Abstract
Background There is a need for a comprehensive and critical review of the literature to inform scientific debates about the public health effects of waterpipe smoking. The objective of this study was therefore to systematically review the medical literature for the effects of waterpipe tobacco smoking on health outcomes.
Methods We conducted a systematic review using the Cochrane Collaboration methodology for conducting systematic reviews. We rated the quality of evidence for each outcome using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology.
Results Twenty-four studies were eligible for this review. Based on the available evidence, waterpipe tobacco smoking was significantly associated with lung cancer [odds ratio (OR) = 2.12; 95% confidence interval (CI) 1.32–3.42], respiratory illness (OR = 2.3; 95% CI 1.1–5.1), low birth-weight (OR = 2.12; 95% CI 1.08–4.18) and periodontal disease (OR = 3–5). It was not significantly associated with bladder cancer (OR = 0.8; 95% CI 0.2–4.0), nasopharyngeal cancer (OR = 0.49; 95% CI 0.20–1.23), oesophageal cancer (OR = 1.85; 95% CI 0.95–3.58), oral dysplasia (OR = 8.33; 95% CI 0.78–9.47) or infertility (OR = 2.5; 95% CI 1.0–6.3) but the CIs did not exclude important associations. Smoking waterpipe in groups was not significantly associated with hepatitis C infection (OR = 0.98; 95% CI 0.80–1.21). The quality of evidence for the different outcomes varied from very low to low.
Conclusion Waterpipe tobacco smoking is possibly associated with a number of deleterious health outcomes. There is a need for high-quality studies to identify and quantify with confidence all the health effects of this form of smoking.
Background
Tobacco smoking using waterpipe—also known as arguileh, hookah and shisha—is an older form of tobacco consumption traditional of the eastern Mediterranean region (Figure 1).1 The waterpipe device indirectly heats the tobacco. The resulting smoke then passes through a column of water before being inhaled through the mouth using a pipe.2 These two unique features of this form of smoking are assumed to minimize its tobacco-related health hazards. Also, waterpipe tobacco smoke is produced at a much lower temperature than cigarette smoke, suggesting that the toxins may be different for these different forms of tobacco smoking.3 Another unique feature is the social nature of waterpipe smoking, whereby family members and friends may share the same device, which has been hypothesized to cause the transmission of communicable diseases.
Figure 1
Annotated figure of a waterpipe device
Waterpipe tobacco smoking has been spreading globally at a remarkable pace.4 Although its use in the eastern Mediterranean region has been, until recently, confined to adults, it is now gaining popularity among university students5,6 and teenagers.7–9 Waterpipe smoking is also spreading to Western countries such as Australia,10 the UK,11 Canada12 and the USA,13 where it is also affecting young people and adolescents.14,15
Waterpipe tobacco smoking is generally considered as a public health threat and the American Lung Association has recently called it the ‘emerging deadly trend’.2 In fact, waterpipe smoking has been suspected to be a risk factor for a number of tobacco-related diseases such as lung cancer,16 oesophageal cancers,17 cardiovascular disease18 and adverse pregnancy outcomes.19 In addition, the waterpipe device may expose its user (via its non-tobacco components) to metals and cancer-causing chemicals.20,21 The potential association of waterpipe smoking with communicable diseases such as hepatitis C22 and tuberculosis23 (via its shared and repetitive use without proper sanitation) has also been investigated.
While one article reviewed the effects of waterpipe smoking on health outcomes,24 we have identified no published systematic review or meta-analysis of the topic. The objective of this study was therefore to systematically review the medical literature for the effects of waterpipe tobacco smoking on health outcomes.
Methods
Eligibility criteria
We included observational studies (i.e. cohort studies, case–control studies and cross-sectional studies) assessing the association between waterpipe tobacco smoking and health outcomes. We excluded case reports, case series, outbreak investigations and abstracts. We also excluded studies assessing physiological outcomes [e.g. Forced Expiratory Volume in 1 Second (FEV1)], assessing waterpipe use for non-tobacco smoking purposes (e.g. marijuana smoking), not distinguishing waterpipe smoking from other forms of smoking, and not reporting any measure of association.
Search strategy
In June 2008, we electronically searched the following databases: MEDLINE (1950 onwards; access via OVID), EMBASE (1980 onwards; access via OVID) and ISI the Web of Science using a detailed search strategy with no language restrictions ( Appendix 1). We designed the strategy based on a preliminary review of relevant articles, an extensive Internet search for waterpipe synonyms and the search strategy of a systematic review on interventions for waterpipe smoking cessation.25 Two medical librarians reviewed and commented on the search strategy. We also reviewed the reference lists of included and relevant papers and used the ‘Related Articles’ feature in PubMed.
Selection process
Two reviewers independently screened the title and abstract of identified citations for potential eligibility using a standardized screening guide. We retrieved the full texts of citations judged potentially eligible by at least one reviewer. Then, two reviewers independently screened the full texts for eligibility using a standardized and pilot-tested form. They resolved their disagreements by discussion or by consulting a third reviewer.
