Upload
balamand
View
0
Download
0
Embed Size (px)
Citation preview
RISK FACTORS
The effects of waterpipe tobacco smoking onhealth outcomes: a systematic reviewElie A Akl,1,2� Swarna Gaddam,2 Sameer K Gunukula,2 Roland Honeine,1
Philippe Abou Jaoude1 and Jihad Irani3
1Department of Medicine, State University of New York at Buffalo, Buffalo, NY, USA, 2Department of Family Medicine, StateUniversity of New York at Buffalo, Buffalo, NY, USA and 3Faculty of Health Sciences, University of Balamand, Beirut, Lebanon
�Corresponding author. Department of Medicine, State University of New York at Buffalo, ECMC-CC 142, 462 Grider Street,Buffalo, NY 14215, USA. E-mail: [email protected]
Accepted 5 January 2010
Background There is a need for a comprehensive and critical review of theliterature to inform scientific debates about the public healtheffects of waterpipe smoking. The objective of this study wastherefore to systematically review the medical literature for theeffects of waterpipe tobacco smoking on health outcomes.
Methods We conducted a systematic review using the Cochrane Collabora-tion methodology for conducting systematic reviews. We rated thequality of evidence for each outcome using the Grading of Recom-mendations Assessment, Development and Evaluation (GRADE)methodology.
Results Twenty-four studies were eligible for this review. Based on theavailable evidence, waterpipe tobacco smoking was significantlyassociated with lung cancer [odds ratio (OR)¼ 2.12; 95% confi-dence interval (CI) 1.32–3.42], respiratory illness (OR¼ 2.3; 95%CI 1.1–5.1), low birth-weight (OR¼ 2.12; 95% CI 1.08–4.18) andperiodontal disease (OR¼ 3–5). It was not significantly associatedwith bladder cancer (OR¼ 0.8; 95% CI 0.2–4.0), nasopharyngealcancer (OR¼ 0.49; 95% CI 0.20–1.23), oesophageal cancer(OR¼ 1.85; 95% CI 0.95–3.58), oral dysplasia (OR¼ 8.33; 95% CI0.78–9.47) or infertility (OR¼ 2.5; 95% CI 1.0–6.3) but the CIs didnot exclude important associations. Smoking waterpipe in groupswas not significantly associated with hepatitis C infection(OR¼ 0.98; 95% CI 0.80–1.21). The quality of evidence for thedifferent outcomes varied from very low to low.
Conclusion Waterpipe tobacco smoking is possibly associated with a numberof deleterious health outcomes. There is a need for high-qualitystudies to identify and quantify with confidence all the healtheffects of this form of smoking.
Keywords Oesophageal neoplasms, infant, low birth-weight, lung neoplasms,hepatitis C, tobacco, waterpipe
BackgroundTobacco smoking using waterpipe—also known asarguileh, hookah and shisha—is an older form of
tobacco consumption traditional of the easternMediterranean region (Figure 1).1 The waterpipedevice indirectly heats the tobacco. The resulting
Published by Oxford University Press on behalf of the International Epidemiological Association
� The Author 2010; all rights reserved. Advance Access publication 4 March 2010
International Journal of Epidemiology 2010;39:834–857
doi:10.1093/ije/dyq002
834
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
smoke then passes through a column of water beforebeing inhaled through the mouth using a pipe.2 Thesetwo unique features of this form of smoking areassumed to minimize its tobacco-related health haz-ards. Also, waterpipe tobacco smoke is produced at amuch lower temperature than cigarette smoke, sug-gesting that the toxins may be different for these dif-ferent forms of tobacco smoking.3 Another uniquefeature is the social nature of waterpipe smoking,whereby family members and friends may share thesame device, which has been hypothesized to causethe transmission of communicable diseases.
Waterpipe tobacco smoking has been spreadingglobally at a remarkable pace.4 Although its use inthe eastern Mediterranean region has been, untilrecently, confined to adults, it is now gaining popu-larity among university students5,6 and teenagers.7–9
Waterpipe smoking is also spreading to Westerncountries such as Australia,10 the UK,11 Canada12
and the USA,13 where it is also affecting youngpeople and adolescents.14,15
Waterpipe tobacco smoking is generally consideredas a public health threat and the American LungAssociation has recently called it the ‘emergingdeadly trend’.2 In fact, waterpipe smoking has beensuspected to be a risk factor for a number oftobacco-related diseases such as lung cancer,16 oeso-phageal cancers,17 cardiovascular disease18 and
adverse pregnancy outcomes.19 In addition, the water-pipe device may expose its user (via its non-tobaccocomponents) to metals and cancer-causing chemi-cals.20,21 The potential association of waterpipe smok-ing with communicable diseases such as hepatitis C22
and tuberculosis23 (via its shared and repetitive usewithout proper sanitation) has also been investigated.
While one article reviewed the effects of waterpipesmoking on health outcomes,24 we have identified nopublished systematic review or meta-analysis of thetopic. The objective of this study was therefore to sys-tematically review the medical literature for theeffects of waterpipe tobacco smoking on healthoutcomes.
MethodsEligibility criteriaWe included observational studies (i.e. cohort studies,case–control studies and cross-sectional studies)assessing the association between waterpipe tobaccosmoking and health outcomes. We excluded casereports, case series, outbreak investigations andabstracts. We also excluded studies assessing physio-logical outcomes [e.g. Forced Expiratory Volume in 1Second (FEV1)], assessing waterpipe use fornon-tobacco smoking purposes (e.g. marijuana smok-ing), not distinguishing waterpipe smoking fromother forms of smoking, and not reporting any mea-sure of association.
Search strategyIn June 2008, we electronically searched the followingdatabases: MEDLINE (1950 onwards; access viaOVID), EMBASE (1980 onwards; access via OVID)and ISI the Web of Science using a detailed searchstrategy with no language restrictions (Appendix 1).We designed the strategy based on a preliminaryreview of relevant articles, an extensive Internetsearch for waterpipe synonyms and the search strat-egy of a systematic review on interventions for water-pipe smoking cessation.25 Two medical librariansreviewed and commented on the search strategy. Wealso reviewed the reference lists of included and rel-evant papers and used the ‘Related Articles’ feature inPubMed.
Selection processTwo reviewers independently screened the title andabstract of identified citations for potential eligibilityusing a standardized screening guide. We retrievedthe full texts of citations judged potentially eligibleby at least one reviewer. Then, two reviewers indepen-dently screened the full texts for eligibility using astandardized and pilot-tested form. They resolvedtheir disagreements by discussion or by consulting athird reviewer.
Figure 1 Annotated figure of a waterpipe device
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 835
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Data abstractionTwo reviewers used a standardized and pilot-testedform to independently abstract data. They resolvedtheir disagreements by discussion or by consultinga third reviewer. The abstracted data related tostudy design, population, exposure, outcomes, meth-odological features, results and funding. For exposuremeasurement, we assessed whether the instrumentwas standardized to measure the unique smoking pat-terns and characteristics of waterpipe smoking. Werecorded the results of analyses restricted to waterpipeonly smokers (i.e. people who smoked only water-pipe) as well as the results of analyses that includedall waterpipe smokers (i.e. people who smoked water-pipe and other forms of tobacco). We recorded theeffect measures derived from the regression modelsthat adjusted for the maximum number of covariates.We rated the overall quality of evidence for each out-come using the Grading of RecommendationsAssessment, Development and Evaluation (GRADE)approach (Appendix 2).26
Data analysisWe calculated the kappa statistic to evaluate theagreement between the two reviewers assessing full
texts for eligibility. We conducted meta-analyses forthe outcomes for which at least two studies reportedeffect estimates of their association with waterpipetobacco smoking. When the authors reportedodds ratios (ORs) for more than one analysis weselected the OR according to the following order:(i) analysis adjusted for other forms of tobaccosmoking (choosing the one adjusting for the maxi-mum number of confounders); (ii) analysisrestricted to waterpipe only smokers; (iii) analysisnot adjusted for other forms of tobacco smokingand not restricted to waterpipe only smokers.We used the ORs to calculate the correspondingln(ORs) and standard errors. We then pooled, foreach outcome, the ln(ORs) of eligible studies usingthe generic inverse variance and the random effectsmodel in Review Manager Version 5.0.20. Wemeasured homogeneity across study results usingthe I2 statistic.27 We checked for possible publicationbias using inverted funnel plots.
ResultsFigure 2 shows the study flow. We included 23reports on: lung cancer (n¼ 6),16,28–32 bladder cancer
1658 citations identified
1570 citations screened for retrieval 25 papers excluded: • report of already reported data (4) • case reports or case series (4) • review of the literature (1) • inappropriate study design (3) • no distinction from other forms of smoking (4) • no outcome of interest reported (5)
• no measure of association reported (2)
• abstract (1)
• about long stem pipes, not waterpipes (1)
44 potentially eligible papers retrieved
23 reports included in systematic review
88 duplicate citations
10 studies (11 analyses) included in meta-analyses
2 papers published after search date and identified 2 papers identified through checking of citations list
Figure 2 Study flow diagram
836 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
(n¼ 1),33 oesophageal cancers (n¼ 1),17 nasopharyn-geal cancer (n¼ 1),34 dysplasia of the oral mucosa(n¼ 1),35 pregnancy outcomes (n¼ 3),19,36,37 peri-odontal disease (n¼ 5),38–42 hepatitis C infection(n¼ 3),22,43,44 respiratory illness (n¼ 1)45 and infertil-ity (n¼ 1).46 Agreement between reviewers for studyeligibility was excellent (kappa¼ 0.963).
Lung cancerFive of the six included studies were case–controlstudies assessing lung cancer diagnosis,16,28–31 andone was a retrospective cohort study assessing lungcancer mortality (Table 1).32 One was conducted inNorthern India,28 one was conducted in Tunisia,31
whereas rule four reported data from the same pop-ulation in China. While in the recent global epidemicthe tobacco is processed and flavoured and indirectlyheated by the charcoal, in most of the included stu-dies (conducted in China and India) the tobacco istypically unprocessed and burned directly by charcoal.
The pooled OR for the association of waterpipetobacco smoking with lung cancer diagnosis was2.12 [95% confidence interval (CI) 1.32–3.42;I2¼ 0%] (Figure 3). We judged the overall quality of
evidence to be very low (Appendix 3). The calculatedcrude relative risk (RR) for the association with lungcancer mortality was 4.39 (3.82–5.04). We judged theoverall quality of evidence to be very low (Appendix3). A sensitivity analysis restricted to studies with nomajor methodological limitations, included the studyby Hsairi et al.31 for which the OR was 3.0 (95% CI1.2–7.6).
Bladder cancerThe one included case–control study33 reported apotentially protective association between waterpipetobacco smoking and bladder cancer diagnosis of 0.8(95% CI 0.2–4.0) (Table 1). A stratified analysis by theamount of cigarette smoking suggested a potentialinteraction between waterpipe and cigarette smokingbut the CI overlapped and no interaction test wasreported. An interaction test we ran was statisticallynot significant (P¼ 0.55). We judged the overall qual-ity of evidence to be very low (Appendix 3).
Oesophageal cancerThe one study assessing oesophageal cancer diagnosiswas a case–control study (Table 1).17 The reported ORfor the association of waterpipe tobacco smoking withoesophageal cancer diagnosis was 1.85 (95% CI 0.95–3.58). We judged the overall quality of evidence to below (Appendix 3).
Nasopharyngeal cancerThe one included study was a case–control study(Table 1). The OR for the association of waterpipetobacco smoking with nasopharyngeal cancer was
0.49 (95% CI 0.20–1.23). We judged the overall qual-ity of evidence to be very low (Appendix 3).
Oral dysplasiaThe one included study was a cross-sectionalstudy that recruited exclusively subjects who prac-ticed ‘takhzeen al-qat’, i.e. a practice that is distinctfrom waterpipe smoking and consists of chewinga green-leaved plant for its stimulant effect(Table 1).35 The OR for the association of waterpipetobacco smoking with dysplasia of the oral mucosa onchewing side was 8.33 (95% CI 0.78–9.47). There wereno events observed in either group on non-chewingside. We judged the overall quality of evidence to bevery low (Appendix 3).
Pregnancy outcomesOf the three included studies, two were retrospectivecohort studies19,36 and one was a case–control study(Table 2).37 All three studies assessed lowbirth-weight. One study also reported on Apgarscore, pulmonary problems, malformations and peri-natal complications.19 The pooled OR for the associa-tion of waterpipe tobacco smoking with lowbirth-weight was 2.12 (95% CI 1.08–4.18; I2
¼ 0%)(Figure 4). The reported OR for the association ofwaterpipe tobacco smoking with newborn pulmonaryproblems was 3.65 (95% CI 1.52, 8.75). The associa-tions were not significant for Apgar scores at 1 and 5min, malformations or perinatal complications. Wejudged the overall quality of evidence for pregnancyoutcomes to be low (Appendix 3).
Periodontal diseaseOf the five included studies,38–42 four were conductedin the same (or in a subgroup of the same) group ofparticipants (Table 3).39–42 These four studies werecross-sectional and evaluated the same outcome (i.e.periodontal disease) using different outcome mea-surement (periodontal bone height loss, plaqueindex and gingivitis, deepening of the sulci or pockets,vertical periodontal bone loss). We did not pool thefour related studies evaluating periodontal disease astheir data were derived from the same participants.Their results were consistent in showing a statisticallysignificant association of waterpipe tobacco smokingwith periodontal disease (OR¼ 3–5). We judged theoverall quality of evidence to be low (Appendix 3).
