatinghandwashing to the risk of infectious intestinal ordiarrhoeal diseases in the community. Medline, CABAbstracts, Embase, Web of Science and the CochraneLibrary were systematically searched for papersrelated to handwashing, use of soap, as well as dis-ease terms such as diarrhoea, typhoid, enteric, chol-era, shigellosis, dysentery, and mortality. Searcheswere also undertaken by hand with reference listsfrom these papers, the authors’ own collections andreview articles. No limitations were placed on date orgeographical location. The search was updated in2008.
Studies were retained for the meta-analysis if theyprovided point estimates and 95% CIs (or the meansto calculate them) of the risk of not washing hands.Intervention trials not solely concerned with hand-washing were excluded. Where both crude andadjusted odds ratios were presented, adjusted valueswere used. The risk values for studies with severalmeasures of handwashing practice were combinedby averaging, if they concerned the same samplegroup. If they concerned different groups, they weretreated as if they were separate studies. Similarly,studies with two different outcome measures wereentered into the meta-analysis as if they were sepa-rate studies.
Water qualityFollowing the Cochrane peer-reviewed protocol,12 wesearched the specialized register of the CochraneInfectious Diseases Group, CENTRAL, Medline,Embase and LILACS for all randomized andquasi-randomized controlled trials of interventions toimprove water quality for the prevention of diarrhoealdisease, regardless of language, publication status, ordate, up to December 2005. Interventions includedany measure to improve the microbial quality ofdrinking water, unless undertaken in response to epi-demic diarrhoea. The primary outcome was diarrhoeain adults or children. We hand searched conferenceproceedings, contacted researchers and organizationsworking in the specialty, and checked the referencesof identified studies. Two reviewers independentlyexamined the electronic records for potentially eligiblestudies and the full text of potentially eligible reports.Disagreements were resolved by a third reviewer.
Measures of effect reported were risk ratios, rateratios, odds ratios and longitudinal prevalence ratios(number of days or weeks with diarrhoea divided bynumber of days or weeks under observation in aperson). The results are presented separately bystudy type and also by type of intervention, whethersource- or household-based. A random effects inverse
i194 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
variance method was used on the log scale to calcu-late pooled estimates.
Excreta disposalAgain we followed the Cochrane approved protocol13
to search for interventions to dispose of humanexcreta so as to reduce direct or indirect human con-tact. This includes any steps to remove or containfaeces, such as simple pit latrines, bucket latrines,hanging toilets and composting toilets, and shouldbe contrasted with open defecation. Diarrhoea wasagain the outcome measure, whether or not microbio-logically confirmed. We defined diarrhoea and anepisode in accordance with the case definitions usedin each trial. We excluded trials that had no clinicaloutcomes; for example, trials that only reported onmicrobiological pathogens in the stool.
We searched the following databases up to April2007: Cochrane Infectious Disease Group SpecializedRegister; CENTRAL; MEDLINE; EMBASE; LILACS,and also Chinese-language databases available underthe Wan Fang portal using the Chinese equivalents ofour search terms where appropriate. We searched themetaRegister of Controlled Trials (mRCT) using ‘diar-rhea’, ‘diarrhoea’ and ‘sanitation or latrine or toilet orprivy or disposal’ as search terms, and also a number ofrelevant conference proceedings. Other researchers andrelevant international agencies were also contactedin the search for unpublished and ongoing trials. Thereference lists of studies identified as above were alsoscanned for any further relevant studies.
GeneralAll studies which met final inclusion and exclusioncriteria were double data abstracted into a standar-dized form for each outcome of interest.14 Weabstracted key variables with regard to the studyidentifiers and context, study design and limitations,intervention specifics, and outcome effects. Eachstudy was assessed and graded according to the adap-tation by the Child Health Epidemiology ReferenceGroup (CHERG) of the GRADE technique.15 Studiesreceived an initial score of high if a randomized orcluster randomized trial. The grade was decreased onegrade for each study design limitation. In addition,studies reporting an intent-to-treat analysis or withstatistically significant strong levels of association(480% reduction) receive 1–2 grade increases. Anystudy with a final grade of very low was excludedon the basis of inadequate study quality. The processis summarized in Tables 1, 2 and 3. To save space, wehave not listed all of the studies in all three reviewswith the bibliography of this article; instead, adetailed listing is provided in three correspondingSupplementary Tables 1, 2 and 3 available at IJEonline. The numbers in Tables 1, 2 and 3 refer tothese supplementary tables.
For the outcome of interest, namely the reductionof diarrhoea mortality, we applied the CHERG Rules
for Evidence Review14 to the collective diarrhoea mor-bidity and mortality outcomes to generate a finalestimate of effect.
ResultsHandwashing with soapThe search identified a total of 38 studies (Figure 1) but21 were not suitable for data extraction, either becausethey did not specify whether soap was used, or did notpresent data permitting a calculation of effect. Of the17 remaining studies, 7 were intervention trials and10 were observational (Table 1). All of the interven-tions or exposure measures related to handwashingbefore eating or food handling, or after defecation orhandling of child stools, or a combination of these.
Only one mortality study was found.16 The numberof events was not stated, but the wide confidenceintervals (reduction of þ62% to �43%) suggest therewere very few, and the study was observational, withweaknesses in the outcome definition (deaths fromother infectious diseases besides diarrhoea wereincluded) and ascertainment of compliance.
We therefore turned to morbidity studies. None ofthe intervention studies gave adequate compliancedata, so that the effect reported (a reduction of47%) is that of handwashing promotion rather thanof handwashing itself; i.e. the effect of the interven-tion, not the individual’s response to it. However, thiswas slightly greater than the pooled effect (a reduc-tion of 43%) of all studies in the review (Figure 2aand b). In other words, the observational studies,which did report the effect of individuals’ behaviour,did not find a greater effect. The original reviewfound that the effect was remarkably consistentacross studies of higher methodological quality, andstudies with severe forms of diarrhoea as an outcome.In each of these categories, the pooled estimate ofeffect is of a reduction within the range of 42–48%(Table 1). The combined effect of the more severediarrhoeas was a 48% reduction. Two studies oflaboratory confirmed shigellosis,17,18 a more severeand a ‘harder’ outcome than self-reported diarrhoea,had a pooled effect of reduction by 59%. On the otherhand, three studies were conducted in settings wherewater use was known to be constrained; (i) alow-income area of Lima, Peru, where vendors sellwater expensively from tanker trucks, (ii) a refugeecamp in Malawi and (iii) a setting in Burundi wheremedian water usage was only 5 l per capita per day.The reductions in risk were 11, 26 and 41%, respec-tively, all of them less than the combined effect of43% found in the review as a whole.