The fifth included study was a cohort study with7 days follow-up after surgical removal of mandibularthird molars and evaluated the outcome of drysocket.38 Dry socket, or alveolar osteitis, is the mostcommon complication following tooth extractions. Itis caused by the dislodgement of the blood clot at thesite of the tooth extraction, exposing underlying boneand nerves and causing increasing pain. The reportedRR for the association of waterpipe tobacco smoking
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 837
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
1C
hara
cter
isti
cso
fin
clu
ded
stu
die
sass
essi
ng
the
effe
cto
fw
ate
rpip
eto
bacc
osm
ok
ing
on
can
cer
ou
tco
mes
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
Qia
o1
6
�S
tud
yd
esig
n:
case
–co
ntr
ol
stu
dy
�F
un
din
g:
Ch
ina
Sta
teS
cien
cean
dT
ech
no
logy
Co
mm
issi
on
;an
dth
eU
SN
ati
on
al
Can
cer
Inst
itu
tegra
nt
�S
etti
ng
an
dp
erio
d:
Gej
iuci
ty,
Yu
nn
an
Pro
vin
ce,
Ch
ina,
inte
rvie
ws
con
-d
uct
edin
19
85
�C
ase
s:1
07
lun
gca
nce
rca
ses
inm
ale
s3
5–8
0ye
ars
old
rep
ort
edto
Lab
or
Pro
tect
ion
Inst
itu
teo
fth
eY
un
nan
Tin
Co
rpo
rati
on
(YT
C)
du
rin
g1
96
7–8
4�
Co
ntr
ols
:1
07
con
tro
lsch
ose
nsy
stem
ati
call
yb
yse
lect
ing
ever
y2
0th
per
son
fro
mth
eli
sto
fall
livi
ng
past
or
pre
sen
tw
ork
ers
of
the
YT
C;
on
eco
ntr
ol
per
case
�T
ype:
Ch
ines
ew
ate
rp
ipes
,ci
gare
ttes
�M
easu
rem
ent
too
l:se
lf-d
evel
op
edq
ues
-ti
on
nair
e,n
ost
an
-d
ard
izati
on
rep
ort
ed;
cum
ula
tive
con
-su
mp
tio
nca
lcu
late
das
pip
eye
ars
�E
xp
osu
rele
vels
:C
ase
s:m
ean
pip
eye
ars¼
17
7(r
an
ge
0–5
60
);C
on
tro
ls:
mea
np
ipe
years¼
12
2(r
an
ge
0–4
80
)
Lu
ng
can
cer
dia
gn
osi
s�
Mea
sure
men
tto
ol:
det
ecte
db
yra
dio
logy
an
dco
nfi
rmed
by
his
tolo
gy
or
cyto
logy
�B
lin
din
go
fo
utc
om
ead
ju-
dic
ato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
seri
eso
fin
cid
ent
case
sb
ut
un
clea
rw
het
her
all
inci
den
tca
ses
rep
ort
ed,
con
tro
lsw
ere
sele
cted
fro
mth
esa
me
stu
dy
base
as
case
s�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
eev
alu
ati
on
:ye
s;st
an
dard
ized
exp
osu
rem
easu
rem
ent:
no
.S
urr
ogate
sw
ere
inte
r-vi
ewed
for
10
%o
fca
ses
an
d6
%o
fco
ntr
ols
�C
on
fou
nd
ing:
matc
hin
gfo
rage;
ad
just
men
tfo
rage;
no
ad
just
men
tre
po
rted
for
rad
on
�P
art
icip
ati
on
rate
:n
ot
rep
ort
ed
OR
com
pare
dw
ith
nev
ersm
ok
ing:
�1
.9(9
5%
CI
0.4
–9
.4)
(wate
rpip
eo
nly
smo
kin
g)
�2
.1(9
5%
CI
0.5
–9
.1)
(wate
rpip
ean
dci
gare
tte
smo
kin
g)
�0
.9(9
5%
CI
0.1
–5
.4)
(cig
are
tte
on
lysm
ok
ing)
Sta
tist
icall
ysi
gn
ific
an
td
ose
resp
on
seto
wate
rp
ipe
smo
kin
gO
R¼
3.4
(1.3
–8
.1)
by
qu
art
ero
fp
ipe-
years
Lu
bin
29
�S
tud
yd
esig
n:
case
–co
ntr
ol
stu
dy
�F
un
din
g:
Ch
ina
Sci
ence
an
dT
ech
no
logy
Co
mm
issi
on
;an
dth
eU
SN
ati
on
al
Can
cer
Inst
itu
te�
Un
clea
rw
het
her
this
isa
rep
ort
on
asu
b-
po
pu
lati
on
of
Qia
o1
8
�S
etti
ng
an
dp
erio
d:
Gej
iuci
ty,
Yu
nn
an
Pro
vin
ce,
Ch
ina,
stu
dy
con
du
cted
in1
98
5�
Case
s:7
4lu
ng
can
cer
case
sin
male
sw
ith
mea
nage
62
years
(ran
ge
35
–8
0)
ali
veat
the
tim
eo
fth
est
ud
yre
po
rted
toL
ab
or
Pro
tect
ion
Inst
itu
teo
fth
eY
TC
du
rin
g1
98
1–8
4�
Co
ntr
ols
:7
4co
ntr
ols
cho
sen
fro
mth
eli
sto
fall
livi
ng
past
or
pre
sen
tw
ork
ers
of
the
YT
C;
on
eco
ntr
ol
per
case
�T
ype:
Ch
ines
ew
ate
rp
ipes
,ci
gare
ttes
�M
easu
rem
ent
too
l:se
lf-d
evel
op
edq
ues
-ti
on
nair
e,n
ost
an
-d
ard
izati
on
rep
ort
ed;
cum
ula
tive
con
-su
mp
tio
nca
lcu
late
das
pip
eye
ars
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
cts:
nu
mb
ero
fp
ipe
years
(case
s/co
ntr
ols
):0
(6/1
6);
1–1
14
(18
/2
3);
11
5–2
20
(21
/1
5);
52
20
(29
/20
)
Lu
ng
can
cer
dia
gn
osi
s�
Mea
sure
men
tto
ol:
case
sco
nfi
rmed
by
an
ind
epen
-d
ent
pan
elo
fp
ath
olo
gis
ts,
clin
icia
ns
an
dcy
tolo
gis
ts�
Bli
nd
ing
of
ou
tco
me
ad
ju-
dic
ato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
seri
eso
fin
cid
ent
case
sb
ut
un
clea
rw
het
her
all
inci
den
tca
ses
rep
ort
ed;
excl
ud
edth
ose
wh
oh
ad
die
db
yth
eti
me
of
the
stu
dy;
con
tro
lsw
ere
sele
cted
fro
mth
esa
me
stu
dy
base
as
case
s�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
eev
alu
ati
on
:ye
s;st
an
dard
ized
exp
osu
rem
easu
rem
ent:
no
;va
li-
date
dco
nfo
un
der
mea
-su
rem
ent
(ars
enic
):n
o�
Co
nfo
un
din
g:
matc
hin
gfo
rage
(wit
hin
5-y
ear
age
gro
up
s)�
Part
icip
ati
on
rate
:n
ot
rep
ort
ed�
Data
wer
esp
ars
ean
dau
tho
rsw
ere
un
ab
leto
ad
just
for
imp
ort
an
tco
n-
fou
nd
ing
fact
ors
OR
com
pare
dw
ith
no
tob
acc
osm
ok
ing:
�3
.6(w
ate
rpip
eo
nly
smo
kin
g)
�1
.2(c
igare
tte
on
lysm
ok
ing)
�3
.5(m
ixed
smo
kin
g)
(co
nti
nu
ed)
838 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
1C
on
tin
ued
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
Lu
bin
30
�S
tud
yd
esig
n:
case
–co
ntr
ol
stu
dy
�F
un
din
g:
no
tre
po
rted
�S
etti
ng
an
dp
erio
d:
Gej
iuci
ty,
Yu
nn
an
Pro
vin
ce,
Ch
ina
�C
ase
s:4
27
male
lun
gca
nce
rp
ati
ents
,m
ean
age
63
years
(ran
ge
35
–7
5);
rep
ort
edb
etw
een
19
84
an
d1
98
8to
the
Can
cer
Reg
istr
yo
fth
eL
ab
or
Pro
tect
ion
Inst
itu
teo
fY
TC
or
toth
eG
ejiu
Cit
yC
an
cer
Reg
istr
y;3
39
fro
mY
un
nan
Tin
Co
rpo
rati
on
an
d8
8fr
om
Gej
iuC
ity
�C
on
tro
ls:
10
11
male
con
-tr
ols
,tw
oco
ntr
ols
per
case
,m
ean
age
62
years
(ran
ge
35
–7
5),
77
0fr
om
YT
Can
d2
41
fro
mG
ejiu
city
�T
ype:
Ch
ines
ew
ate
rpip
e,ci
gare
ttes
�M
easu
rem
ent
too
l:se
lf-d
evel
op
edst
an
-d
ard
ized
stru
ctu
red
qu
esti
on
nair
e,n
ost
an
dard
izati
on
rep
ort
ed;
con
sum
p-
tio
nca
lcu
late
das
pip
eye
ars
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
cts:
mea
nd
ura
tio
no
fp
ipe
on
lysm
ok
ing
is4
1ye
ars
Lu
ng
can
cer
dia
gn
osi
s�
Mea
sure
men
tto
ol:
con
-fi
rmed
by
ind
epen
den
tp
an
elo
fp
ath
olo
gis
ts,
clin
icia
ns,
rad
iolo
gis
tsan
dcy
tolo
gis
ts�
Bli
nd
ing
of
ou
tco
me
ad
ju-
dic
ato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
seri
eso
fin
cid
ent
case
s,co
ntr
ols
wer
eco
mm
un
ity
base
dan
dse
lect
edfr
om
the
sam
est
ud
yb
ase
as
case
s;�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
eev
alu
ati
on
:ye
s;st
an
dard
ized
exp
osu
rem
easu
rem
ent:
no
�C
on
fou
nd
ing:
matc
hin
gfo
rage
(wit
hin
5-y
ear
age
gro
up
s);
ad
just
men
tfo
rage,
nu
mb
ero
fye
ars
of
wo
rku
nd
ergro
un
d,
sou
rce
of
sub
ject
,ty
pe
of
resp
on
den
t�
Part
icip
ati
on
rate
:7
2%
for
case
s,8
8%
for
con
tro
ls
OR
com
pare
dw
ith
no
smo
kin
g:
�1.8
(95%
CI
0.8
–4.2
)(w
ate
rpip
eon
lysm
ok
ing)
�2.6
(95%
CI
1.1
–6.2
)(c
igare
tte
on
ly)
OR
sfo
rp
ipe
use
rssh
ow
incr
easi
ng
tren
dw
ith
incr
ease
dd
ura
tio
no
fu
se(t
est
for
tren
dst
a-
tist
icall
ysi
gn
ific
an
t)
Hsa
iri3
1
�S
tud
yd
esig
n:
case
–co
ntr
ol
stu
dy
�F
un
din
g:
no
tre
po
rted
�S
etti
ng
an
dp
erio
d:
Ari
an
a,
Tu
nis
,D
ecem
ber
19
88
toM
ay
19
89
�C
ase
s:1
10
lun
gca
nce
rp
ati
ents
;ep
ider
mo
id(5
6%
),an
ap
last
icsm
all
cell
(17
%),
un
dif
fere
n-
tiate
d(1
3%
),ad
eno
carc
i-n
om
a(1
2%
),an
ap
last
icla
rge
cell
(3%
),�
Co
ntr
ols
:1
10
resi
den
tso
fth
esa
me
are
a
�T
ype:
wate
rpip
e,p
ipe,
inh
ale
dto
bacc
o,
can
nab
is�
Mea
sure
men
tto
ol:
self
-dev
elo
ped
too
l,n
ost
an
dard
izati
on
rep
ort
ed�
Ex
po
sure
leve
lso
fin
clu
ded
sub
ject
s:n
ot
rep
ort
ed
Lu
ng
can
cer
dia
gn
osi
s�
Mea
sure
men
tto
ol:
his
tolo
gi-
call
yco
n-
firm
edin
70
%o
fca
ses,
‘ver
yp
rob
ab
le’
in3
0%
of
case
sb
ase
do
ncl
ini-
cal,
rad
iolo
gic
al
an
den
do
-sc
op
icsu
spic
ion
�B
lin
din
go
fo
utc
om
ead
ju-
dic
ato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
seri
eso
fin
cid
ent
case
s,co
ntr
ols
wer
eco
mm
un
ity
base
dan
dse
lect
edfr
om
the
sam
est
ud
yb
ase
as
case
s.�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
eev
alu
ati
on
:ye
s;st
an
dard
ized
exp
osu
rem
easu
rem
ent:
no
�C
on
fou
nd
ing:
matc
hin
gfo
rage,
sex
,co
nsu
mp
tio
no
fci
gare
ttes
per
day;
ad
just
-m
ent
for
age,
sex
,ci
gare
tte
con
sum
pti
on
,o
ccu
pati
on
al
risk
s�
Part
icip
ati
on
rate
:n
ot
rep
ort
ed
OR
com
pare
dw
ith
no
smo
kin
g:
�3.0
(95%
CI
1.2
–7.6
)a
Gu
pta
28
�S
tud
yd
esig
n:
case
–co
ntr
ol
stu
dy
�F
un
din
g:
Inte
rnati
on
al
Agen
cy
�S
etti
ng
an
dp
erio
d:
No
rth
ern
Ind
ia;
recr
uit
-m
ent,
Jan
uary
19
95
toJu
ne
19
97
�T
ype:
cigare
ttes
,b
idis
,h
oo
kah
�M
easu
rem
ent
too
l:se
lf-d
evel
op
edto
ol,
no
stan
dard
izati
on
Lu
ng
can
cer
dia
gn
osi
s�
Mea
sure
men
tto
ol:
det
ecte
db
yra
dio
logy
�S
elec
tio
nb
ias:
seri
eso
fh
osp
ital
base
din
cid
ent
case
s,co
mm
un
ity-
base
dco
ntr
ols
sele
cted
fro
mth
esa
me
stu
dy
base
as
case
s
OR
com
pare
dw
ith
no
smo
kin
g:
�1
.94
(95
%C
I0
.85
–4
.44
)(h
oo
kah
smo
kin
g)
inm
en
(co
nti
nu
ed)
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 839
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
1C
on
tin
ued
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
for
Res
earc
ho
nC
an
cer,
Lyo
nF
ran
ce;
an
dth
eIn
tern
ati
on
al
can
cer
rela
ted
tech
no
logy
tran
sfer
fell
ow
ship
aw
ard
�C
ase
s:2
65
inci
den
tca
ses
of
lun
gca
nce
r�
Co
ntr
ols
:5
25
visi
tors
an
datt
end
an
tso
fth
ep
ati
ents
;tw
oco
ntr
ols
per
case
�P
art
icip
an
tsw
ere
85
%m
ale
,o
fd
iffe
ren
tre
ligio
ns,
ages
ran
gin
gfr
om
<5
0to
47
0li
vin
gin
rura
lo
ru
rban
are
as
rep
ort
ed;
cum
ula
tive
con
sum
pti
on
calc
u-
late
das
pack
-yea
req
uiv
ale
nt
of
cigar-
ette
s(1
hook
ah¼
4ci
gare
ttes
)st
rati
fied
info
ur
gro
up
s(0
–9
;1
0–1
9;
20
–2
9;
30þ
)�
Ex
po
sure
leve
lso
fin
clu
ded
sub
ject
s:in
com
ple
tely
rep
ort
ed
an
dco
nfi
rmed
by
his
tolo
gy
or
cyto
logy
�B
lin
din
go
fo
utc
om
ead
ju-
dic
ato
r:n
ot
rep
ort
ed
�In
form
ati
on
bia
s:o
bje
ctiv
eo
utc
om
eev
alu
ati
on
:ye
s;st
an
dard
ized
exp
osu
rem
easu
rem
ent:
no
�C
on
fou
nd
ing:
matc
hin
gfo
rage
an
dse
x;
ad
just
men
tfo
rage,
an
ded
uca
tio
n.
No
ad
just
men
tfo
ran
yco
n-
fou
nd
ing
fact
or
incl
ud
ing
oth
erfo
rms
of
tob
acc
oco
nsu
mp
tion
�P
art
icip
ati
on
rate
:n
ot
rep
ort
ed�
Mis
sin
gd
ata
:5
4.3
%fo
rex
po
sure
�N
um
ber
sfo
rw
om
enw
ere
too
small
tod
eriv
est
ab
leri
skes
tim
ate
s
Haze
lto
n3
2
�S
tud
yd
esig
n:
retr
o-
spec
tive
coh
ort
stu
dy
�F
un
din
g:
Nati
on
al
Inst
itu
teo
fH
ealt
h
�S
etti
ng
an
dp
erio
d:
Gej
iuci
ty,
Yu
nn
an
Pro
vin
ce,
Ch
ina,
12
years
foll
ow
-up
(19
76
–8
8)
�E
xp
ose
d:
12
89
male
wate
rpip
e-o
nly
smo
ker
san
d2
30
6m
ale
wate
rpip
ean
dci
gare
tte
smo
ker
s�
No
n-e
xp
ose
d:
84
16
male
s
�T
ype:
Ch
ines
ew
ate
rpip
e,ci
gare
tte
�M
easu
rem
ent
tool:
no
tre
po
rted
,ca
lcu
-la
ted
as
cum
ula
tive
exp
osu
rean
dca
te-
go
rize
din
toq
uart
iles
of
lian
g/m
on
thye
ar
�M
easu
rem
ent
tim
ep
oin
ts:ev
ery
5ye
ars
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
cts:
cum
ula
tive
dose
qu
art
iles
:0
(n¼
84
16
),0
.16
–4
.44
(n¼
87
7),
4.4
4–6
.25
(n¼
88
8),
6.2
5–9
.21
(n¼
90
6),
9.2
1–8
2.1
9(n¼
92
4)
Lu
ng
can
cer
mo
rtali
ty�
Mea
sure
men
tto
ol:
no
tre
po
rted
�B
lin
din
go
fo
utc
om
ead
ju-
dic
ato
r:n
ot
rep
ort
ed�
Inci
den
ce:
7%
�S
elec
tio
nb
ias:
smo
ker
sw
ere
rep
rese
nta
tive
of
the
stu
dy
base
,n
on
-sm
ok
ers
sele
cted
fro
mth
esa
me
com
mu
nit
yas
smo
ker
s�
Info
rmati
on
bia
s:u
nli
kel
yfo
rm
ort
ali
tyo
utc
om
e;st
an
dard
ized
exp
osu
rem
easu
rem
ent:
no
�C
on
fou
nd
ing:
no
ad
just
-m
ent
(no
ad
just
edR
Rre
po
rted
;cr
ud
eR
Rca
lcu
-la
ted
fro
mre
po
rted
data
)�
Part
icip
ati
on
rate
:n
ot
rep
ort
ed�
8%
lost
tofo
llo
w-u
p
�R
Rco
mp
are
dw
ith
no
smo
kin
g¼
4.3
9(9
5%
CI
3.8
2–5
.04
)(w
ate
rp
ipe
smo
kin
g)
Bed
wan
i33
�S
tud
yd
esig
n:
case
–co
ntr
ol
stu
dy
�F
un
din
g:
Nati
on
al
Res
earc
hC
ou
nci
l-A
pp
lica
zio
ni
clin
ich
ed
ella
rice
rca
on
colo
-gic
a(C
NR
-AC
RO
Pro
ject
);an
dIt
ali
an
�S
etti
ng
an
dp
erio
d:
Gre
ate
rA
lex
an
dri
a,
Egyp
t;st
ud
yco
nd
uct
ed,
Jan
uary
19
94
toJu
ly1
99
6�
Case
s:1
51
male
inci
den
tca
ses
of
inva
sive
bla
dd
erca
nce
rw
ith
med
ian
age
61
years
(ran
ge
31
–7
4)
ad
mit
ted
toa
net
wo
rko
f
�T
ype:
wate
rpip
e,ci
gare
tte,
hash
ish
�M
easu
rem
ent
tool:
self
-dev
elo
ped
stru
c-tu
red
qu
esti
on
nair
e;st
an
dard
izati
on
not
rep
ort
ed;
part
icip
an
tsca
tego
rize
das
ever
smo
ker
san
dn
ever
Bla
dd
erca
nce
rd
iagn
osi
s�
Mea
sure
men
tto
ol:
cen
trall
yre
view
edh
is-
top
ath
olo
gic
al
dia
gn
osi
s�
Bli
nd
ing
of
ou
tco
me
�S
elec
tio
nb
ias:
ho
spit
al-
base
dca
ses
of
bla
dd
erca
nce
rco
nfi
rmed
wit
hin
the
year
pre
ced
ing
inte
r-vi
ew,
ho
spit
al-
base
dco
n-
tro
lsse
lect
edfr
om
the
sam
est
ud
yb
ase
as
case
s.S
elec
tio
nre
late
dto
exp
o-
sure
statu
s(i
nth
eca
seo
f
OR
com
pare
dw
ith
nev
ersm
ok
ing:
�0
.8(9
5%
CI
0.2
–4
.0)
(wate
rpip
esm
ok
ing)
�5
.4(9
5%
CI
2.4
–1
2.1
)(<
20
/day
cigare
tte
smo
kin
g) (c
on
tin
ued
)
840 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
1C
on
tin
ued
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
Ass
oci
ati
on
for
Can
cer
Res
earc
hgen
eral
an
dte
ach
ing
ho
spit
als
.�
Co
ntr
ols
:1
57
male
sad
mit
ted
for
acu
te,
no
n-
neo
pla
stic
,n
on
-uri
nary
tract
,n
on
-sm
ok
ing-r
elate
dco
nd
itio
ns,
med
ian
age
50
years
(ran
ge
32
–7
4)
�3
9fe
male
case
san
d3
0fe
male
con
tro
lsw
ere
excl
ud
edb
ecau
se‘o
nly
on
efe
male
case
was
asm
ok
er’
smo
ker
s.A
ge
at
start
ing,
du
rati
on
of
hab
it,
ave
rage
dail
yco
nsu
mp
tio
nex
tract
edfr
om
his
-to
ryb
ut
no
tre
po
rted
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
cts:
no
tre
po
rted
ad
jud
icato
r:n
ot
rep
ort
edfe
male
san
dp
ati
ents
wit
hn
on
-sm
ok
ing-r
elate
dco
nd
itio
ns)
�In
form
ati
on
bia
s:o
bje
ctiv
eo
utc
om
eev
alu
ati
on
:ye
s;st
an
dard
ized
exp
osu
rem
easu
rem
ent:
no
�C
on
fou
nd
ing:
adju
stm
entf
orag
e,ed
uca
tion
,typ
eofh
ou
se,
his
tory
of
sch
isto
som
iasi
s,h
igh
-ris
kocc
up
atio
n,
tob
acco
smok
ing
�P
art
icip
ati
on
rate
:4
95
%o
fel
igib
lesu
bje
cts
part
ici-
pate
din
the
inte
rvie
w
�7
.6(9
5%
CI
3.4
–1
6.8
)fo
r5
20
/d
ay
cigare
tte
smo
kin
g
Nasr
oll
ah
zad
eh1
7
�T
ype
of
des
ign
:ca
se–
con
tro
lst
ud
y�
Fu
nd
ing:
Dig
esti
veD
isea
seR
esea
rch
Cen
ter:
Teh
ran
Un
iver
sity
of
Med
ical
Sci
ence
s(D
DR
C/T
UM
S);
an
dN
ati
on
al
Can
cer
Inst
itu
teat
Nati
on
al
Inst
itu
teo
fH
ealt
h
�S
etti
ng
an
dp
erio
d:
Atr
ak
Cli
nic
inG
on
bad
Cit
y,ea
ster
nG
ole
stan
Pro
vin
ceo
fIr
an
,D
ecem
ber
20
03
toJu
ne
20
07
�C
ase
s:3
00
oes
op
hagea
lsq
uam
ou
sce
llca
rcin
om
aca
ses,
50
%m
ale
s,m
ean
age
64
.5ye
ars
�C
on
tro
ls:
57
1co
ntr
ols
,tw
op
op
ula
tio
n-b
ase
dm
atc
hed
con
tro
lsu
bje
cts
per
case
for
90
%o
fca
ses
�T
ype:
ho
ok
ah
,ci
gar-
ette
s,n
ass
chew
ing,
op
ium
�M
easu
rem
ent
too
l:se
lf-d
evel
op
edq
ues
-ti
on
nair
e,te
sted
for
reli
ab
ilit
yan
dva
lid
-it
y,cu
mu
lati
veco
n-
sum
pti
on
calc
ula
ted
as
ho
ok
ah
-yea
rsan
dca
tego
rize
din
ton
ever
use
rs,4
32
ho
ok
ah
-yea
rs,4
32
ho
ok
ah
-yea
rs�
Ex
po
sure
leve
lso
fin
clu
ded
sub
ject
s:m
edia
no
f3
2h
oo
kah
-yea
rs
Oes
op
hagea
lsq
uam
ou
sce
llca
rcin
om
ad
iagn
osi
s�
Mea
sure
men
tto
ol:
his
to-
path
olo
gic
all
yco
nfi
rmed
�B
lin
din
go
fo
utc
om
ead
ju-
dic
ato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
seri
eso
fin
cid
ent
case
s(7
0%
of
inci
-d
ent
case
sre
gis
tere
dw
ith
loca
lca
nce
rre
gis
try
wer
ere
ferr
edto
the
Atr
ak
Cli
nic
),co
ntr
ols
wer
ese
lect
edfr
om
the
sam
est
ud
yb
ase
as
case
s�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
eev
alu
ati
on
:ye
s;st
an
dard
ized
exp
osu
rem
easu
rem
ent:
yes
�C
on
fou
nd
ing:m
atch
ing
for
age,
sex,r
esid
ence
;ad
just
-m
ent
for
edu
cati
on
,eth
ni-
city
,oth
erty
pes
of
tob
acco
use
,tota
lin
tak
eof
fru
itan
dve
get
able
s�
Part
icip
ati
on
rate
:n
ot
rep
ort
ed
OR
com
pare
dw
ith
no
smo
kin
g:
�1
.85
(95
%C
I0
.95
–3
.58
)(h
oo
kah
smo
kin
g)
�1
.69
(95
%C
I0
.76
–3
.77
)(h
oo
kah
on
lysm
ok
ing)
�1
.50
(95
%C
I0
.92
–2
.43
)(c
igare
tte
on
lysm
ok
ing)
�T
est
for
tren
dsi
gn
if-
ican
tfo
rin
ten
sity
(P¼
0.0
3)
bu
tn
ot
for
du
rati
on
,to
tal
am
ou
nt,
or
age
start
ed)
Fen
g3
4
�S
tud
yd
esig
n:
case
–co
ntr
ol
stu
dy
�F
un
din
g:
Ass
oci
ati
on
for
Inte
rnati
on
al
Can
cer
Res
earc
h
�S
etti
ng
an
dp
erio
d:
five
ho
spit
als
inA
lger
ia,
Mo
rocc
o,
an
dT
un
isia
,Ja
nu
ary
20
02
toM
arc
h2
00
5�
Case
s:6
36
inci
den
tca
ses
of
naso
ph
ary
ngea
lca
nce
r
�T
ype:
cigare
tte,
shis
ha,
snu
ff,
can
nab
is�
Mea
sure
men
tto
ol:
inte
rvie
ws,
self
-d
evel
op
edq
ues
tio
n-
nair
e,n
ost
an
dard
i-za
tio
nre
po
rted
;
Naso
ph
ary
ngea
lca
rcin
om
a�
Mea
sure
men
tto
ol:
case
sid
enti
fied
by
clin
icia
nin
the
On
colo
gy
an
d
�S
elec
tio
nb
ias:
seri
eso
fh
osp
ital-
base
din
cid
ent
case
s(2
00
1–0
4),
con
tro
lsw
ere
ho
spit
al-
base
do
rfr
ien
ds
an
dfa
mil
yo
fca
ses,
an
dre
cru
ited
fro
mth
esa
me
stu
dy
base
as
case
s
OR
com
pare
dw
ith
nev
ersm
ok
ing:
0.4
9(9
5%
CI
0.2
0–1
.23
)(e
ver
shis
ha
smo
kin
g)
(co
nti
nu
ed)
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 841
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
1C
on
tin
ued
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
�C
on
tro
ls:
61
5co
ntr
ols
[pati
ents
ho
spit
ali
zed
for
no
n-c
an
cer
dis
ease
s(6
1%
)an
dfr
ien
ds
an
dfa
mil
yo
fn
on
-can
cer
pati
ents
(39
%)]
part
icip
an
tsca
tego
r-iz
edas
ever
smo
ker
san
dn
ever
smo
ker
s.A
ges
of
start
ing,
an
dq
uit
tin
gd
ail
yco
n-
sum
pti
on
extr
act
edfr
om
his
tory
bu
tn
ot
rep
ort
ed�
Ex
po
sure
leve
lso
fin
clu
ded
sub
ject
s:n
ot
rep
ort
ed
rad
ioth
erap
yd
epart
men
ts�
Bli
nd
ing
of
ou
tco
me
ad
ju-
dic
ato
r:n
ot
rep
ort
ed
�In
form
ati
on
bia
s:o
bje
ctiv
eo
utc
om
em
easu
rem
ent:
yes,
stan
dard
ized
exp
osu
rem
easu
rem
ent:
no
�C
on
fou
nd
ing:
matc
hin
gfo
rh
osp
ital,
age,
sex
,h
ou
se-
ho
ldty
pe
(urb
an
/ru
ral)
;ad
just
men
tfo
rage
�P
art
icip
ati
on
rate
:4
90
%
Ali
35
�S
tud
yd
esig
n:
cro
ss-
sect
ion
al
stu
dy
�F
un
din
g:
no
tre
po
rted
�S
etti
ng
an
dp
erio
d:
Yem
en;
per
iod
no
tre
po
rted
�P
op
ula
tio
n:
Yem
eni
volu
n-
teer
s(2
7m
enan
dsi
xw
om
en)
all
chew
ers
of
al-
qat,
mea
nage
38
.5ye
ars
(ran
ge
22
–5
8),
all
of
them
free
fro
man
ysy
stem
ati
cd
isea
ses.