Subsequent to our initial review,10 a Cochranereview of the effect of handwashing on diarrhoeawas published.19 Most of the studies included werein institutional settings, but five were conducted inthe community. In two of those, the intervention
PREVENTION OF DIARRHOEA i195
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
Ta
ble
1Q
uali
tyass
essm
ent
of
tria
lso
fh
an
dw
ash
ing
wit
hso
ap
for
the
pre
ven
tio
no
fd
iarr
ho
ea
No
of
stu
die
s(s
up
ple
me
nta
ryta
ble
ref.
)
Qu
ali
tya
sse
ssm
en
tN
oo
fe
ve
nts
Eff
ect
Des
ign
Lim
itati
on
sC
on
sist
ency
(base
don
the
het
erogen
eity
of
the
met
a)
Gen
erali
zab
ilit
yto
pop
ula
tion
of
inte
rest
Gen
erali
zab
i-li
tyof
inte
rven
tion
Inte
rven
tion
Con
trol
RR
(95%
CI)
Ou
tcom
e1:
Dia
rrh
oea
mort
ali
ty;
Qu
ali
ty:
very
low
115
Ob
serv
ati
on
al
(case
con
trol)
Not
ran
dom
ized
,n
op
lace
bo,
un
reli
ab
leex
posu
rem
easu
re(�
2)
Not
ap
pli
cab
leO
nly
on
est
ud
y(�
0.5
)N
ot
giv
enN
ot
giv
en0.9
7(0
.38–1.4
3)a
Ou
tcom
e2:
Dia
rrh
oea
morb
idit
y;Q
uali
ty:
very
low
71–7
RC
T/q
uasi
-RC
TIn
ad
equ
ate
ran
dom
izati
on
,n
op
lace
bo,
com
pli
an
cen
ot
mea
sure
d,
no
base
lin
ein
cid
ence
(�1)
Tes
tsfo
rh
eter
ogen
eity
gav
esi
gn
ific
an
tre
sult
(�0.5
)
1st
ud
yin
US
A,
1A
ust
rali
a;
3st
ud
ies
are
all
ages
(�0.5
)
Mod
erate
an
dse
vere
morb
idit
y(�
0.5
)
739
b1157
b0.5
3(0
.37–0.7
6)c
31,2
,10
Hig
hm
eth
od
o-
logic
al
qu
ali
ty(t
rials
&ob
serv
ati
on
al
stu
die
s)d
1of
3ra
nd
om
ized
,n
op
lace
bo,
1h
igh
loss
tofo
llow
-up
,1
com
pli
ance
not
ass
esse
d(�
0.5
)
Het
erogen
eity
not
sign
ific
an
t.R
and
om
effe
cts
mod
esu
sed
for
poole
des
tim
ate
s.B
oth
fix
edan
dra
nd
om
effe
ctm
od
els
gav
eth
esa
me
poole
des
tim
ate
1st
ud
yin
US
A,
1A
sia,
1L
ati
nA
m,
1st
ud
yis
all
ages
(�0.5
)
Mod
erate
an
dse
vere
morb
idit
y(�
0.5
)
55
b109
b0.5
8(0
.49–0.6
9)c
Ou
tcom
e3:
Sev
ere
ou
tcom
es(h
osp
itali
zed
ente
ric
infe
ctio
nca
ses,
chole
ra,
shig
ello
sis,
typ
hoid
an
dd
eath
s);
Qu
ali
ty:
very
low
92,4
,6,1
1,1
2,1
4,1
5,1
6,1
8R
CT
/qu
asi
-RC
T/
ob
serv
ati
on
al
Inad
equ
ate
ran
dom
izati
on
,n
op
lace
bo,
com
pli
an
cen
ot
ass
esse
d,
no
base
lin
ein
cid
ence
(�1)
Tes
tsfo
rh
eter
ogen
eity
gav
esi
gn
ific
an
tre
sult
(�0.5
)
Most
lyA
sia;
most
stu
die
sare
all
ages
(�0.5
)
Sev
ere
morb
idit
yan
dca
use
-sp
ecif
icm
ort
ali
ty(�
0.5
)
492
b798
b0.5
2(0
.34–0.6
5)c
Ou
tcom
e4:
Sh
igel
losi
s;Q
uali
ty:
low
22,6
RC
T/q
uasi
-RC
TIn
ad
equ
ate
ran
dom
izati
on
,n
op
lace
bo,
com
pli
an
cen
ot
mea
sure
d,
no
base
lin
ein
cid
ence
(�1)
Het
erogen
eity
not
sign
ific
an
t.R
and
om
effe
cts
mod
elu
sed
for
poole
des
tim
ate
s(F
ixed
effe
ctm
od
elgave
the
sam
ep
oole
des
tim
ate
)
Both
stu
die
sin
Asi
a,
both
all
ages
(�0.5
)
Sev
ere
morb
idit
y17
51
0.4
1(0
.27–0.6
2)c
So
urc
eo
fsy
stem
ati
cre
view
an
dm
easu
res
of
effe
ct:
Ref
.10
aE
ffec
tco
mes
fro
mu
nad
just
edo
dd
sra
tio
of
0.9
7re
po
rted
inH
oq
ue
etal
.16
bN
um
ber
of
even
tsn
ot
ava
ilab
lefo
rall
stu
die
s.c R
an
do
mef
fect
met
a-a
naly
sis.
dIn
clu
des
tria
lsw
ith
base
lin
esan
dco
ncu
rren
tco
ntr
ol
gro
up
san
do
bse
rvati
on
al
stu
die
sw
ith
ad
equ
ate
con
tro
lfo
rco
nfo
un
din
g.
i196 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
Ta
ble
2Q
uali
tyass
essm
ent
of
tria
lsof
wate
rq
uali
tyim
pro
vem
ent
for
the
pre
ven
tion
of
dia
rrh
oea
No
of
stu
die
s(s
up
ple
me
n-
tary
tab
lere
f.)