11
wate
rpip
esm
ok
ers,
11
cigare
tte
smo
ker
s,1
1n
on
-sm
ok
ers
�T
ype:
cigare
tte
smo
kin
g,
wate
rpip
esm
ok
ing
�M
easu
rem
ent
too
l:in
terv
iew
s,st
an
dar-
diz
ati
on
no
tre
po
rted
,su
bje
cts
cate
go
rize
das
hea
vyci
gare
tte
smo
ker
s(4
20
/day)
,n
on
-sm
o-
ker
san
dw
ate
rpip
esm
ok
ers
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
cts:
no
tre
po
rted
Dys
pla
sia
of
ora
lm
uco
sad
iagn
osi
s�
Mea
sure
men
tto
ol:
his
to-
path
olo
gic
al
exam
inati
on
of
bu
ccal
mu
cosa
bio
psi
es�
Bli
nd
ing
of
ou
tco
me
ad
ju-
dic
ato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
volu
nte
erre
cru
itm
ent
into
the
stu
dy
�In
form
ati
on
bia
s:o
bje
ctiv
eo
utc
om
em
easu
rem
ent:
yes,
stan
dard
ized
exp
osu
rem
easu
rem
ent:
no
�C
on
fou
nd
ing:
no
matc
hin
go
rad
just
men
tin
the
an
a-
lysi
sre
po
rted
�P
art
icip
ati
on
rate
:n
ot
rep
ort
ed
OR
com
pare
dw
ith
no
smo
kin
g:
�8
.33
(95
%C
I0
.78
–9
.47
)(w
ate
rpip
esm
ok
ing)
Res
ult
sare
rest
rict
edto
the
chew
ing
sid
e;n
oev
ents
inei
ther
gro
up
on
no
n-c
hew
ing
sid
e.O
Rca
lcu
late
dfr
om
rep
ort
edn
um
ber
s
aN
inet
yfi
vep
erce
nt
ad
just
edO
Ran
dC
Ire
po
rted
as
5.7
(1.2
–7
.6).
As
the
effe
ctes
tim
ate
was
no
tco
nsi
sten
tw
ith
the
po
siti
on
of
the
low
eran
du
pp
erb
ord
ers
of
the
CI
an
dgiv
enth
eu
nad
just
edO
Rw
as
4.8
,an
das
we
wer
eu
nab
leto
con
tact
the
au
tho
rs,
we
use
dO
R¼
3.0
wh
ich
we
calc
ula
ted
base
do
nth
eC
Io
f1
.2–7
.6.
842 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
with dry socket was 3.7 (P¼ 0.001). We judged theoverall quality of evidence to be low (Appendix 3).
Infectious diseaseWe identified three studies on hepatitis C. The threecross-sectional studies were conducted in Egypt andrestricted to male participants exposed to groupwaterpipe tobacco smoking (Table 4).22,43,44 Thepooled OR for the association of group waterpipetobacco with hepatitis C was 0.98 (95% CI 0.80–1.21; I2
¼ 0%) (Figure 5). We judged the overall qual-ity of evidence to be very low (Appendix 3).
We did not identify any eligible study assessing theassociation between waterpipe smoking and thetransmission of tuberculosis. We did identify tworeports of outbreak investigations suggesting an asso-ciation between tuberculosis and sharing tobaccowaterpipe23 and marijuana waterpipe.47
Respiratory illnessOne cross-sectional study evaluated the associationbetween waterpipe smoking and respiratory illness(defined as perennial rhinitis and including nasalcongestion and wheezing) (Table 5).45 The studyreported significant association of waterpipe tobacco
smoking (OR¼ 2.3; 95% CI 1.1–5.1) and of waterpipeand/or cigarette smoking (OR¼ 2.5; 95% CI 1.6–3.8)with respiratory illness. We judged the overall qualityof evidence to be very low (Appendix 3).
InfertilityOne case–control study evaluated the associationbetween waterpipe smoking and male factor infertility(based on semen analysis) (Table 5).46 The reportedOR for the association of waterpipe tobacco smokingwith male factor infertility was not statistically signif-icant (OR¼ 2.5; 95% CI 1.0–6.3). We judged the over-all quality of evidence to be very low (Appendix 3).
DiscussionWe systematically reviewed the medical literature forthe effects of waterpipe tobacco smoking on healthoutcomes. Based on the available evidence, waterpipetobacco smoking was significantly associated withlung cancer, respiratory illness, low birth-weight andperiodontal disease. It was not significantly associatedwith bladder cancer, nasopharyngeal cancer, oesopha-geal cancer, oral dysplasia or infertility but the CIs did
Study or Subgroup
Gupta 2001
Hsairi 1993
Lubin 1992
Qiao 1989
Total (95% CI)
Heterogeneity: Tau² = 0.00; Chi ² = 0.74, df = 3 (P = 0.86); I ² = 0%
Test for overall effect: Z = 3.09 (P = 0.002)
log[OR]
0.662688
1.098612
0.587787
0.64185389
SE
0.422432
0.474253
0.432295
0.81574276
Weight
33.2%
26.3%
31.7%
8.9%
100.0%
IV, Random, 95% CI
1.94 (0.85, 4.44)
3.00 (1.18, 7.60)
1.80 (0.77, 4.20)
1.90 (0.38, 9.40)
2.12 (1.32, 3.42)
OR
IV, Random, 95% CI
0.01 0.1 1 10 100Protective effect
OR
Harmful effect
Figure 3 Association between waterpipe tobacco smoking and lung cancer
Study or Subgroup
Aghamolaei 2007
Nuwayhid 1998
Tamim 2008
Total (95% CI)
Heterogeneity: Tau ² = 0.00; Chi ² = 1.19, df = 2 (P = 0.55); I ² = 0%
Test for overall effect: Z = 2.17 (P = 0.03)
log[OR]
1.252763
0.774727
0.277632
SE
0.653538
0.546211
0.614272
Weight
28.1%
40.2%
31.8%
100.0%
IV, Random, 95% CI
3.50 (0.97, 12.60)
2.17 (0.74, 6.33)
1.32 (0.40, 4.40)
2.12 (1.08, 4.18)
OR
IV, Random, 95% CI
0.01 0.1 1 10 100Protective effect
OR
Harmful effect
Figure 4 Association between waterpipe tobacco smoking and low birth-weight
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 843
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
2C
hara
cter
isti
cso
fin
clu
ded
stu
die
sass
essi
ng
the
effe
cto
fw
ate
rpip
eto
bacc
osm
ok
ing
on
pre
gn
an
cyo
utc
om
es
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
Nu
wayh
id1
9
�S
tud
yd
esig
n:
retr
o-
spec
tive
coh
ort
stu
dy
�F
un
din
g:
Hari
riF
ou
nd
ati
on
;an
dU
niv
ersi
tyR
esea
rch
Bo
ard
at
the
Am
eric
an
Un
iver
sity
of
Bei
rut
�S
etti
ng
an
dp
erio
d:
ho
spit
als
inL
eban
on
;1
99
3an
d1
99
5�
Part
icip
an
ts:
pre
g-
nan
tw
om
end
eli-
veri
ng
inh
osp
itals
.1
06
narg
hil
esm
o-
ker
s(m
ean
age
27
.6ye
ars
),2
77
cigare
tte
smo
ker
s(m
ean
age
29
.1)
an
d5
12
no
n-
smo
ker
s(m
ean
age
28
.5)
�T
ype
of
exp
osu
re:
narg
hil
e,ci
gare
ttes
�M
easu
rem
ent
too
l:in
terv
iew
s,st
an
dar-
diz
ati
on
no
tre
po
rted
;ca
tego
rize
din
to<
1vs
51
per
day;
an
dst
art
edsm
ok
ing
du
rin
gfi
rst
vsse
con
do
rth
ird
trim
este
r�
Mea
sure
men
tti
me
po
ints
:n
ot
rep
ort
ed�
Ex
po
sure
leve
lso
fin
clu
ded
sub
ject
s:<
1p
erd
ay
(n¼
38
);5
1p
erd
ay
(n¼
67
)p
erd
ay;
firs
ttr
ime-
ster
(n¼
78
);se
con
do
rth
ird
trim
este
r(n¼
25
)
Lo
wb
irth
-wei
gh
t,A
pgar
sco
re<
7(1
an
d5
min
),p
ulm
on
ary
pro
ble
ms,
malf
orm
ati
on
s,p
eri-
nata
lco
mp
lica
tio
ns
�M
easu
rem
ent
too
l:m
edic
al
reco
rds;
low
bir
th-w
eigh
td
efin
edas
ab
irth
wei
gh
t<
25
00
g�
Bli
nd
ing
of
ou
tco
me
ad
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
un
clea
rw
het
her
sam
ple
isre
pre
sen
tati
ve(d
ata
coll
ecti
on
ina
few
nu
mb
ero
fh
osp
itals
)�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
sb
ut
base
do
nm
edic
al
reco
rds,
stan
dard
ized
exp
osu
rem
easu
rem
ent:
no
�C
on
fou
nd
ing:
ad
just
-m
ent
for
age,
ho
spit
al
loca
tio
n,
ges
tati
on
al
age,
exp
osu
reto
pass
ive
smo
kin
g,
an
dp
revi
ou
slo
wb
irth
-wei
gh
t�
Part
icip
ati
on
rate
:n
ot
rep
ort
ed�
<1
0%
mis
sin
gd
ata
OR
for
narg
hil
esm
ok
ing
com
pare
dw
ith
no
smo
kin
g�
low
bir
th-w
eigh
t¼
2.1
7(9
5%
CI
0.7
4–6
.33
)�
low
bir
th-w
eigh
t¼
2.3
6(9
5%
CI
0.5
2–1
0.7
3)
(<1
/day)
�lo
wb
irth
-wei
gh
t¼
2.0
7(9
5%
CI
0.5
4–7
.94
)(5
1/d
ay)
�lo
wb
irth
-wei
gh
t¼
2.9
3(9
5%
CI
0.9
7–8
.83
)(n
arg
hil
esm
ok
ing
start
edin
firs
ttr
imes
ter)
�T
est
for
tren
dfo
rlo
wb
irth
-wei
gh
tn
ot
sign
ific
an
tfo
rin
ten
sity
(P<
0.1
8)
�A
pgar
sco
re<
7at
1m
in¼
1.7
3(9
5%
CI
0.7
3–4
.14
)�
Ap
gar
sco
re<
7at
5m
in¼
3.3
9(9
5%
CI
0.5
4–2
1.4
2)
�P
ulm
on
ary
pro
ble
ms¼
3.6
5(9
5%
CI
1.5
2–8
.75
)�
Malf
orm
ati
on
s¼
2.0
1(9
5%
CI
0.5
9–6
.88
)�
Per
inata
lco
mp
lica
tio
ns¼
1.6
7(9
5%
CI
0.8
2–3
.41
)O
Rfo
rci
gare
tte
smo
kin
gco
mp
are
dw
ith
no
smo
kin
g�
Lo
wb
irth
-wei
gh
t¼
2.0
0(9
5%
CI
0.9
6–4
.20
)�
low
bir
th-w
eigh
t¼
2.2
5(9
5%
CI
1.0
4–4
.86
)(n
arg
hil
esm
ok
ing
start
edin
firs
ttr
imes
ter)
�A
pgar
sco
re<
7at
1m
in¼
1.5
9(9
5%
CI
0.8
2–3
.07
)�
Ap
gar
sco
re<
7at
5m
in¼
2.6
2(9
5%
CI
0.5
6–1
2.2
9)
�P
ulm
on
ary
pro
ble
ms¼
1.7
6(0
.80
–3
.87
)�
Malf
orm
ati
on
s¼
1.3
6(0
.49
–3
.80
)�
Per
inata
lco
mp
lica
tio
ns¼
1.5
0(0
.87
–2
.57
)
(co
nti
nu
ed)
844 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
2C
on
tin
ued
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
Agh
am
ola
ei3
7
�S
tud
yd
esig
n:
case
–co
ntr
ol
stu
dy
�F
un
din
g:
no
tre
po
rted
�S
etti
ng
an
dp
erio
d:
Sh
ari
ati
Ho
spit
al
of
Ban
dar
Ab
bas,
So
uth
of
Iran
;p
erio
dn
ot
rep
ort
ed�
Case
s:6
0In
tra-u
ter-
ine
gro
wth
reta
rda-
tio
nin
fan
ts�
Co
ntr
ols
:6
0n
orm
al
bir
thin
fan
ts�
Incl
ud
edsu
bje
cts:
all
term
infa
nts
wit
hges
tati
on
al
age
37
–4
2w
eek
s;2
9m
ale
san
d3
1fe
male
sin
each
gro
up
.‘N
on
eo
fin
clu
ded
mo
ther
ssm
ok
edan
du
sed
alc
oh
ol
du
rin
gp
reg-
nan
cyan
dn
on
eo
fth
emh
ad
dia
bet
es’
�T
ype:
ho
ok
ah
smo
k-
ing
by
mo
ther
�M
easu
rem
ent
too
l:st
ruct
ure
dq
ues
tio
n-
nair
ead
min
iste
red
tom
oth
ers,
stan
dar-
diz
ati
on
no
tre
po
rted
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
cts:
no
tre
po
rted
Intr
au
teri
ne
gro
wth
reta
rdati
on
(IU
GR
)�
Def
init
ion
:te
rmin
fan
tsw
ith
ab
irth
-w
eigh
t<
25
00
g�
Mea
sure
men
tto
ol:
bir
thw
eigh
td
eter
-m
ined
up
to1
0m
ino
fd
eliv
ery
usi
ng
ad
igit
al
bab
ysc
ale
�B
lin
din
go
fo
utc
om
ead
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
seri
eso
fin
cid
ent
case
so
fIU
GR
inm
ate
rnit
yw
ard
s,co
ntr
ols
sele
cted
fro
mth
esa
me
stu
dy
base
as
the
case
s�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
s,st
an
dar-
diz
edex
po
sure
mea
sure
men
t:n
o�
Co
nfo
un
din
g:
exp
lore
d1
8p
ote
nti
al
risk
fact
ors
an
dad
just
edfo
rm
ate
rnal
wo
rk,
hyp
er-
ten
sio
n,
an
ten
ata
lca
re�
Part
icip
ati
on
rate
:n
ot
rep
ort
ed
�O
Rfo
rh
oo
kah
smo
kin
gd
uri
ng
pre
gn
an
cyco
mp
are
dw
ith
no
ho
ok
ah
smo
kin
gd
uri
ng
pre
gn
an
cy¼
3.5
(95
%C
I1
.1–1
2.6
)
Tam
im3
6
�S
tud
yd
esig
n:
retr
osp
ecti
veco
ho
rtst
ud
y�
Fu
nd
ing:
Leb
an
ese
Nati
on
al
Co
un
cil
for
Sci
enti
fic
Res
earc
h;
the
Med
ical
Pra
ctic
eP
lan
;an
dth
eU
niv
ersi
tyR
esea
rch
Bo
ard
at
the
Am
eric
an
Un
iver
sity
of
Bei
rut
�S
etti
ng
an
dp
erio
d:
six
majo
rh
osp
itals
ingre
ate
rB
eiru
t,L
eban
on
,A
ugu
st2
00
0to
Au
gu
st2
00
3�
Po
pu
lati
on
:3
78
sin
-gle
ton
new
bo
rns
tow
om
enex
clu
sive
narg
hil
esm
ok
ers,
92
9si
ngle
ton
new
-b
orn
sto
wo
men
excl
usi
veci
gare
tte
smo
ker
s,8
4si
ngle
-to
nn
ewb
orn
sto
wo
men
smo
kin
gb
oth
typ
eso
fto
bacc
o,
72
01
sin
-gle
ton
new
bo
rns
tow
om
enn
on
-sm
ok
ers
�T
ype:
narg
hil
e,ci
gare
ttes
�M
easu
rem
ent
too
l:in
terv
iew
s,q
ues
tio
n-
nair
ep
ilo
tte
sted
;n
arg
hil
esm
ok
ers
cate
go
rize
din
to4
1vs
41
per
day;
cate
-go
rize
din
to<
1vs
51
per
day;
an
dst
art
edsm
ok
ing
du
rin
gfi
rst
vsse
con
do
rth
ird
trim
este
r�
Mea
sure
men
tti
me
po
ints
:n
ot
rep
ort
ed�
Ex
po
sure
leve
lso
fin
clu
ded
sub
ject
s:4
1p
erd
ay
(n¼
23
3);4
1p
erd
ay
(n¼
14
5);
firs
ttr
imes
ter
(n¼
30
9);
seco
nd
or
thir
dtr
i-m
este
r(n¼
69
)
Lo
wb
irth
-wei
gh
t�
Mea
sure
men
tto
ol:
data
coll
ecti
on
fro
mo
bst
etri
can
dn
urs
ery
chart
s,lo
wb
irth
-w
eigh
td
efin
edas
42
50
0g
�B
lin
din
go
fo
utc
om
ead
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
sam
ple
rep
rese
nta
tive
of
po
pu
-la
tio
ns
of
six
majo
rh
osp
itals
inG
reate
rB
eiru
t�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
sb
ut
base
do
nm
edic
al
reco
rds,
stan
dard
ized
exp
osu
rem
easu
rem
ent:
no
(pil
ot
test
edo
nly
)�
Co
nfo
un
din
g:
ad
just
-m
ent
for
mate
rnal/
pate
rnal
edu
cati
on
,m
oth
er’s
wo
rkin
gst
atu
s,m
oth
er’s
age,
pari
ty,
dia
bet
es,
ble
ed-
ing,
pre
gn
an
cyh
yper
-te
nsi
ved
iso
rder
s,ges
tati
on
al
age,
an
dp
ass
ive
smo
kin
g�
Part
icip
ati
on
rate
:n
ot
rep
ort
ed
OR
for
narg
hil
eo
nly
smo
kin
gco
mp
are
dw
ith
no
smo
kin
g�
low
bir
th-w
eigh
t¼
1.3
2(9
5%
CI
0.3
9–4
.40
)(O
Rp
oo
led
for
41
day
an
d4
1d
ay)
�lo
wb
irth
-wei
gh
t¼
0.7
(95
%C
I0.3
–1.6
)(4
1d
ay)
�lo
wb
irth
-wei
gh
t¼
2.4
(95
%C
I1.2
–5.0
)(4
1d
ay)
�lo
wb
irth
-wei
gh
t¼
1.2
(95
%C
I0
.6–2
.2)
(narg
hil
esm
ok
ing
start
edin
firs
ttr
imes
ter)
�O
Rfo
rci
gare
tte
on
lysm
ok
ing
com
pare
dw
ith
no
smo
kin
g¼
1.3
2(9
5%
CI
0.9
1–1
.90
)
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 845
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
3C
hara
cter
isti
cso
fin
clu
ded
stu
die
sass
essi
ng
the
effe
cto
fw
ate
rpip
eto
bacc
osm
ok
ing
on
per
iod
on
tal
dis
ease
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
Natt
o3
9
�S
tud
yd
esig
n:
cro
ss-
sect
ion
al
stu
dy
�F
un
din
g:
Min
istr
yof
Hea
lth
inS
au
di
Ara
bia
;an
dS
au
di
Ara
bia
nC
ult
ura
lB
ure
au
Off
ice
inB
on
n
�S
etti
ng
an
dp
erio
d:
Jed
dah
,S
au
di
Ara
bia
,p
erio
dn
ot
rep
ort
ed�
Part
icip
an
ts:
35
5vo
lun
-te
ers,
28
%fe
male
s,m
ean
age
36
.9ye
ars
(ran
ge
17
–6
0),
havi
ng
20
teet
han
dn
ot
pre
gn
an
t
�T
ype:
wate
rpip
e,ci
gare
ttes
�M
easu
rem
ent
too
l:st
an
dard
ized
qu
esti
on
-n
air
ead
min
iste
red
inin
terv
iew
s,st
an
dard
iza-
tio
nn
ot
rep
ort
ed;
cal-
cula
ted
as
run
-yea
rs(p
rod
uct
of
wate
rpip
eru
ns
per
day
wit
hye
ars
of
smo
kin
g;
aru
nis
the
com
ple
tio
no
fth
ew
ate
rpip
esm
ok
ing
un
til
the
tob
acc
ois
bu
rnt)
;h
eavy
exp
osu
rew
as
def
ined
as5
40
run
-yea
rs�
Ex
po
sure
leve
lso
fin
clu
ded
sub
ject
:m
ean
exp
osu
reo
f5
6.8
run
-ye
ars
for
wate
rpip
esm
ok
ers
Per
iod
on
tal
dis
ease
mea
sure
das
per
iod
-o
nta
lb
on
eh
eigh
tlo
ss�
Mea
sure
men
tto
ol:
dig
ital
pan
ora
mic
rad
iogra
ph
s.B
on
elo
ssd
efin
edas
bo
ne
hei
gh
t4
70
%.
Mea
nb
on
eh
eigh
tp
erin
div
idu
al
isth
era
tio
of
the
per
iod
-o
nta
lb
on
eh
eigh
tto
the
roo
tle
ngth
�B
lin
din
go
fo
utc
om
ead
jud
icato
r:ye
s
�S
elec
tio
nb
ias:
volu
n-
teer
sre
cru
ited
by
new
spap
eran
no
un
cem
ents
�In
form
ati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
s,st
an
dar-
diz
edex
po
sure
mea
sure
men
t:n
o�
Co
nfo
un
din
g:
ad
just
-m
ent
for
age
on
ly�
Part
icip
ati
on
rate
:n
ot
rep
ort
ed
OR
com
pare
dw
ith
no
smo
kin
g�
3.5
(95%
CI
1.6
–7.6
)(w
ate
rpip
eon
lysm
ok
ing)
�1.0
(95%
CI
0.3
–3.1
)(l
igh
tw
ate
rpip
eo
nly
smo
kin
g)
�7
.5(9
5%
CI
3.0
–1
8.3
)(h
eavy
wate
r-p
ipe
on
lysm
ok
ing)
�T
he
ass
oci
ati
on
bet
wee
nw
ate
rpip
esm
ok
ing
an
db
on
eh
eigh
tre
main
edst
a-
tist
icall
ysi
gn
ific
an
taft
erco
ntr
oll
ing
for
‘ed
uca
tion
as
asu
r-ro
gate
for
soci
o-e
co-
no
mic
stan
dard
an
do
ther
vari
ab
les’
�D
iffe
ren
ces
bet
wee
nli
gh
tan
dh
eavy
exp
osu
res
wer
est
a-
tist
icall
ysi
gn
ific
an
t(P
<0
.00
1)
Natt
o4
0
�S
tud
yd
esig
n:
cro
ss-
sect
ion
al
stu
dy
�F
un
din
g:
no
tre
po
rted
�S
etti
ng
an
dp
erio
d:
Wes
tern
part
of
Sau
di
Ara
bia
,p
erio
dn
ot
rep
ort
ed�
Part
icip
an
ts:
24
4vo
lun
-te
ers,
34
%fe
male
s,m
ean
age
37
.4ye
ars
(ran
ge
25
–7
0),
havi
ng
20
teet
han
dn
ot
pre
gn
an
t
�T
ype:
wate
rpip
e,ci
gare
ttes
�M
easu
rem
ent
too
l:st
an
dard
ized
qu
esti
on
-n
air
ead
min
iste
red
inin
terv
iew
s,st
an
dard
iza-
tio
nn
ot
rep
ort
ed;
cal-
cula
ted
as
run
-yea
rs(p
rod
uct
of
wate
rpip
eru
ns
per
day
wit
hye
ars
of
smo
kin
g;
aru
nis
the
com
ple
tio
no
fth
ew
ate
rpip
esm
ok
ing
un
til
the
tob
acc
ois
bu
rnt)
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
ct:
no
tre
po
rted
Per
iod
on
tal
dis
ease
mea
sure
das
pla
qu
ein
dex
an
dgin
giv
itis
�M
easu
rem
ent
too
l:P
laq
ue:
clin
ical
exam
inati
on
,p
re-
sen
ceo
fvi
sib
led
enta
lp
laq
ue
was
reco
rded
acc
ord
ing
toth
ecr
iter
iao
fS
iln
ess
an
dL
oe
�B
lin
din
go
fo
utc
om
ead
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
24
4o
fth
e3
55
volu
nte
ers
par-
tici
pati
ng
inan
oth
erst
ud
y38
volu
nte
ered
for
acl
inic
al
exam
inati
on
�In
form
ati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
s,st
an
dar-
diz
edex
po
sure
mea
sure
men
t:n
o�
Co
nfo
un
din
g:
ad
just
-m
ent
for
age,
den
tal
care
hab
it�
Part
icip
ati
on
rate
:6
8%
�T
her
ew
as
an
ove
rall
sign
ific
an
tass
oci
a-
tio
nb
etw
een
smo
k-
ing
an
dp
laq
ue
ind
exan
dgin
giv
al
ind
ex.
No
effe
ctes
tim
ate
sre
po
rted
(co
nti
nu
ed)
846 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
3C
on
tin
ued
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
Natt
o4
1
�S
tud
yd
esig
n:
cro
ss-
sect
ion
al
stu
dy
�F
un
din
g:
Min
istr
yo
fH
ealt
hin
Sau
di
Ara
bia
;an
dS
au
di
Ara
bia
nC
ult
ura
lB
ure
au
Off
ice
inB
on
n
�S
etti
ng
an
dp
erio
d:
Jed
dah
,S
au
di
Ara
bia
,p
erio
dn
ot
rep
ort
ed�
Po
pu
lati
on
:2
62
volu
nte
ers,
35
%fe
male
s,m
ean
age
36
.5(r
an
ge
17
–6
0).
Part
icip
an
tsw
ere
req
uir
edto
have
20
teet
han
dn
ot
be
pre
gn
an
t.5
1ci
tize
ns
wit
hp
erio
do
nta
ld
isea
sean
d2
11
citi
zen
sw
ith
ou
tp
erio
do
nta
ld
isea
se
�T
ype:
wate
rpip
e,ci
gare
ttes
�M
easu
rem
ent
too
l:st
an
dard
ized
qu
esti
on
-n
air
ead
min
iste
red
inin
terv
iew
s,st
an
dard
iza-
tio
nn
ot
rep
ort
ed;
cal-
cula
ted
as
run
-yea
rs(p
rod
uct
of
wate
rpip
eru
ns
per
day
wit
hye
ars
of
smo
kin
g;
aru
nis
the
com
ple
tio
no
fth
ew
ate
rpip
esm
ok
ing
un
til
the
tob
acc
ois
bu
rnt)
;h
eavy
exp
osu
rew
as
def
ined
as5
40
run
-yea
rs�
Ex
po
sure
leve
lso
fin
clu
ded
sub
ject
:m
ean
exp
osu
reo
f3
5.8
run
-ye
ars
for
wate
rpip
esm
ok
ers
Per
iod
on
tal
dis
ease
mea
sure
das
dee
pen
-in
go
fth
esu
lci
or
po
cket
s�
Mea
sure
men
tto
ol:
clin
ical
exam
inati
on
pro
bin
gth
ed
epth
of
the
sulc
io
rp
ock
ets
wit
ha
2-m
mgra
du
-ate
dp
erio
do
nta
lp
rob
e;p
erio
do
nta
ld
isea
sed
efin
edas
51
0si
tes
wit
ha
pro
bin
gd
epth
55
mm
�B
lin
din
go
fo
utc
om
ead
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
26
2o
fth
e3
55
volu
nte
ers
par-
tici
pati
ng
inan
oth
erst
ud
y18
volu
nte
ered
for
acl
inic
al
exam
inati
on
�In
form
ati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
s,st
an
dar-
diz
edex
po
sure
mea
sure
men
t:n
o�
Co
nfo
un
din
g:
ad
just
-m
ent
for
age
on
ly�
Part
icip
ati
on
rate
:7
4%
OR
com
pare
dw
ith
no
smo
kin
g�
5.1
(95
%C
I2
.0–1
3.5
(wate
rpip
eo
nly
smo
kin
g)
�8
.2(9
5%
CI
2.9
–2
2.9
)(h
eavy
wate
r-p
ipe
smo
kin
g)
�2
.9(9
5%
CI
1.7
–4
.8)
(lig
ht
wate
rpip
esm
ok
ing)
�3
.8(9
5%
CI
1.3
–1
1.0
)(c
igare
tte
on
lysm
ok
ing)
�5
.0(9
5%
CI
1.6
–1
6.2
),P
<0
.00
1(h
eavy
cigare
tte
on
lysm
ok
ing)
�1
.5(9
5%
CI
1.1
–2
.0),
P<
0.0
01
(lig
ht
cigare
tte
on
lysm
ok
ing)
�D
iffe
ren
ces
bet
wee
nli
gh
tan
dh
eavy
exp
osu
rew
ere
sta-
tist
icall
ysi
gn
ific
an
t
Balj
oo
n4
2
�S
tud
yd
esig
n:
cro
ss-
sect
ion
al
stu
dy
�F
un
din
g:
Min
istr
yo
fH
ealt
hin
Sau
di
Ara
bia
;an
dS
au
di
Ara
bia
nC
ult
ura
lB
ure
au
Off
ice
inB
on
n
�S
etti
ng
an
dp
erio
d:
Jed
dah
,S
au
di
Ara
bia
,p
erio
dn
ot
rep
ort
ed�
Po
pu
lati
on
:1
03
volu
nte
ers
wit
hve
rtic
al
per
iod
on
tal
bo
ne
loss
an
d1
59
ind
ivid
-u
als
wit
ho
ut
vert
ical
per
i-o
do
nta
lb
on
elo
ss
�T
ype:
wate
rpip
e,ci
gare
ttes
�M
easu
rem
ent
too
l:st
an
dard
ized
qu
esti
on
-n
air
ead
min
iste
red
inin
terv
iew
s,st
an
dard
iza-
tio
nn
ot
rep
ort
ed;
cal-
cula
ted
as
run
-yea
rs(p
rod
uct
of
wate
rpip
eru
ns
per
day
wit
hye
ars
of
smo
kin
g;
aru
nis
the
com
ple
tio
no
fth
ew
ate
rpip
esm
ok
ing
un
til
the
tob
acc
ois
bu
rnt)
;h
eavy
exp
osu
rew
as
def
ined
as5
40
run
-yea
rs
Per
iod
on
tal
dis
ease
mea
sure
das
vert
ical
per
iod
on
tal
bo