Qu
ali
tya
sse
ssm
en
tN
oo
fe
ve
nts
Eff
ect
Des
ign
Lim
itati
on
sC
on
sist
ency
(bas
edon
the
het
erogen
eity
of
the
met
a)
Gen
erali
zab
ilit
yto
pop
ula
tion
of
inte
rest
Gen
erali
zab
ilit
yof
inte
rven
tion
Inte
rven
tion
Con
trol
RR
(95%
CI)
Ou
tcom
e1:
Dia
rrh
oea
lm
ort
ali
tyfo
rall
ages
;Q
uali
ty:
very
low
28,2
7R
CT
an
dq
uasi
-RC
TN
op
lace
bo,
not
bli
nd
ed,
on
era
nd
om
ized
,se
lf-r
eport
edan
dse
lf-d
efin
edd
iarr
hoea
(�1.5
)
NA
Both
Afr
ica,
1is
all
ages
,1
un
der
5s
(�0.5
)
1all
-cau
sem
ort
al-
ity,
1d
iarr
hoea
lm
ort
ali
ty
34
55
Ran
gin
gfr
om
0.6
1to
0.1
5a
Ou
tcom
e2:
Dia
rrh
oea
lm
orb
idit
yfo
rall
ages
;Q
uali
ty:
very
low
38
(30
stu
die
s,5
wit
hm
ult
iple
arm
s;arm
sco
un
ted
as
ad
dit
ion
al
tria
ls)1
–30
RC
Tan
dq
uasi
-RC
TO
nly
thre
eb
lin
ded
an
dp
lace
bo,
few
ad
equ
atel
yra
nd
om
ized
,m
ost
self
-rep
ort
edd
iarr
hoea
wit
hex
cess
ive
reca
llp
erio
d.
(�1.5
)
Tes
tfo
rh
eter
o-
gen
eity
gav
esi
gn
ific
an
tre
sult
(�0.5
)
Most
lyA
fric
a,
Lati
nA
mer
ica
an
dA
sia,
all
ages
(�0.5
)
Mod
erate
an
dse
vere
morb
idit
y(�
0.5
)
41
485
b64
224
b0.5
8(0
.46–0.7
2)c
27
(19
stu
die
s,5
wit
hm
ult
iple
arm
s)1–19
RC
TO
nly
thre
eb
lin
ded
an
dw
ith
pla
-ce
bo,
few
ad
equ
atel
yra
nd
o-
miz
ed(�
1)
Sig
nif
ican
th
eter
ogen
eity
(�0.5
)
Most
lyA
fric
aan
dL
ati
nA
mer
ica,
all
ages
(�0.5
)
Mod
erate
an
dse
vere
morb
idit
y(�
0.5
)
23
151
45
098
0.5
4(0
.38–0.8
8)c
41,5
,12
Dou
ble
bli
nd
stu
die
sIn
ad
equ
ate
ran
dom
izati
on
,h
igh
loss
tofo
llow
-up
,co
vera
ge
not
mea
sure
d(�
0.5
)
Het
erogen
eity
not
sign
ific
an
t(P¼
0.3
0)
1L
at
Am
,1
Afr
ica,
1U
SA
,all
ages
(�0.5
)
Mod
erate
an
dse
vere
morb
idit
y(�
0.5
)
1912
1987
0.9
3(0
.70–1.3
3)c
11
20–30
Qu
asi
-RC
TIn
ad
equ
ate
lyra
nd
om
ized
,n
op
lace
bo,
most
lyse
lf-r
eport
edw
ith
exce
ssiv
ere
call
per
iod
(�2)
Sig
nif
ican
th
eter
ogen
eity
(�0.5
)
Most
lyA
sia
an
dA
fric
a,
all
ages
(�0.5
)
Mod
erate
an
dse
vere
morb
idit
y(�
0.5
)
18
334
b19
125
b0.6
2(0
.48–0.8
9)c
620,2
1,2
3,2
4,2
7,3
0Q
uasi
-RC
Tso
urc
e-b
ase
din
terv
enti
on
s
Inad
equ
ate
lyra
nd
om
ized
,n
op
lace
bo,
most
lyse
lf-r
eport
ed,
exce
ssiv
ere
call
per
iod
(�2)
Sig
nif
ican
th
eter
ogen
eity
(�0.5
)
4A
sia,
2A
fric
a,
all
ages
(�0.5
)M
od
erate
an
dse
vere
morb
idit
y(�
0.5
)
18
164
18
723
0.7
3(0
.53–1.0
1)c
32
1–19,2
2,2
5,2
6,2
8,2
9R
CT
an
dq
uasi
-RC
Th
ou
se-h
old
-base
din
terv
enti
on
s
On
lyth
ree
bli
nd
edan
dp
lace
bo,
few
ad
equ
atel
yra
nd
om
ized
,m
ost
self
-rep
ort
edd
iarr
hoea
wit
hex
cess
ive
reca
llp
erio
d.
(�1.5
)
Sig
nif
ican
th
eter
ogen
eity
(�0.5
)
Most
lyA
fric
aan
dL
ati
nA
mer
ica,
all
ages
(�0.5
)
Mod
erate
an
dse
vere
morb
idit
y(�
0.5
)
23
321
b45
500
b0.5
7(0
.46–0.7
0)c
Ou
tcom
e3:
Dia
rrh
oea
lm
orb
idit
yfo
rch
ild
ren
<5
years
;Q
uali
ty:
very
low
29
1–4,8
–13,1
5–20,
22–24,2
6,2
7R
CT
san
dq
uasi
-RC
Ts
On
lyth
ree
bli
nd
edan
dp
lace
bo,
few
ad
equ
atel
yra
nd
om
ized
,m
ost
self
-rep
ort
edd
iarr
hoea
wit
hex
cess
ive
reca
llp
erio
d(�
1.5
)
Sig
nif
ican
th
eter
ogen
eity
(�0.5
)
Most
lyL
ati
nA
mer
ica
an
dA
fric
a,
un
der
5s
Mod
erate
an
dse
vere
morb
idit
y(�
0.5
)
17
687
b39
165
b0.6
0(0
.44–0.8
1)c
420,2
3,2
4,2
7Q
uasi
-RC
Tso
urc
e-b
ase
din
terv
enti
on
s
Inad
equ
ate
lyra
nd
om
ized
,n
op
lace
bo,
most
self
-rep
ort
ed,
exce
ssiv
ere
call
per
iod
(�2)
Tes
tfo
rh
eter
o-
gen
eity
gav
eP¼
0.0
8
2A
fric
a,
2A
sia,
un
der
5s
Mod
erate
an
dse
vere
morb
idit
y(�
0.5
)
1603
1924
0.8
5(0
.71–1.0
2)c
25
1–4,8
,11–13,
15–19,2
6,2
7R
CT
an
dq
uasi
-RC
Th
ou
seh
old
-base
din
terv
enti
on
s
On
lyth
ree
bli
nd
edan
dp
lace
bo,
few
ad
equ
atel
yra
nd
om
ized
,m
ost
self
-rep
ort
edd
iarr
hoea
wit
hex
cess
ive
reca
llp
erio
d.