ne
loss
�M
easu
rem
ent
too
l:in
tra-o
ral
rad
io-
gra
ph
s;ve
rtic
al
per
i-o
do
nta
lb
on
elo
ssd
efin
edas
on
e-si
ded
bo
ne
reso
rpti
on
of
the
inte
rden
tal
mar-
gin
al
bo
ne5
2m
mth
at
had
aty
pic
al
an
gu
lati
on
tow
ard
sei
ther
the
mes
ial
or
dis
tal
asp
ect
of
the
roo
t�
Bli
nd
ing
of
ou
tco
me
ad
jud
icato
r:ye
s
�S
elec
tio
nb
ias:
26
2o
fth
e3
55
volu
nte
ers
par-
tici
pati
ng
inan
oth
erst
ud
y,5
1,5
2vo
lun
teer
edto
part
icip
ate
inth
isst
ud
y�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
s,st
an
dar-
diz
edex
po
sure
mea
sure
men
t:n
o�
Co
nfo
un
din
g:
ad
just
-m
ent
for
age
on
ly�
Part
icip
ati
on
rate
:7
4%
OR
com
pare
dw
ith
no
smo
kin
g�
2.9
(95
%C
I1
.2–7
.0)
(wate
rpip
eo
nly
smo
kin
g)
�4
3.3
(95
%C
I1
2.1
–7
1.6
)(h
eavy
wate
r-p
ipe
smo
kin
g)
�0
.6(9
5%
CI
0.3
–1
.4)
(lig
ht
wate
rpip
esm
ok
ing)
�6
.6(9
5%
CI
2.6
–1
7.1
)(c
igare
tte
on
lysm
ok
ing)
�1
8.3
(95
%C
I6
.2–
53
.9),
P<
0.0
01
(hea
vyci
gare
tte
on
lysm
ok
ing) (c
on
tin
ued
)
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 847
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
3C
on
tin
ued
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
ct:
mea
nex
po
sure
of
56
.8ru
n-
years
for
wate
rpip
esm
ok
ers
�1
.3(9
5%
CI
0.5
–3
.4),
P<
0.0
01
(lig
ht
cigare
tte
on
lysm
ok
ing)
�D
iffe
ren
ces
bet
wee
nli
gh
tan
dh
eavy
exp
osu
res
wer
est
a-
tist
icall
ysi
gn
ific
an
t(P
<0
.00
1)
Al-
Bel
asy
38
�S
tud
yd
esig
n:
coh
ort
stu
dy
wit
h7
days
foll
ow
-up
aft
ersu
r-gic
al
rem
ova
lo
fm
an
dib
ula
rth
ird
mo
lars
�F
un
din
g:
no
tre
po
rted
�S
etti
ng
an
dp
erio
d:
Man
sou
raE
gyp
t,Ja
nu
ary
20
00
toF
ebru
ary
20
02
�E
xp
ose
d:
10
0m
ale
shis
ha
smo
ker
s,m
ean
age
28
.7ye
ars
(ran
ge
22
–3
9),
10
0m
ale
cigare
tte
smo
ker
s,m
ean
age
28
.7ye
ars
(ran
ge
20
–3
8)
�N
on
-ex
po
sed
:1
00
male
sn
on
-sm
ok
ers,
mea
nage
27
.7ye
ars
(ran
ge
20
–3
7)
�In
clu
ded
sub
ject
sw
ere
req
uir
edto
be
hea
lth
yn
ot
tak
ing
med
icati
on
sat
the
tim
eo
fth
est
ud
yan
dw
ith
un
ilate
ral
hig
hm
eso
an
gu
-la
rim
pact
ion
sw
ith
exp
ose
do
cclu
sal
surf
ace
s.P
ati
ents
wit
hre
cen
tan
ti-
bio
tic
use
or
the
med
ial
nee
dfo
ran
tib
ioti
cp
ro-
ph
ylax
isw
ere
excl
ud
ed
�T
ype:
shis
ha,
cigare
ttes
�M
easu
rem
ent
too
l:se
lf-
rep
ort
,st
an
dard
izati
on
no
tre
po
rted
�M
easu
rem
ent
tim
ep
oin
ts:
on
day
of
the
surg
ery,
on
the
firs
tan
dse
con
dp
ost
-op
erati
ved
ays
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
cts:
no
tre
po
rted
Dry
sock
et�
Mea
sure
men
tto
ol:
clin
ical
dia
gn
osi
so
nth
eb
asi
so
fco
nst
an
tra
dia
tin
gp
ain
no
tre
liev
edb
yth
ean
alg
esic
,acc
om
pa-
nie
db
ya
den
ud
edso
cket
or
nec
roti
ccl
ot
an
da
feti
dsm
ell
�B
lin
din
go
fo
utc
om
ead
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
on
lyh
ealt
hy
pati
ents
no
tta
kin
gm
edic
ati
on
sw
ere
incl
ud
ed,
no
n-
smo
ker
sse
lect
edfr
om
the
sam
eco
mm
un
ity
as
smo
ker
s�
Info
rmati
on
bia
s:ob
ject
ive
ou
tcom
em
ea-
sure
men
t:ye
s,st
an
dar-
diz
edex
po
sure
mea
sure
men
t:n
o;
sam
ed
ura
tio
nfo
llo
w-u
pfo
rb
oth
gro
up
s�
Con
fou
nd
ing:
no
matc
hin
gor
ad
just
men
tin
the
an
aly
sis
rep
ort
ed�
Part
icip
ati
on
rate
:n
ot
rep
ort
ed�
Att
riti
on
rate
:n
ot
rep
ort
ed
�‘S
his
ha
smo
ker
sh
ad
thre
eti
mes
the
risk
of
no
n-s
mo
ker
sfo
rd
evel
op
ing
dry
sock
et’
(RR¼
3.7
;P¼
0.0
01
);tr
end
of
dev
elo
pin
gd
ryso
cket
wit
hfr
e-q
uen
cyo
fsh
ish
asm
ok
ing
was
stati
s-ti
call
ysi
gn
ific
an
t(P¼
0.0
01
)�
Co
mp
are
dw
ith
no
n-
smo
ker
s,sh
ish
asm
ok
ers
wh
osm
ok
edth
ed
ay
of
surg
ery
or
the
firs
td
ay
aft
ersu
rger
yh
ad
asi
gn
ific
an
tly
incr
ease
din
cid
ence
of
dry
sock
et(d
ay
0,
P¼
0.0
01
;d
ay
1,
P¼
0.0
05
)
848 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
4C
hara
cter
isti
cso
fin
clu
ded
stu
die
sass
essi
ng
the
effe
cto
fw
ate
rpip
eto
bacc
osm
ok
ing
on
hep
ati
tis
Cin
fect
ion
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
Hab
ib2
2
�S
tud
yd
esig
n:
cro
ss-
sect
ion
al
stu
dy
�F
un
din
g:
Hep
ati
tis
CP
reve
nti
on
Pro
ject
(US
AID
gra
nt)
�S
etti
ng
an
dp
erio
d:
Agh
ou
rE
lS
ou
gh
ra,
aru
ral
vill
age
inN
ile
Del
ta,
in1
99
7�
Po
pu
lati
on
:m
ale
vill
age
inh
ab
itan
tssc
reen
edfo
rH
epati
tis
C.
Fo
ur-
hu
nd
red
an
dfi
fty-
five
test
edp
osi
tive
for
Hep
ati
tis
Can
d1
37
2te
sted
neg
ati
vefo
rH
epati
tis
C(s
tud
yals
ore
cru
ited
fem
ale
sb
ut
rest
rict
edgo
zaan
a-
lysi
sto
male
s)
�T
ype:
Gro
up
go
zasm
ok
ing
(ass
esse
din
men
on
ly)
�M
easu
rem
ent
too
l:q
ues
tio
nn
air
e,st
an
-d
ard
izati
on
no
tre
po
rted
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
cts:
no
tre
po
rted
Hep
ati
tis
Cin
fect
ion
�M
easu
rem
ent
too
l:se
con
dgen
erati
on
En
zym
eIm
mu
no
ass
ay
(Ab
bo
ttH
CV
EIA
2.0
)�
Bli
nd
ing
of
ou
tco
me
ad
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
sam
ple
was
‘sys
tem
ati
call
yse
lect
ed’
�In
form
ati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
s,st
an
dar-
diz
edex
po
sure
mea
sure
men
t:n
o�
Co
nfo
un
din
g:
ad
just
-m
ent
for
age,
sex
,m
ar-
ital
statu
s,ed
uca
tio
n,
his
tory
of
inva
sive
med
ical
pro
ced
ure
san
dd
enta
lp
roce
du
res
�P
art
icip
ati
on
rate
:h
alf
of
ho
use
ho
lds
sele
cted
�0
.15
%m
issi
ng
data
�T
wo
sub
ject
s(g
oza
smo
ker
s)ex
clu
ded
for
age
<2
0ye
ars
�It
isn
ot
clea
rh
ow
fem
ale
sub
ject
sw
ere
dea
ltw
ith
inth
ere
gre
ssio
nan
aly
ses
�O
Rfo
rgro
up
go
zasm
ok
ing
com
pare
dw
ith
no
n-g
rou
pgo
zasm
ok
ing¼
1.1
95
%C
I(0
.7–1
.5)
(male
s5
20
years
old
)
Med
hat4
4
�S
tud
yd
esig
n:
cro
ss-
sect
ion
al
stu
dy
�F
un
din
g:
Hep
ati
tis
CP
reve
nti
on
Pro
ject
[US
AID
gra
nt]
�S
etti
ng
an
dp
erio
d:
Co
mm
un
ity
inU
pp
erE
gyp
t,p
erio
dn
ot
rep
ort
ed.
�P
op
ula
tio
n:
male
vill
age
inh
ab
itan
tssc
reen
edfo
rH
epati
tis
C.
30
8te
sted
po
siti
vefo
rH
epati
tis
Can
d2
40
9te
sted
neg
ati
vefo
rH
epati
tis
C(s
tud
yals
ore
cru
ited
fem
ale
sb
ut
rest
rict
edgo
zaan
a-
lysi
sto
male
s)
�T
ype:
gro
up
go
zasm
ok
ing
(ass
esse
din
men
on
ly)
�M
easu
rem
ent
too
l:q
ues
tio
nn
air
e,st
an
-d
ard
izati
on
no
tre
po
rted
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
cts:
no
tre
po
rted
Hep
ati
tis
Cin
fect
ion
�M
easu
rem
ent
too
l:se
con
dgen
erati
on
En
zym
eIm
mu
no
ass
ay
(Ab
bo
ttH
CV
EIA
2.0
).O
fth
e5
23
part
icip
an
ts,
51
4w
ith
po
siti
vean
ti-
HC
Vte
stre
sult
sw
ere
test
edfo
rH
CV
RN
Ab
ya
on
e-st
epre
vers
etr
an
scri
pta
se-
po
lym
erase
chain
react
ion
met
ho
d�
Bli
nd
ing
of
ou
tco
me
ad
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
sam
ple
rep
rese
nts
62
.8%
of
vil-
lage
inh
ab
itan
ts�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
s,st
an
dar-
diz
edex
po
sure
mea
sure
men
t:n
o�
Co
nfo
un
din
g:
ad
just
-m
ent
for
age,
inje
ctio
nh
isto
ry,
ho
spit
al
or
clin
ical
exp
erie
nce
s,d
enta
ltr
eatm
ents
,o
bst
etri
cex
po
sure
s(w
om
en5
20
years
),sh
ari
ng
razo
rsw
ith
infa
mil
y(m
ale
so
nly
),ci
rcu
mci
sio
n,
shavi
ng
at
com
mu
nit
yb
arb
er(m
eno
nly
),ea
rp
ierc
ing
�O
Rfo
rgro
up
go
zasm
ok
ing
com
pare
dw
ith
no
n-g
rou
pgo
zasm
ok
ing¼
0.9
(0.4
–2
.0)
(male
s4
30
years
old
)�
OR
for
gro
up
go
zasm
ok
ing
com
pare
dw
ith
no
n-g
rou
pgo
zasm
ok
ing¼
0.8
(0.5
–1
.2)
(male
s4
30
years
old
) (co
nti
nu
ed)
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 849
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
4C
on
tin
ued
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
(fem
ale
so
nly
),ta
tto
o,
cau
tery
�P
art
icip
ati
on
rate
:6
03
3(5
3.7
%)
of
the
11
22
7vi
llage
inh
ab
itan
tsw
ere
incl
ud
edin
the
stu
dy
�It
isn
ot
clea
rh
ow
fem
ale
sub
ject
sw
ere
dea
ltw
ith
inth
ere
gre
ssio
nan
aly
ses
El-
Sad
aw
y43
�S
tud
yd
esig
n:
cro
ss-
sect
ion
al
stu
dy
�F
un
din
g:
no
tre
po
rted
�S
etti
ng
an
dp
erio
d:
Urb
an
an
dru
ral
are
as
of
Sh
ark
iaG
ove
rno
rate
,E
gyp
t;p
erio
dn
ot
rep
ort
ed�
Po
pu
lati
on
:m
ale
vill
age
inh
ab
itan
tssc
reen
edfo
rh
epati
-ti
sC
.2
17
test
edp
osi
tive
for
Hep
ati
tis
Can
d5
65
test
edn
egati
vefo
rH
epati
tis
C(s
tud
yals
ore
cru
ited
fem
ale
sb
ut
rest
rict
edgo
zaan
a-
lysi
sto
male
s)
�T
ype:
Gro
up
go
zasm
ok
ing
(ass
esse
din
men
on
ly)
�M
easu
rem
ent
too
l:sp
ecif
icall
yd
esig
ned
qu
esti
on
nair
e;p
ilo
tte
sted
bef
ore
surv
eyfo
rre
liab
ilit
y�
Ex
po
sure
leve
lso
fin
clu
ded
sub
ject
s:n
ot
rep
ort
ed
Hep
ati
tis
Cin
fect
ion
�M
easu
rem
ent
too
l:an
tib
od
yto
HC
Vass
esse
db
yM
EIA
an
dH
CV
RN
Ate
sted
by
real-
tim
eP
CR
�B
lin
din
go
fo
utc
om
ead
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
use
d‘s
trati
fied
ran
do
msa
m-
pli
ng,
wh
ich
incl
ud
edsy
stem
ati
csa
mp
lin
g�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
s;h
ow
ever
,w
hil
e2
5.8
%te
sted
po
siti
veb
yM
EIA
,7
.66
%te
stp
osi
tive
by
PC
R;
stan
dard
ized
exp
osu
rem
easu
rem
ent:
no
(pil
ot
test
edo
nly
)�
Co
nfo
un
din
g:
un
clea
rw
het
her
rep
ort
edre
sult
sw
ere
ad
just
edfo
rp
ote
nti
al
con
fou
nd
ers
�P
art
icip
ati
on
rate
no
tre
po
rted
�M
issi
ng
data
no
tre
po
rted
�O
Rfo
rgro
up
go
zasm
ok
ing
com
pare
dw
ith
no
n-g
rou
pgo
zasm
ok
ing¼
1.0
2(0
.64
–1
.62
)(m
ale
s)
ME
IA,
mic
ro-p
art
icle
enzy
me
imm
un
oass
ay;
PC
R,
po
lym
erase
chain
react
ion
.
850 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
not exclude important associations. Smoking water-pipe in groups was not significantly associated withhepatitis C infection. The overall quality of evidencevaried from very low to low.