(�1.5
)
Sig
nif
ican
th
eter
ogen
eity
(�0.5
)
Most
lyL
ati
nA
mer
ica
an
dA
fric
a,
un
der
5s
Mod
erate
an
dse
vere
morb
idit
y(�
0.5
)
16
084
b37
241
b0.5
6(0
.39–0.8
1)c
Sou
rce:
Cla
sen
T,
20
06
,P
hD
Dis
sert
ati
on
,U
niv
ersi
tyo
fL
on
do
n.1
1
aW
ed
idn
ot
calc
ula
tep
oo
led
esti
mate
sfo
rth
est
ud
ies
on
wate
rq
uali
tyre
po
rtin
gm
ort
ali
ty.
bN
um
ber
of
even
tsn
ot
ava
ilab
lefo
ro
ne
or
mo
rest
ud
ies.
c Ran
do
mef
fect
sin
vers
eva
rian
cem
eth
od
on
the
log
scale
was
use
dto
calc
ula
tep
oo
led
esti
mate
s.H
eter
ogen
eity
was
exam
ined
usi
ng
the�
2te
stw
ith
a1
0%
leve
lo
fst
ati
stic
al
sign
ific
an
cean
dth
eI2
test
for
con
sist
ency
.
PREVENTION OF DIARRHOEA i197
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
did not involve soap and was not focussed on hand-washing. Pooling the other three20–22 gave a reductionin diarrhoea by 43% (95% CI 25–56%).
Water quality interventionsTwo studies with a mortality outcome were found andincluded, following the Cochrane protocol12
(see Figure 3 and Table 2). One of these23 had ascer-tained diarrhoea mortality, but the interventionincluded health education about oral rehydrationtherapy for diarrhoea, which alone could explain the85% reduction in diarrhoea mortality that wasobserved. Moreover, only two villages had been ran-domized. The other trial,24 whose authors admit thatit was not designed to detect a mortality outcome,found a significant reduction (RR of death¼ 0.58,P < 0.036) only for all causes and all ages by poolingtwo intervention arms (total 59 events). Pooling thestudies would be questionable, as the interventionswere very different.
The serious limitations of the mortality studies ledus to consider the morbidity studies. All took self-reported diarrhoea as the outcome; the Cochraneprotocol had excluded trials that had no clinical out-comes, such as trials that report only on microbiolo-gical pathogens in the stool. Similar results wereobtained for diarrhoea in all age groups and withthe outcome limited to children aged <5 years. Thetype of study seemed to have little bearing on themeasure of effect, which ranged from 46% in rando-mized controlled trials (RCTs) to 38% of quasi-RCTs.The review protocol approved by the CochraneCollaboration had limited its scope to trials, so thatno observational studies were included. Interventionsbased at the water source had less effect (27% for allages, 15% for young children) than household-basedones, with an effect of 43–44%, depending on the agerange. Two studies implemented in settings wherethe water before treatment had <10 faecal coli-forms/100 ml (a concentration classified by WHO as
17 suitable for data extraction
38 relevant studies
identified21 not suitable (eg soap not mentioned)
7 intervent-ion studies
10 observat-ional studies
Figure 1 Synthesis of study identification in review of theeffect of handwashing with soap on diarrhoea mortality andmorbidity10
Ta
ble
3Q
uali
tyass
essm
ent
of
tria
lso
fex
cret
ad
isp
osa
lfo
rth
ep
reve
nti
on
of
dia
rrh
oea
No
of
stu
die
s(s
up
ple
me
nta
ryta
ble
ref.
)
Qu
ali
tya
sse
ssm
en
tN
oo
fe
ve
nts
Eff
ect
Des
ign
Lim
itati
on
sC
on
sist
ency
(base
do
nth
eh
eter
ogen
eity
of
the
met
a)
Gen
erali
z-ab
ilit
yto
po
pu
lati
on
of
inte
rest
Gen
erali
z-ab
ilit
yo
fin
terv
enti
on
Inte
rven
tio
nC
on
tro
lR
R(9
5%
CI)
Ou
tco
me
1:
Dia
rrh
oea
lm
orb
idit
yfo
rall
ages
;Q
uali
ty:
very
low
41
–4
Qu
asi
-RC
TIn
ad
equ
ate
ran
do
miz
ati
on
,n
op
lace
bo
,n
ob
ase
lin
e,co
mp
lian
cen
ot
ass
esse
d,
lon
gre
call
for
self
-rep
ort
edd
iarr
ho
ea(�
1.5
)
No
tap
pli
cab
leA
llC
hin
a,
all
ages
(�0
.5).
Lim
ited
rele
van
ceo
uts
ide
Ch
ina
Sev
ere
tom
od
erate
mo
rbid
ity
(�0
.5)
24
71
39
25
Ran
gin
gfr
om
0.9
2to
0.3
7;
med
ian
0.4
9to
0.8
0a
Ou
tco
me
2:
Dia
rrh
oea
lm
orb
idit
y,ages
0–3
years
Qu
ali
ty:
low
15
Bef
ore
/aft
erst
ud
yIn
ad
equ
ate
ran
do
miz
ati
on
,an
dn
op
lace
bo
No
tap
pli
cab
leU
rban
,h
igh
cove
rage
of
on
-plo
tw
ate
r
Sew
erage,
no
to
n-s
ite
low
-co
stsy
stem
s
No
tgiv
en;
24
sen
tin
elare
as
stu
die
d
No
tgiv
enC
ity-
wid
ere
du
ctio
no
f2
1%
(19
%,
26
%),
an
d4
3%
(39
%,
46
%)
inth
eh
igh
risk
are
as
Sou
rce:
Cla
sen
TF
,B
ost
oen
K,
Sch
mid
tW
-Pet
al.
Inte
rven
tio
ns
toim
pro
veex
cret
ad
isp
osa
lfo
rp
reve
nti
ng
dia
rrh
oea
.C
och
ran
eD
atab
ase
Syst
Rev
DR
AF
T(A
pri
l2
00
9).
aW
ed
idn
ot
calc
ula
tep
oo
led
esti
mate
sfo
rth
esa
nit
ati
on
tria
ls.