To our knowledge, this is the first meta-analysis ofstudies assessing the association of waterpipe smok-ing with health outcomes. While Knishkowy et al.24
reviewed the health effects of waterpipe tobaccosmoking, they did not use systematic methods, for-mally evaluate the quality of the evidence or use sta-tistical methods to pool the results across studies. Themain strength of our study is the use of the CochraneCollaboration methodology for conducting systematicreviews, i.e. using a very sensitive and comprehensivesearch strategy, a duplicate and independent selectionprocess, a duplicate and independent data abstractionprocess, and a rigorous appraisal of the methodolog-ical quality of included studies. We also used theGRADE approach to rate the overall quality of evi-dence for each outcome.
For most of the outcomes of interest the evidencewas either lacking, indirect or of lower quality. Interms of indirectness, lung cancer studies were con-ducted mostly in China and India where tobacco istypically unprocessed and burned directly by charcoal.The practice of waterpipe smoking that is involved inthe recent global epidemic involves tobacco that isprocessed and flavoured and indirectly heated by thecharcoal. In terms of quality, the one methodologicalstudy limitation that affected our rating of the qualityof evidence for most outcomes was measurementbias, which has been reported in similar systematicreviews.48 In fact, only one study used a standardizedexposure measurement tool in spite of the fact thatthe practice of waterpipe smoking can vary widely.Variables include the quantity of tobacco used, thetype of tobacco used, the concomitant use of othersubstances, the frequency of smoking sessions, andthe length of sessions, the number of years of smok-ing. In addition, no study reported using a standard-ized measurement tool for other forms of tobacco
smoking in spite of the variety of these forms andthe need to account for passive smoking and pastsmoking history.
The other methodological study limitation thataffected our rating of the quality of evidence formany outcomes was the inappropriate handling ofconfounding, particularly for other forms of tobaccosmoking and for factors such as radon exposureamong miners as a risk factor for lung cancer.Furthermore, although many studies reported sepa-rate ORs for waterpipe and cigarette smoking, nonereported tests for interactions between these twoforms of smoking. This type of information isneeded given that a significant proportion of water-pipe smokers also smoke cigarettes.11,49
In spite of the many methodological study limita-tions for lung cancer and pregnancy outcomes, thefindings showed consistency (i.e. the low heterogene-ity) across studies. Similarly, the lack of associationbetween smoking waterpipe in groups and hepatitis Cinfection was characterized by the consistency acrossstudies and a relatively narrow CI. These findings areconsistent with current pathophysiological knowledgeof the spread of hepatitis C by blood-to-blood contact.While the lack of evidence for an association betweenwaterpipe smoking with bladder cancer, nasopharyn-geal cancer, oral dysplasia and infertility could reflecta true absence of association, it could also beexplained by lack of power to detect any existingassociation given their relatively wide CIs. Indeed,the lack of studies and of insufficient data on suchlong-term health effects is related to the short epide-miological time frame of the recent waterpipeepidemic.
The one study that reported on coronary heartdisease was excluded because of its publicationexclusively as an abstract.18 The quality of the evi-dence from that study was very low and the ORwas 2.2 (95% CI 0.9–5.4) for ever smokers and 0.7(95% CI 0.3–1.9) for current smokers compared withnever smokers.
Study or Subgroup
El-Sadawy 2004
Habib 2001
Medhat 2002 A
Medhat 2002 B
Total (95% CI)
Heterogeneity: Tau² = 0.00; Chi² = 1.57, df = 3 (P = 0.67); I ² = 0%
Test for overall effect: Z = 0.18 (P = 0.86)
log[OR]
0.019803
0.09531
–0.10536
–0.22314
SE
0.236032
0.158242
0.407402
0.20687
Weight
20.6%
45.8%
6.9%
26.8%
100.0%
IV, Random, 95% CI
1.02 (0.64, 1.62)
1.10 (0.81, 1.50)
0.90 (0.41, 2.00)
0.80 (0.53, 1.20)
0.98 (0.80, 1.21)
OR
IV, Random, 95% CI
0.01 0.1 1 10 100Protective effect
OR
Harmful effect
Figure 5 Association between waterpipe tobacco smoking and hepatitis C infection
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 851
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Ta
ble
5C
hara
cter
isti
cso
fin
clu
ded
stu
die
sass
essi
ng
the
effe
cto
fw
ate
rpip
eto
bacc
osm
ok
ing
on
oth
ero
utc
om
es
Stu
dy
Po
pu
lati
on
Ex
po
sure
Ou
tco
me
sM
eth
od
olo
gic
al
fea
ture
sR
esu
lts
Tam
im4
5
�S
tud
yd
esig
n:
cro
ss-
sect
ion
al
stu
dy
�F
un
din
g:
no
tre
po
rted
�S
etti
ng
an
dp
erio
d:
stu
den
tsin
the
seco
nd
an
dth
ird
inte
rmed
iate
class
esfr
om
five
pri
-m
ary
sch
oo
lsin
Gre
ate
rB
eiru
tare
a,
inA
pri
l2
00
0�
Po
pu
lati
on
:6
25
stu
-d
ents
wit
hm
ean
age
13
years
(ran
ge
10
–1
5).
14
3st
ud
ents
(85
male
s)re
po
rted
resp
irato
rytr
act
sym
pto
ms,
48
2st
ud
ents
(23
7m
ale
s)re
po
rted
no
sym
pto
ms
�T
ype:
seco
nd
-han
dex
po
sure
toci
gar-
ette
s,n
arg
hil
e�
Mea
sure
men
tto
ol:
qu
esti
on
nair
e,st
an
-d
ard
izati
on
no
tre
po
rted
�E
xp
osu
rele
vels
of
incl
ud
edsu
bje
cts:
no
tre
po
rted
Res
pir
ato
ryil
lnes
sth
rou
gh
ou
tth
eye
ar
(no
tse
aso
nal)
incl
ud
ing
nasa
lco
n-
ges
tio
no
rw
hee
zin
g�
Mea
sure
men
tto
ol:
qu
esti
on
nair
eass
es-
sin
gcl
inic
al
con
di-
tio
nth
rou
gh
ou
tth
eye
ar
�B
lin
din
go
fo
utc
om
ead
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
sam
ple
rep
rese
nta
tive
of
po
pu
-la
tio
ns
of
the
five
pri
-m
ary
sch
oo
ls�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:n
o,
stan
dar-
diz
edex
po
sure
mea
sure
men
t:n
o�
Co
nfo
un
din
g:
no
matc
hin
go
rad
just
men
tin
the
an
aly
sis
rep
ort
ed�
Part
icip
ati
on
rate
:n
ot
rep
ort
ed
OR
com
pare
dw
ith
no
exp
osu
reat
ho
me
�2
.39
5%
CI
(1.1
–5
.1)
(narg
hil
eo
nly
exp
osu
re)
�2
.5(1
.6–3
.8)
(nar-
gh
ile
an
d/o
rci
gare
tte
exp
osu
re)
�3
.2(1
.9–5
.4)
(cig
ar-
ette
on
lyex
po
sure
)
Inh
ort
n4
6
�S
tud
yd
esig
n:
case
–co
ntr
ol
stu
dy
Fu
nd
ing:
Nati
on
al
Sci
ence
Fou
nd
ati
on
,th
eF
ulb
righ
tIn
stit
ute
for
Inte
rnati
on
al
Ed
uca
tio
n,
the
US
Dep
art
men
to
fE
du
cati
on
Fu
lbri
gh
t-H
ays
Do
cto
ral
Dis
sert
ati
on
Res
earc
hA
bro
ad
Pro
gra
man
dth
eS
oro
pti
mis
tsIn
tern
ati
on
al
Fo
un
der
Reg
ion
Fel
low
ship
Pro
gra
m
�S
etti
ng
an
dp
erio
d:
infe
rtil
ity
clin
ico
fth
eU
niv
ersi
tyo
fA
lex
an
dri
a’s
Sh
atb
yH
osp
ital,
the
pu
bli
co
bst
etri
cs/g
ynaec
olo
gy
teach
ing
ho
spit
al,
rura
lare
as
of
the
no
rth
wes
-te
rnN
ile
Del
tare
gio
n,
Egyp
t;O
cto
ber
19
88
toD
ecem
ber
19
89
�C
ase
s:h
usb
an
ds
of
40
fem
ale
pati
ents
of
the
un
iver
sity
infe
rtil
ity
clin
icw
ho
wer
ed
iag-
no
sed
of
male
-fact
or
infe
rtil
ity
(pro
ble
ms
of
sem
enan
dsp
erm
)�
Co
ntr
ols
:h
usb
an
ds
of
90
fert
ile
fem
ale
pati
ents
of
the
un
iver
-si
tyh
osp
ital
�P
art
icip
an
ts:
15
–4
5ye
ars
old
,lo
wer
an
dlo
wer
-mid
dle
soci
o-e
co-
no
mic
class
es
�T
ype:
hu
sban
ds’
wate
rpip
ean
dci
gar-
ette
smo
kin
gb
ehavi
ou
rs�
Mea
sure
men
tto
ol:
inte
rvie
ws,
self
-d
evel
op
edse
mi-
stru
ctu
red
qu
esti
on
-n
air
e,n
ost
an
dard
i-za
tio
nre
po
rted
;p
art
icip
an
tsca
tego
r-iz
edas
regu
lar
smo
-k
ers
an
dn
ever
smo
ker
s�
Ex
po
sure
leve
lso
fin
clu
ded
sub
ject
s:n
ot
rep
ort
ed
Co
up
lein
fert
ilit
yass
o-
ciate
dw
ith
male
-fa
ctor
infe
rtil
ity
�M
easu
rem
ent
too
l:m
edic
al
reco
rds
of
sem
enan
aly
ses
�B
lin
din
go
fo
utc
om
ead
jud
icato
r:n
ot
rep
ort
ed
�S
elec
tio
nb
ias:
pre
vale
nt
case
so
fin
fert
ilit
y,co
n-
tro
lsw
ere
ho
spit
al-
base
dan
dre
cru
ited
fro
mth
esa
me
stu
dy
base
as
case
s�
Info
rmati
on
bia
s:o
bje
ctiv
eo
utc
om
em
ea-
sure
men
t:ye
s,st
an
dar-
diz
edex
po
sure
mea
sure
men
t:n
o(s
eco
nd
-han
din
form
ati
on
)�
Co
nfo
un
din
g:
matc
hin
gfo
rage
gro
up
,so
cio
-ec
on
om
iccl
ass
;ad
just
-m
ent
for
cigare
tte
smo
kin
g,
tea
dri
nk
ing,
mari
tal
du
rati
on
,h
us-
ban
d’s
age,
hu
sban
d’s
edu
cati
on
�P
art
icip
ati
on
rate
:49
8%
OR
com
pare
dw
ith
nev
ersm
ok
ing:
2.5
(1.0
–6
.3)
(reg
ula
rw
ate
rpip
esm
ok
ing)
852 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
This study indicates the need for more research, asthe World Health Organization (WHO) has alreadyadvised.50 The overall quality of evidence for the dif-ferent outcomes was mostly low, limiting our confi-dence in the results. There is a need for high-qualitycohort studies to identify and quantify with confi-dence all the health effects of waterpipe smokingand to explore its interaction with other forms ofsmoking. The high quality of future studies willdepend on the use of standardized exposure measure-ment tools,49 another area of research need.
The study also has important implications for publichealth practice. Public health policy makers have todeal with suggestive but generally weak evidence ofthe association of waterpipe tobacco smoking withdeleterious health outcomes. One approach could beto prioritize public health resources to deal with otherpublic health crises (such as youth cigarette smoking)
for which the evidence is much stronger, while await-ing further evidence regarding waterpipe smoking.Another approach could be to devote the necessaryresources to control at a relatively early stage therapid growth of the waterpipe trend. Whatever thedecision policy makers make, it should be based onan objective consideration of the current available evi-dence in the proper public health context.
AcknowledgementsThe authors thank Ms Ann Grifasi for her adminis-trative assistance and Dr Ranime Saliba for her assis-tance. Those acknowledged have confirmed theiragreement.
Conflict of interest: None declared.
KEY MESSAGES
� Waterpipe tobacco smoking is possibly associated with a number of deleterious health outcomesincluding lung cancer, respiratory illness, low birth-weight and periodontal disease.
� There is a need for higher quality evidence to determine with confidence all the health effects ofwaterpipe tobacco smoking
References1 Chaouachi K. The medical consequences of narghile
(hookah, shisha) use in the world. Revue d Epidemiologieet de Sante Publique 2007;55:165–70.
2 American Lung Association. An Emerging Deadly Trend;Waterpipe Tobacco Use. American Lung Association:Washington, DC, 2007.
3 Sukumar A, Subramanian R. Elements in hair and nailsof residents from a village adjacent to New-Delhi—influ-ence of place of occupation and smoking-habits. BiologicalTrace Element Res 1992;34:99–105.
4 Maziak W, Ward KD, Soweid RAA, Eissenberg T. Tobaccosmoking using a waterpipe: a re-emerging strain in aglobal epidemic. Tobacco Control 2004;13:327–33.
5 Chaaya M, El Roueiheb Z, Chemaitelly H, Azar G, Nasr J,Al-Sahab B. Argileh smoking among universitystudents: a tobacco epidemic. Nicotine Tobacco Res 2004;6:457–63.
6 Al-Turki YA, Al-Turki YA. Smoking habits among medicalstudents in Central Saudi Arabia. Saudi Med J 2006;27:700–3.
7 Al-Haddad N, Hamadeh RR. Smoking among secondary-school boys in Bahrain: prevalence and risk factors. EastMediterr Health J 2003;9:78–86.
8 Kelishadi R, Ardalan G, Gheiratmand R et al.Smoking behavior and its influencing factors in a natio-nal-representative sample of Iranian adolescents:CASPIAN study. Preventive Med 2006;42:423–6.
9 El-Roueiheb Z, Tamim H, Kanj M, Jabbour S, Alayan I,Musharrafieh U. Cigarette and waterpipe smoking among
Lebanese adolescents, a cross-sectional study, 2003–2004.Nicotine Tobacco Res 2008;10:309–14.
10 Carroll T, Poder N, Perusco A. Is concern about waterpipetobacco smoking warranted? Aust N Z Publ Health 2008;32:181–2.
11 Jackson D, Aveyard P. Waterpipe smoking in students:prevalence, risk factors, symptoms of addiction, andsmoke intake. Evidence from one British university. BMCPublic Health 2008;8:174.
12 Roskin J, Aveyard P. Canadian and English students’beliefs about waterpipe smoking: a qualitative study.BMC Public Health 2009;9:10.
13 Eissenberg T, Ward KD, Smith-Simone S, Maziak W.Waterpipe tobacco smoking on a US collegecampus: Prevalence and correlates. J Adolesc Health 2008;42:526–9.
14 Weglicki LS, Templin TN, Rice VH, Jamil H, Hammad A.Comparison of cigarette and water-pipe smoking by Araband non-Arab-American youth. Am J Prev Med 2008;35:334–9.