i198 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
(a)
NOTE: Weights are from random effects analysis
Overall
14
20
17
10
7B
13
StudyID
18
22
12
7A6
15A15B
11
9B
16
8
19
21
9A
0.56 (0.45, 0.70)
0.34 (0.26, 0.46)
1.03 (0.61, 1.74)
0.03 (0.00, 0.46)
0.35 (0.09, 1.33)
0.43 (0.23, 0.79)
0.81 (0.41, 1.62)
0.74 (0.55, 0.99)
0.25 (0.08, 0.82)
0.89 (0.79, 1.01)
0.63 (0.35, 1.11)0.52 (0.36, 0.76)
0.38 (0.33, 0.43)0.40 (0.23, 0.69)
0.21 (0.08, 0.54)
0.93 (0.39, 2.21)
0.59 (0.30, 1.14)
0.98 (0.89, 1.08)
ES (95% CI)
0.58 (0.42, 0.80)
0.50 (0.36, 0.69)
0.70 (0.53, 0.92)
.05 .25 .5 1 2Relative risk
NOTE: Weights are from random effects analysis
Overall
15B
7A
8
9B
7B
15A
11
StudyID
9A
21
6
0.53 (0.37, 0.76)
ES (95% CI)
0.40 (0.23, 0.69)
0.63 (0.35, 1.11)
0.98 (0.89, 1.08)
0.93 (0.39, 2.21)
0.43 (0.23, 0.79)
0.38 (0.33, 0.43)
0.21 (0.08, 0.54)
0.70 (0.53, 0.92)
0.50 (0.36, 0.69)
0.52 (0.36, 0.76)
.05 .25 .5 1 2Relative risk
(b)
Figure 2 Forest plots of (a) all studies in handwashing review, and (b) intervention studies only. Numbers on y-axis arereferences to studies in Supplementary Table 1. The diamond represents the combined relative risk and 95% CI fromrandom effects model10
PREVENTION OF DIARRHOEA i199
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
‘low risk’) had a combined effect of reducing risk by61% (95% CI 30–67%) with insignificant heterogene-ity. The greatest difference of all was between the 31unblinded and the 4 blinded trials,25–28 with the lattergiving a reduction of only 7% in diarrhoea risk, whichwas not statistically significant (Figure 4).
Excreta disposalOur initial attempt at this review nearly foundered.The initial search13 produced seven quasi-randomizedintervention studies with diarrhoea morbidity as theoutcome measure, and one with diarrhoea mortality(Figure 5, Table 4). The mortality study has alreadybeen mentioned;23 the intervention included watersupply and education about oral rehydration therapy,which could explain the observed reduction in mor-tality. All eight studies involved an intervention toimprove water supply as well as excreta disposal, sothat in none of them is it possible to assess the impactof sanitation as an intervention per se.
A renewed effort to locate suitable studies produceda further four,29–32 in which the intervention involvedexcreta disposal alone. All four were conducted inChina (see the following URL http://maps.google.com/maps/ms?ie=UTF8&hl=en&msa=0&msid=109120590372693546058.000465c8157b73053ba7f&z=6for a map of the study sites), and published inChinese (Table 3). The reductions in diarrhoea mor-bidity in the four studies were 63, 51, 20 and 8%,respectively. In all but the last, the confidence intervaldid not include zero reduction. We did not calculatepooled estimates for the sanitation trials, because
most studies randomized a very small number of vil-lages, which makes calculation of confidence intervalsimpossible. Also, the interventions tested were verydifferent from one another, which also makes poolingquestionable.
These four studies are hardly ideal; they arequasi-randomized, not full RCTs; the control groupsdid not lack sanitation altogether, but mainly usedsome sort of pit latrine; and third, the published stu-dies are not available for scrutiny except by those ableto read Chinese. We therefore widened our scopefurther to include before/after studies. The unit ofintervention is effectively the community, neighbour-hood or village, but only one such study consideredenough such units for statistical tests to be applied atthis level. This recent study33 is, strictly speaking, abefore/after design, but is in many ways akin to a
.1 .5 1 1.5
Combined
(a)
.1 .5
Relative risk
Relative risk
1 1.5
Combined
(b)
Figure 4 Forest plots of (a) all 35 studies in water qualityreview, and (b) the four blinded studies only35
41 trials from 32 studiesincluded in meta-analysis
908 excluded onbasis of abstract
71 retrieved for evaluation
38 excluded33 met inclusion criteria 4 had 2 trial arms 1 had 3 trial arms 1 had 4 trial arms- yielding 42 included trials
1 excluded frommeta-analysis;inadequateinformation
979 relevant studies identified: 939 by database search 40 by hand
Figure 3 Synthesis of study identification in review of theeffect of water quality interventions on diarrhoea mortalityand morbidity11
i200 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
quasi-randomized trial in 24 neighbourhoods of theCity of Salvador, Brazil. It found a city-wide reductionof 21% (95% CI 19–26%), and 43% (CI 39–46%) in thehigh risk areas.
DiscussionHandwashing with soapOnce the nature of the studies has been consistentlydefined (in the present case, community-based stu-dies with an intervention focussed on handwashingand involving soap), the studies in the literature areremarkably consistent, showing a reduction in diar-rhoea by 42–48%. We took 48%, the reduction foundfor the more severe diarrhoeas, as the figure to pro-pose for LiST.
The problem encountered with water quality inter-ventions, that the blinded studies do not support thepositive picture drawn by the others, raises the ques-tion of whether this problem applies to handwashingwith soap. After all, one cannot persuade people towash their hands without their knowledge! The dif-fering results of the blinded and unblinded studiessuggest that much of the apparent impact of watertreatment is attributable to bias, but by the sametoken that possibility cannot be ignored in the caseof handwashing.
Blinding can be applied to two groups; the subjects,and those who assess their status. It can refer to eachsubject’s exposure allocation, or to the nature of theexposure or the disease outcome. Indeed, theCochrane review of handwashing19 noted that one
7 trialsincluded in CHERGreview
45 excluded onbasis of abstract
20 retrieved for evaluation
8 excluded12 trials met inclusion criteria (outcomes included diarrhoeal disease, helminth infestations and schistosomiasis)
5 excluded as outcomes were other than diarrhoea
65 relevant studies identified
Figure 5 Synthesis of study identification in review of theeffect of excreta disposal on diarrhoea morbidity
Ta
ble
4R
esu
lts
of
init
ial
searc
hfo
rev
iden
ceo
fth
eef
fect
of
san
itati
on
ind
iarr
ho
eaco
ntr
ol
Au
tho
rY
ea
rD
esi
gn
Co
un
try
Se
ttin
gR
an
do
miz
ed
?O
the
rin
terv
en
tio
ns
at
sam
eti
me
Ris
kre
du
ctio
n
Ou
tco
me
1:
Dia
rrh
oea
mo
rtali
tyM
esso
uet
al.
19
97
Qu
asi
-RC
TIv
ory
Co
ast
rura
lU
ncl
ear
Pu
bli
cta
ps,
2h
/wee
kh
ealt
hed
uca
tio
nab
ou
td
iar-
rho
eap
reve
nti
on
&O
RT
85
%
Ou
tco
me
2:
Dia
rrh
oea
mo
rbid
ity
Azi
zet
al.