15 Weglicki LS, Templin T, Hammad A et al. Tobacco usepatterns among high school students: do Arab Americanyouth differ? Ethn Dis 2007;17(2 Suppl 3):S3-22–S3-4.
16 Qiao YL, Taylor PR, Yao SX et al. Relation of radonexposure and tobacco use to lung cancer among tinminers in Yunnan Province, China. Am J Ind Med 1989;16:511–21.
17 Nasrollahzadeh D, Kamangar F, Aghcheli K et al. Opium,tobacco, and alcohol use in relation to oesophageal squa-mous cell carcinoma in a high-risk area of Iran. Br JCancer 2008;98:1857–63.
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 853
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
18 Jabbour S, El-Roueiheb Z, Sibai AM. Nargileh (Water-Pipe) smoking and incident coronary heart disease: acase-control study. Ann Epidemiol 2003;13:570.
19 Nuwayhid IA, Yamout B, Azar G et al. Narghile (hubble-bubble) smoking, low birth weight, and other pregnancyoutcomes. Am J Epidemiol 1998;148:375–83.
20 Shihadeh A, Saleh R. Polycyclic aromatic hydrocarbons,carbon monoxide, ‘‘tar’’, and nicotine in the mainstreamsmoke aerosol of the narghile water pipe. Food & ChemToxicol 2005;43:655–61.
21 Shihadeh A. Investigation of mainstream smoke aerosolof the argileh water pipe. Food Chem Toxicol 2003;41:143–52.
22 Habib M, Mohamed MK, Abdel-Aziz F et al. Hepatitis Cvirus infection in a community in the Nile Delta: riskfactors for seropositivity. Hepatology 2001;33:248–53.
23 Steentoft J, Wittendorf J, Andersen JR, Amitai Y.Tuberculosis and water pipes as source of infection.Ugeskr Laeger 2006;168:904–7.
24 Knishkowy B, Amitai Y, Knishkowy B, Amitai Y. Water-pipe (narghile) smoking: an emerging health risk beha-vior. Pediatrics 2005;116:e113–9.
25 Maziak W, Ward KD, Eissenberg T. Interventions forwaterpipe smoking cessation. Cochrane Database ofSystematic Rev 2007;4:CD005549.
26 Guyatt GH, Oxman AD, Vist GE et al. GRADE: an emer-ging consensus on rating quality of evidence and strengthof recommendations. BMJ 2008;336:924–6.
27 Higgins JPT, Thompson SG, Deeks JJ, Altman DG.Measuring inconsistency in meta-analysis. BMJ 2003;327:557–60.
28 Gupta D, Boffetta P, Gaborieau V et al. Risk factors oflung cancer in Chandigarh, India. Indian J Med Res2001;113:142–50.
29 Lubin JH, Qiao YL, Taylor PR et al. Quantitative evalua-tion of the radon and lung cancer association in a casecontrol study of Chinese tin miners. Cancer Res 1990;50:174–80.
30 Lubin JH, Li JY, Xuan XZ et al. Risk of lung cancer amongcigarette and pipe smokers in southern China. Int J Cancer1992;51:390–5.
31 Hsairi M, Achour N, Zouari B. Facteurs etiologiques ducancer bronchique primitif en Tunisie. La Tunisie Medicale1993;71:265–8.
32 Hazelton WD, Luebeck EG, Heidenreich WF,Moolgavkar SH. Analysis of a historical cohort ofChinese tin miners with arsenic, radon, cigarette smoke,and pipe smoke exposures using the biologically basedtwo-stage clonal expansion model. Radiat Res 2001;156:78–94.
33 Bedwani R, ElKhwsky F, Renganathan E et al.Epidemiology of bladder cancer in Alexandria, Egypt:Tobacco smoking. Int J Cancer 1997;73:64–7.
34 Feng BJ, Khyatti M, Ben-Ayoub W et al. Cannabis,tobacco and domestic fumes intake are associated withnasopharyngeal carcinoma in North Africa. Br J Cancer2009;101:1207–12.
35 Ali AA. Histopathologic changes in oral mucosa ofYemenis addicted to water-pipe and cigarette smoking
in addition to takhzeen al-qat. Oral Surg Oral Med OralPathol Oral Radiol Endod 2007;103:e55–9.
36 Tamim H, Yunis KA, Chemaitelly H et al. NationalCollaborative Perinatal Neonatal Network Beirut L.Effect of narghile and cigarette smoking on newbornbirthweight. BJOG 2008;115:91–7.
37 Aghamolaei T, Eftekhar H, Zare S. Risk factors associatedwith Intrauterine Growth Retardation (IUGR) in BandarAbbas. J Med Sci 2007;7:665–9.
38 Al-Belasy FA. The relationship of ‘‘shisha’’ (water pipe)smoking to postextraction dry socket. J Oral MaxillofacSurg 2004;62:10–4.
39 Natto S, Baljoon M, Bergstrom J. Tobacco smoking andperiodontal bone height in a Saudi Arabian population. JClin Periodontol 2005;32:1000–6.
40 Natto S, Baljoon M, Abanmy A et al. Tobacco smokingand gingival health in a Saudi Arabian population. OralHealth Prev 2004;2:351–7.
41 Natto S, Baljoon M, Bergstrom J. Tobacco smokingand periodontal health in a Saudi Arabian population.J Periodontol 2005;76:1919–26.
42 Baljoon M, Natto S, Abanmy A et al. Smoking and verticalbone defects in a Saudi Arabian population. Oral HealthPrev 2005;3:173–82.
43 El-Sadawy M, Ragab H, el-Toukhy H et al. Hepatitis Cvirus infection at Sharkia Governorate, Egypt: seropreva-lence and associated risk factors. J Egypt Soc Parasitol2004;34(Suppl 1):367–84.
44 Medhat A, Shehata M, Magder LS et al. Hepatitis C in acommunity in Upper Egypt: risk factors for infection. AmJ Trop Med Hyg 2002;66:633–8.
45 Tamim H, Musharrafieh U, El Roueiheb Z et al. Exposureof children to environmental tobacco smoke (ETS) and itsassociation with respiratory ailments. J Asthma 2003;40:571–6.
46 Inhorn MC, Buss KA. Ethnography, epidemiology andinfertility in Egypt. Soc Sci Med 1994;39:671–86.
47 Munckhof WJ, Konstantinos A, Wamsley M et al. A clus-ter of tuberculosis associated with use of a marijuanawater pipe. Int J Tuberc Lung Dis 2003;7:860–5.
48 Pirie K, Beral V, Peto R et al. Passive smoking and breastcancer in never smokers: prospective study and meta-analysis. Int J Epidemiol 2008;37:1069–79.
49 Maziak W, Ward KD, Afifi Soweid RA, Eissenberg T.Standardizing questionnaire items for the assessment ofwaterpipe tobacco use in epidemiological studies. PublicHealth 2005;119:400–4.
50 WHO Study Group on Tobacco Product Regualation(TobReg). Advisory Note. Waterpipe Tobacco Smoking:Health Effects, Research Needs and Recommended Actions byRegulators. WHO: Geneva, 2005.
51 Onder M, Oztas M, Arnavut O. Nargile (Hubble-Bubble)smoking-induced hand eczema. Int J Dermatol 2002;41:771–2.
52 Szyper-Kravitz M, Lang R, Manor Y, Lahav M. Early inva-sive pulmonary aspergillosis in a leukemia patient linkedto aspergillus contaminated marijuana smoking. LeukemiaLymphoma 2001;42:1433–7.
854 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Appendix 1 Electronic search strategiesMEDLINE (1950 onward)
Waterpipe�.mp.‘‘water pipe�’’.mp.shisha�.mp.sheesha�.mp.hooka�.mp.huqqa�.mp.guza�.mp.goza�.mp.narghil�.mp.nargil�.mp.arghil�.mpargil�.mp(hubbl� adj3 bubbl�).mp.or/1–13
EMBASE (1988 onward)Waterpipe�.mp.‘‘water pipe�’’.mp.shisha�.mp.sheesha�.mp.
hooka�.mp.huqqa�.mp.guza�.mp.goza�.mp.narghil�.mp.nargil�.mp.arghil�.mpargil�.mp(hubbl� adj3 bubbl�).mp.or/1-13
ISI the Web of Science(waterpipe� OR ‘‘water pipe�’’ OR shisha� ORsheesha� OR hooka� OR huqqa� OR guza� OR goza�
OR narghil� OR nargil� OR argil� OR arghil� OR(hubbl� SAME bubbl�)) AND (smoking OR smokeOR health OR disease OR cancer� OR malignan� ORlung� OR pulmonary OR heart OR cardiac OR vascularOR stroke) (in Title or Topic)
Appendix 2 GRADE approach for rating the quality of evidence
Factors that lower the quality ofevidence
(1) Limitations of design� Selection bias� Measurement bias for risk factor (validity of
risk factor measurement instrument, blinding
of risk status data collector and blinding ofrisk status adjudicator)
� Measurement bias for outcome (validity ofoutcome measurement instrument, blinding ofoutcome data collector and blinding outcomeadjudicator)
� Inappropriate handling of confounding(stratification, adjustment)
Quality is first assignedbased on study design
Quality of evidence Quality is lowered ifa Quality is raised ifa
Randomized trials ! 4 ¼ High Limitations of design Large effect�1 Serious þ 1 Large�2 Very serious þ 2 Very large
3 ¼ Moderate Inconsistency Dose response�1 Serious þ 1 Evidence of a gradient�2 Very seriousIndirectness All plausible confounding
Observational study ! 2 ¼ Low �1 Serious þ 1 Would reduce a demonstrated effect, or�2 Very seriousImprecision þ 1 Would suggest a spurious effect
when results show no effect�1 Serious1 ¼ Very low �2 Very serious
Publication bias�1 Likely�2 Very likely
The quality of evidence reflects the extent of confidence that an estimate of effect is correct.a1, move up or down one grade (for example from high to intermediate); 2, move up or down two grades (for example from highto low).
THE EFFECTS OF WATERPIPE TOBACCO SMOKING ON HEALTH 855
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
� Attrition (for cohort studies)(2) Inconsistency
Inconsistency in the estimates of the associationacross studies (i.e. heterogeneity) suggests truedifferences in underlying cause–effect relation-ship that may arise from differences in:� populations (e.g. different association in
populations with differing other specific expo-sures or other genetic profiles);
� risk factor (e.g. larger association with higherexposure or different associations with varyingforms of the same exposure);
� outcomes (e.g. different associations with dif-ferent length of follow-up because of dimin-ishing or increasing association with time).
When heterogeneity exists, but investigators failto identify a plausible explanation, the quality ofevidence should be downgraded by one or twolevels, depending on the magnitude of the incon-sistency in the results.
(3) IndirectnessThe question being addressed by the authors of asystematic review is different from the availableevidence regarding:� population (e.g. general population vs popu-
lation of miners);� risk factor (e.g. cigarette use vs pipe use);� outcome (e.g. intra-uterine growth vs low
birth-weight)(4) Imprecision
Results are imprecise when studies include rela-tively few patients and few events and thus havewide CIs around the estimate of the effect. The95% CI (or alternative estimate of precision)
around the pooled or best estimate of associationincludes both negligible association and appreci-able benefit or appreciable harm.
(5) Publication biasPublication bias is a systematic underestimate oran overestimate of the underlying beneficial orharmful effect due to the selective publicationof studies (publication bias). That is, investiga-tors fail to report studies they have undertaken(typically those that show no effect).
Factors that increase the quality of evidence
(1) Large magnitude of effectWhen methodologically strong observational stu-dies yield large or very large and consistent esti-mates of the magnitude of association, we maybe confident about the results. The larger themagnitude of effect, the stronger becomes theevidence.
(2) Dose–response gradientA gradient dose–response may increase our con-fidence in the findings and thus enhance theassigned quality of evidence.
(3) Plausible confounding, which would reduce ademonstrated effectOn occasion, all plausible confounders or biasesfrom observational studies would result in anunderestimate of the association estimate. If,for instance, only sick patients have an exposure,yet they still fare better, it is likely that the expo-sure effect is even larger than the data suggest.
Appendix 3 Rating the quality of evidence for different associationusing the GRADE framework
Outcome Factors lowering the rating of quality Factors raising the rating ofquality
Finalrating
Lung cancerdiagnosis
Limitations of design: four of five studies didnot appropriately deal with confounding
Dose response: two studiesreported statistically signifi-cant dose–response relation-ships in terms of the totalamount of exposure andduration of exposure
Very low
Indirectness: most studies likely usedunprocessed tobacco burned directly bycharcoal
Lung cancermortality
Limitations of design: included study did notappropriately deal with confounding
None Very low
Indirectness: study likely used unprocessedtobacco burned directly by charcoal
Bladdercancerdiagnosis
Limitations of design: use of prevalent cases None Very lowImprecision
None Low
(continued)
856 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from
Published by Oxford University Press on behalf of the International Epidemiological Association
� The Author 2010; all rights reserved. Advance Access publication 25 March 2010
International Journal of Epidemiology 2010;39:857–859
doi:10.1093/ije/dyq054
Commentary: The waterpipe—a globalepidemic or a passing fadWasim Maziak1,2*
1School of Public Health, University of Memphis, Memphis, TN, USA and 2Syrian Center for Tobacco Studies, Aleppo, Syria
*Corresponding author. Browning Hall 112, 3820 DeSoto Avenue, Memphis 38152-3340, TN, USA. E-mail: [email protected]
Accepted 15 February 2010
Six years ago, we published in Tobacco Control the firstcomprehensive review about the waterpipe (also
known as shisha, hookah, arghile and narghile) inresponse to what seemed to be a burgeoning global
Oesophagealcancer
Dose response: the includedstudy reported statisticallysignificant dose response rela-tionship in terms of intensityof use
Imprecision
Nasopharyng-eal cancer
Limitations of design: investigators usedmatching and adjusted the analysis for anumber of confounding factors but not forother types of smoking
None Very low
ImprecisionOral dysplasia Limitations of design: not reporting dealing
with confounding factors; also the analysiswas stratified by chewing side (i.e. analy-sis of outcome occurring on the chewingside separate from the analysis of outcomeoccurring on the non-chewing side)
None Very low
ImprecisionPregnancy
outcomesNone None Low
Periodontaldisease
Limitations of design: included studies par-tially dealt with confounding factors (threeadjusted for age only and one adjusted forage and dental care habit)
Dose response: two studiesfound a statistically significantdose response relationship interms of total amount ofexposure
Low
Dry socket Limitations in design: study did not ade-quately deal with confounding factors
Dose response: statistically sig-nificant dose response rela-tionship in terms of totalamount of exposure
Low
Hepatitis Cinfection
Limitations in design: study reported discre-pancies between the results of MEIA andPCR
None Very low
Respiratoryillness
Limitations in design: study did not ade-quately deal with confounding factors
None Very low
Infertility NoneImprecision
None Very low
THE WATER PIPE 857
by guest on October 14, 2011
ije.oxfordjournals.orgD
ownloaded from