19
90
Qu
asi
-RC
TB
an
gla
des
hru
ral
No
Imp
rove
dw
ate
rsu
pp
ly,
hyg
ien
eed
uca
tio
n2
5%
Garr
ett
etal
.2
00
8Q
uasi
-RC
TK
enya
rura
lN
oIm
pro
ved
wate
rsu
pp
ly;
ho
use
ho
ldch
lori
nati
on
,h
ygie
ne
edu
cati
on
an
dim
pro
ved
wate
rst
ora
ge
29
%
Hu
ttly
etal
.1
99
0Q
uasi
-RC
TN
iger
iaru
ral
No
Bo
reh
ole
s&
han
dp
um
ps,
hea
lth
&h
ygie
ne
edu
cati
on�
3%
Mes
sou
etal
.1
99
7Q
uasi
-RC
TM
esso
u(b
)m
orb
idit
y
Ivo
ryC
oast
rura
lU
ncl
ear
Pu
bli
cta
ps,
2h
/wee
kh
ealt
hed
uca
tio
nab
ou
td
iarr
ho
eap
reve
nti
on
&O
RT
30
%
Ru
ben
stei
net
al.
19
69
Qu
asi
-RC
TU
SA
rura
lN
oH
ou
seh
old
wate
rco
nn
ecti
on
s6
7%
Xia
oet
al.
19
97
Qu
asi
-RC
TC
hin
aru
ral
No
Hea
lth
edu
cati
on
top
rom
ote
bo
ilin
go
fw
ate
r,h
an
dw
ash
ing,
corr
all
ing
of
live
sto
ck,
fly
con
tro
l,ro
ad
&d
rain
imp
rove
men
ts,
etc
55
%
Zh
an
get
al.
20
00
Qu
asi
-RC
TC
hin
aru
ral
No
,b
ut
the
two
vill
ages
are
sim
ilar
Pip
edw
ate
rsu
pp
ly5
0%
PREVENTION OF DIARRHOEA i201
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
study34 had used a placebo intervention to concealfrom the subjects and assessors which was the inter-vention group and which the control, and had founda smaller effect than the other community-basedtrials. An alternative explanation is that this studydid not focus on handwashing or provide soap. Allof the community-based trials which did not focuson handwashing or provide soap found smaller reduc-tions in diarrhoea than either of those which didthose things. Of the less focussed group, the partiallyblinded study34 found a reduction of 6% (95%CI: �44–15%) while the other, open trial of thiskind35 found 25% (95% CI 15–34%) so the lack ofblinding seems to explain at least part of the differ-ence. Of course, such non-focussed trials wereexcluded from our earlier review,10 and none of theintervention studies focussed on handwashing (Figure2a) used a placebo intervention. The consistency ofeffect between observational and intervention studiesled us to judge that the effect is probably genuine, butmore research is needed to clarify this point.
Water quality interventionsEnthusiasts for point-of-use household-based watertreatment argue that the smaller effect ofsource-based interventions is due to subsequent con-tamination of the water on its way to, or during stor-age in the household. Sceptics point to the lack ofsignificant effect found in the blinded studies, attri-buting the difference to courtesy bias or placeboeffect. They suggest that consumers are less consciousof treatment administered centrally, and so less likelyto show these forms of bias in their self-reporting ofdiarrhoea.
These findings have given rise to a lively debateabout the desirability of scaling-up household-basedwater treatment.36,37 The only certain conclusion isthat the implications are uncertain, and it is unlikelythat a Delphi process would arrive at an amicableconsensus.
There are several reasons to believe that the biasmay apply to water quality but which are less appli-cable to handwashing. First, there is biological plau-sibility; in many settings, a subject would have toingest very large amounts of water in order to
consume an infectious dose of a bacterial pathogen.Second, the anomaly that the reduction in diarrhoeaseems to be independent of the quality of the ambientwater before it is treated. Third, the observationalstudies of drinking-water quality (e.g. see ref.38) donot show such large effects as the point-of-use inter-vention trials. And fourth, most of these trials werefunded by manufacturers of water treatment chemi-cals or equipment.
In these circumstances the data from the recentspate of trials—and hence, any systematic reviewbased on them—do not offer a firm basis for judgingthe effect of water quality improvements. The optionsinclude adopting the effect of the four blinded trials(no significant reduction in any trial), adopting thepooled effect of the source-based intervention studies(three trials among children under five gave a 7%reduction), or keeping the pre-existing consensusview arrived at two decades ago in reviews of obser-vational and source-based intervention studies,4,7 thatwater quality improvements can be expected to beassociated with a reduction of some 17% in diarrhoearisk. This latter figure was proposed for LiST as asso-ciated with use of an improved water supply, which isavailable within a reasonable distance, requiring areturn journey of 30 min or less.
One theme which emerged from the debate was that‘harder’ outcomes such as care-seeking for diarrhoeawould be more objective and less prone to bias than arelatively ‘softer’ outcome such as self-reported diar-rhoea. One well-known water quality trial,39 excludedfrom our review because the outcome included gastro-enteritis without diarrhoea, found that during thestudy period ‘the number of visits to physicians forgastrointestinal symptoms and of hospitalizations wassimilar in both groups.’ In this context, one recenttrial with both handwashing and water qualityarms40 is of particular interest, as the authors docu-mented the effect of each arm both in terms ofself-reported diarrhoea and of care-seeking (seeTable 5). The effect of each intervention is very sim-ilar, whichever the outcome. This suggests, either thatany bias affecting the former outcome also affectedthe latter; or that the former was not susceptible tobias as the blinded studies suggest.
Table 5 Comparison of effects of (i) promotion of hand washing with soap, and (ii) household water treatment, measuredin terms of weekly prevalence of diarrhoea, and of frequency of consulting a practitioner for treatment for diarrhoea22
Intervention group Reduction in weeklyself-reported diarrhoeaprevalence (95% CI)
Reduction in care-seekingfor diarrhoea in children<5 years (95% CI)
Soap and hand washing promotion 45% (12–68%) 48% (15–71%)Bleach water treatment 53% (22–75%) 54% (22–77%)Flocculant-disinfectant water treatment 59% (29–82%) 61% (31–84%)Flocculant-disinfectant plus hand washing with soap 50% (18–72%) 55% (23–77%)
i202 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
Excreta disposalIt is in this area that the evidence is weakest. It is notsurprising that trials are few, as provision of hundredsof latrines is expensive. Observational studies cannotbe trusted, in view of evidence that people inlatrine-owning households (a self-selected group)behave more hygienically than others, even inrespects which have nothing to do with excreta dis-posal.41,42 This is not to say that excreta disposal hasno effect on diarrhoea. Indeed, there is a striking con-sistency between the reductions found in variousreviews of 36%,7 32%,9 20–51% (the four Chinese stu-dies) and 22–43%.33 That being so, there is notenough evidence to justify a departure from the pre-vailing consensus, published nearly two decades ago7
and widely cited with approval since then, that sani-tation reduces diarrhoea risk by about 36%. This thenis the strength of effect proposed for the LiST model.
GeneralThe lack of conclusive evidence of the effect on diar-rhoea of improvements in water, hygiene and sanita-tion in developing countries is not an excuse forinaction.43 We know enough to do a lot of good. Inspite of doubts about the detail, it is clear that suchenvironmental interventions can have a substantialeffect on diarrhoea morbidity, and the very few rele-vant studies44 confirm that they have a similar effectupon diarrhoea mortality. Moreover, water, hygieneand sanitation have other important benefits, includ-ing the emancipation of women from drudgery andthe enhancement of human dignity, and even otherhealth benefits such as the control of trachoma and ofintestinal helminths.
It is not entirely surprising that the evidence isweak; studies of mortality are fraught with ethicaland logistic problems; water supplies and sanitationare expensive interventions to trial; and the engineerswho install them are not accustomed to trials as animportant part of their professional culture. Moreover,it is particularly difficult to blind a trial of an inter-vention involving the provision of hardware or thepromotion of behaviour change, and as we have
seen, it appears that the lack of blinding can lead tosubstantial bias if the outcome is ‘soft’.
In the context of a literature of uneven quality suchas this, these reviews can also teach us that there isno definitive systematic review, or objective set ofrules for conducting one. The examples here showhow the result of a systematic review depends upona number of questions, such as the definition of theoutcome (should only diarrhoea be accepted, andvomiting rejected?), of the intervention (is seweragein Brazil different from pit latrines in Africa?), on therange of languages accepted (in several recentreviews,13,45 more than a third of useful studieswere in Chinese) and on which are regarded as thegravest methodological deficiencies. Each of thosequestions has more than one reasonable answer.Judgement-free data are a myth.
Supplementary dataSupplementary data are available at IJE online.
FundingThe US Fund for UNICEF from the Bill & MelindaGates Foundation (grant 43386, partially) to‘Promote evidence-based decision making in design-ing maternal, neonatal and child health interventionsin low- and middle-income countries’ and UNICEF,Unilever research and WaterAid.
AcknowledgementsStimulating discussions with colleagues in the CHERGhave helped to improve this article; their input isgratefully acknowledged. So are the major contribu-tions of Tom Clasen as principal author of the reviewson water and on sanitation, the patient and construc-tive support of Therese Dooley and the administrativewizardry of Eileen Chappell.
Conflict of interest: None declared.
KEY MESSAGES
� Effect of handwashing with soap is most consistent at roughly 48% reduction in diarrhoea.
� Effect of water quality improvements found in RCTs seems to be affected by bias – not seen inblinded studies.
� Evidence for effect of sanitation is weakest – randomized trials are needed – but may be 36%reduction.
� Though evidence is weak compared with clinical RCTs, it is enough for action.
� Analysing such evidence needs more than algorithms – it requires judgement.
PREVENTION OF DIARRHOEA i203
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
References1 Rosen S, Vincent JR. African Economic Policy Discussion
Paper 69. 2001. Household Water Resources and RuralProductivity in Sub-Saharan Africa: A Review of theEvidence.Cambridge, MA: John F. Kennedy School ofGovernment, Harvard University.
2 Mozynski P. BMA says inadequate sanitation is a globalcrisis. BMJ 2008;336:117.
3 Blum D, Feachem RG. Measuring the impact of watersupply and sanitation investments on diarrhoeal diseases:problems of methodology. Int J Epidemiol 1983;12:357–65.
4 Esrey SA, Feachem RG, Hughes JM. Interventions for thecontrol of diarrheal diseases among young children:improving water supplies and excreta disposal facilities.Bull WHO 1985;63:757–72.
5 Esrey SA, Habicht J-P. Cornell International NutritionMonograph Series 15. The Impact of Improved WaterSupplies and Excreta Disposal Facilities on DiarrhealMorbidity, Growth, and Mortality among Children. Ithaca,NY: Division of Nutritional Sciences, Cornell University,1985.
6 Esrey SA, Potash JB, RobertsL , Shiff C. Health Benefitsfrom Improvements in Water Supply and Sanitation:Survey and Analysis of the Literature on SelectedDiseases. WASH Technical Report 66. Rosslyn, VA:Environmental Health Project, for USAID, 1990.
7 Esrey SA, Potash JB, Roberts L, Shiff C. Effects ofimproved water supply and sanitation on ascariasis, diar-rhea, dracunculiasis, hookworm infection, schistosomia-sis, and trachoma. Bull WHO 1991;69:609–21.
8 Cairncross S, Valdmanis V. Water supply, sanitation andhygiene promotion. In: Jamison DT, Breman JG,Measham AR et al. (eds). Disease Control Priorities inDeveloping Countries. 2nd edn. Washington DC: TheWorld Bank, 2006, pp. 771–92.
9 Fewtrell L, Kaufmann RB, Kay D, Enanoria W, Haller L,Colford JM Jr. Water, sanitation, and hygiene interven-tions to reduce diarrhoea in less developed countries: asystematic review and meta-analysis. Lancet Infect Dis2005;5:42–52.
10 Curtis V, Cairncross S. Effect of washing hands with soapon diarrhoea risk in the community: a systematic review.Lancet Infect Dis 2003;3:275–81.
11 Clasen T, Schmidt W-P, Rabie T, Roberts I, Cairncross S.Interventions to improve water quality for preventingdiarrhoea: a systematic review and meta-analysis. BMJ2007;334:782(Evid Based Med 2007;12;127–28).
12 Clasen T, Roberts I, Rabie T, Cairncross S. Interventionsto improve water quality for preventing diarrhoea(Protocol for a Cochrane Review). The Cochrane Library,Issue 2. Oxford: Update Software, 2004.
13 Clasen T, Bostoen K, Schmidt W-P et al. Interventions toimprove excreta disposal for preventing diarrhoea(Protocol for a Cochrane Review). The Cochrane Library,Issue 2. Oxford: Update Software, 2008.
14 Walker N, Fischer CL, Bahl R et al. The CHERG methodsand procedures for estimating effectiveness of interven-tions on cause-specific mortality. Int J Epidemiol 2010;39(Suppl 1):i88–101.
15 Atkins D, Best D, Briss PA et al. Grading quality of evi-dence and strength of recommendations. BMJ 2004;328:1490.
16 Hoque B, Chakraborty J, Chowdhury J et al. Effectsof environmental factors on child survival inBangladesh: a case control study. Public Health 1999;113:57–64.
17 Khan M. Interruption of shigellosis by handwashing.Trans R Soc Trop Med Hyg 1982;76:164–68.
18 Shahid N, Greenough W, Samadi A, Huq M, Rahaman N.Hand washing with soap reduces diarrhoea and spread ofbacterial pathogens in a Bangladesh village. J DiarrhoealDis Res 1996;14:85–89.
19 Ejemot RI, Ehiri JE, Meremikwu MM, Critchley JA. Handwashing for preventing diarrhoea. Cochrane Database of SystRev. (Issue 1). Art. No.: CD004265.
20 Han AM, Hlaing T. Prevention of diarrhoea and dysenteryby hand washing. Trans R Soc Trop Med Hyg 1989;83:128–31.
21 Luby SP, Agboatwalla M, Painter J, Altaf A,Billhimer WL, Hoekstra RM. Effect of intensive hand-washing promotion on childhood diarrhea in high-riskcommunities in Pakistan: a randomized controlled trial.JAMA 2004;291:2547–54.
22 Luby SP, Agboatwalla M, Painter J et al. Combiningdrinking water treatment and hand washing for diarrheaprevention, a cluster randomised control trial. Trop MedInt Health 2006;11:479–89.
23 Messou E, Sangare SV, Josseran R, Le Corre C, Guelain J.Effect of hygiene measures, water sanitation and oralrehydration therapy on diarrhea in children less thanfive years old in the south of Ivory Coast. Bull SocPathol Exot 1997;90:44–47.
24 Crump JA, Otieno PO, Slutsker L et al. Household basedtreatment of drinking water with flocculant-disinfectantfor preventing diarrhoea in areas with turbid sourcewater in rural western Kenya: cluster randomised con-trolled trial. BMJ 2005;331:478.
25 Kirchhoff LV, McClelland KE, Do Carmo PM, Araujo JG,De Sousa MA, Guerrant RL. Feasibility and efficacy ofin-home water chlorination in rural North-easternBrazil. J Hyg (Lond) 1985;94:173–80.
26 Austin CJ. Investigation of in-house water chlorination and itseffectiveness for rural areas of the Gambia [dissertation]. NewOrleans: Tulane University School of Public Health andTropical Medicine, 1993.
27 Colford JM Jr, Rees JR, Wade TJ et al. Participant blind-ing and gastrointestinal illness in a randomized, con-trolled trial of an in-home drinking water intervention.Emerg Infect Dis 2002;8:29–36.
28 Colford JM Jr, Wade TJ, Sandhu SK et al. A randomized,controlled trial of in-home drinking water intervention toreduce gastrointestinal illness. Am J Epidemiol 2005;161:472–82.
29 Yan Z-S, Wang G-F, Cui C et al. An observation of theeffect on reducing the fly density and diarrhoea of theuse of double urn funnel lavatory in faeces management.Henan Yu Fang Yi Xue Za Zhi [Henan J Prev Med] 1986;1986:73–76.
30 Hu X-R, Liu G, Liu S-P et al. [Field evaluation ofthe effect of diarrhoea control of methanogenesistreatment of human and animal faeces and rubbish ofcellulose nature in the rural areas of Xiang Cheng.Henan Yu Fang Yi Xue Za Zhi [Henan J Prev Med] 1988;1988:11–13.
i204 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
31 Xu J-Z. Observation on the efficacy of three squaresseptic tank lavatory for disease prevention. Huan Jing YuJian Kang Za Zhi [J Environ Health] 1990;1990:250–52.
32 Xu G-X, Zhu X-L. The assessment of the effects for pre-vention of diseases by non-hazardous treatment of nightsoil at experimental spots in rural areas. Wei Sheng Yan Jiu[Wei Sheng Yan Jiu] 1994;23:23–37).
33 Barreto ML, Genser B, Strina A et al. Prado MS, MatosSMA, dos Santos LA, Cairncross S. Effect of city-widesanitation programme on reduction in rate of childhooddiarrhoea in northeast Brazil: assessment by two cohortstudies. The Lancet 2007;370:1622–28.
34 Haggerty PA, Muladi K, Kirkwood BR, Ashworth A,Manunebo M. Community-based hygiene education toreduce diarrhoeal disease in rural Zaire: impact of theintervention on diarrhoeal morbidity. Int J Epidemiol1994;23:1050–59.
35 Stanton BF, Clemens JD. An educational intervention foraltering water-sanitation behaviors to reduce childhooddiarrhea in urban Bangladesh. Am J Epidemiol 1987;125:292–301.
36 Schmidt WP, Cairncross S. Household water treatment inpoor populations: is there enough evidence for scaling upnow? Environ Sci Technol 2009;43:986–92.
37 Clasen T, Bartram J, Colford J et al. Comment on Schmidt& Cairncross. Environ Sci Technol 2009;43:5542–44.
38 Moe CL, Sobsey MD, Samsa GP, Mesolo V. Bacterial indi-cators of risk of diarrhoeal disease from drinking-water inthe Philippines. Bull World Health Organ 1991;69:305–17.
39 Payment P, Richardson L, Siemiatycki J, Dewar R,Edwardes M, Franco E. A randomized trial to evaluatethe risk of gastrointestinal disease due to consumptionof drinking water meeting current microbiological stan-dards. Am J Public Health 1991;81:703–8.
40 Luby SP, Agboatwalla M, Painter J et al. Combiningdrinking water treatment and hand washing for diar-rhoea prevention, a cluster randomised controlled trial.Trop Med Int Health 2006;11:479–89.
41 Hoque BA, Mahalanabis D, Alam MJ, Islam MS. Post-defecation handwashing in Bangladesh: practice andefficiency perspectives. Public Health 1995;109:15–24.
42 Strina A, Cairncross S, Barreto ML, Larrea C, Prado MS.Childhood diarrhea and observed hygiene behavior inSalvador, Brazil. Am J Epidemiol 2003;157:1032–38.
43 Bartram J, Cairncross S. Hygiene, sanitation and water;forgotten foundations of health. Lancet, in press.
44 Victora CG, Smith PG, Vaughan JP et al. Water supply,sanitation and housing in relation to the risk ofinfant mortality from diarrhoea. Int J Epidemiol 1988;17:651–54.
45 Fung IC, Cairncross S. Ascariasis and handwashing. TransR Soc Trop Med Hyg 2009;103:215–22.
PREVENTION OF DIARRHOEA i205
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from
by guest on Decem
ber 17, 2014http://ije.oxfordjournals.org/
Dow
nloaded from