Data abstraction
Two reviewers used a standardized and pilot-tested form to independently abstract data. They resolved their disagreements by discussion or by consulting a third reviewer. The abstracted data related to study design, population, exposure, outcomes, methodological features, results and funding. For exposure measurement, we assessed whether the instrument was standardized to measure the unique smoking patterns and characteristics of waterpipe smoking. We recorded the results of analyses restricted to waterpipe only smokers (i.e. people who smoked only waterpipe) as well as the results of analyses that included all waterpipe smokers (i.e. people who smoked waterpipe and other forms of tobacco). We recorded the effect measures derived from the regression models that adjusted for the maximum number of covariates. We rated the overall quality of evidence for each outcome using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach ( Appendix 2).26
Data analysis
We calculated the kappa statistic to evaluate the agreement between the two reviewers assessing full texts for eligibility. We conducted meta-analyses for the outcomes for which at least two studies reported effect estimates of their association with waterpipe tobacco smoking. When the authors reported odds ratios (ORs) for more than one analysis we selected the OR according to the following order: (i) analysis adjusted for other forms of tobacco smoking (choosing the one adjusting for the maximum number of confounders); (ii) analysis restricted to waterpipe only smokers; (iii) analysis not adjusted for other forms of tobacco smoking and not restricted to waterpipe only smokers. We used the ORs to calculate the corresponding ln(ORs) and standard errors. We then pooled, for each outcome, the ln(ORs) of eligible studies using the generic inverse variance and the random effects model in Review Manager Version 5.0.20. We measured homogeneity across study results using the I2 statistic.27 We checked for possible publication bias using inverted funnel plots.
Results
Figure 2 shows the study flow. We included 23 reports on: lung cancer (n = 6),16,28–32 bladder cancer (n = 1),33 oesophageal cancers (n = 1),17 nasopharyngeal cancer (n = 1),34 dysplasia of the oral mucosa (n = 1),35 pregnancy outcomes (n = 3),19,36,37 periodontal disease (n = 5),38–42 hepatitis C infection (n = 3),22,43,44 respiratory illness (n = 1)45 and infertility (n = 1).46 Agreement between reviewers for study eligibility was excellent (kappa = 0.963).
Lung cancer
Five of the six included studies were case–control studies assessing lung cancer diagnosis,16,28–31 and one was a retrospective cohort study assessing lung cancer mortality (Table 1).32 One was conducted in Northern India,28 one was conducted in Tunisia,31 whereas rule four reported data from the same population in China. While in the recent global epidemic the tobacco is processed and flavoured and indirectly heated by the charcoal, in most of the included studies (conducted in China and India) the tobacco is typically unprocessed and burned directly by charcoal.
Table 1 Characteristics of included studies assessing the effect of waterpipe tobacco smoking on cancer outcomes
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Qiao16 | Setting and period: Gejiu city, Yunnan Province, China, interviews conducted in 1985 Cases: 107 lung cancer cases in males 35–80 years old reported to Labor Protection Institute of the Yunnan Tin Corporation (YTC) during 1967–84 Controls: 107 controls chosen systematically by selecting every 20th person from the list of all living past or present workers of the YTC; one control per case
| Type: Chinese water pipes, cigarettes Measurement tool: self-developed questionnaire, no standardization reported; cumulative consumption calculated as pipe years Exposure levels: Cases: mean pipe years = 177 (range 0–560); Controls: mean pipe years = 122 (range 0–480)
| Lung cancer diagnosis | Selection bias: series of incident cases but unclear whether all incident cases reported, controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no. Surrogates were interviewed for 10% of cases and 6% of controls Confounding: matching for age; adjustment for age; no adjustment reported for radon Participation rate: not reported
| OR compared with never smoking: 1.9 (95% CI 0.4–9.4) (waterpipe only smoking) 2.1 (95% CI 0.5–9.1) (waterpipe and cigarette smoking) 0.9 (95% CI 0.1–5.4) (cigarette only smoking) Statistically significant dose response to water pipe smoking OR=3.4 (1.3–8.1) by quarter of pipe-years |
Lubin29Study design: case–control study Funding: China Science and Technology Commission; and the US National Cancer Institute Unclear whether this is a report on a subpopulation of Qiao18
| Setting and period: Gejiu city, Yunnan Province, China, study conducted in 1985 Cases: 74 lung cancer cases in males with mean age 62 years (range 35–80) alive at the time of the study reported to Labor Protection Institute of the YTC during 1981–84 Controls: 74 controls chosen from the list of all living past or present workers of the YTC; one control per case
| Type: Chinese water pipes, cigarettes Measurement tool: self-developed questionnaire, no standardization reported; cumulative consumption calculated as pipe years Exposure levels of included subjects: number of pipe years (cases/controls): 0 (6/16); 1–114 (18/23); 115–220 (21/15); ≥220 (29/20)
| Lung cancer diagnosis Measurement tool: cases confirmed by an independent panel of pathologists, clinicians and cytologists Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases but unclear whether all incident cases reported; excluded those who had died by the time of the study; controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no; validated confounder measurement (arsenic): no Confounding: matching for age (within 5-year age groups) Participation rate: not reported Data were sparse and authors were unable to adjust for important confounding factors
| OR compared with no tobacco smoking: |
Lubin30 | Setting and period: Gejiu city, Yunnan Province, China Cases: 427 male lung cancer patients, mean age 63 years (range 35–75); reported between 1984 and 1988 to the Cancer Registry of the Labor Protection Institute of YTC or to the Gejiu City Cancer Registry; 339 from Yunnan Tin Corporation and 88 from Gejiu City Controls: 1011 male controls, two controls per case, mean age 62 years (range 35–75), 770 from YTC and 241 from Gejiu city
| Type: Chinese waterpipe, cigarettes Measurement tool: self-developed standardized structured questionnaire, no standardization reported; consumption calculated as pipe years Exposure levels of included subjects: mean duration of pipe only smoking is 41 years
| Lung cancer diagnosis Measurement tool: confirmed by independent panel of pathologists, clinicians, radiologists and cytologists Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases, controls were community based and selected from the same study base as cases; Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age (within 5-year age groups); adjustment for age, number of years of work underground, source of subject, type of respondent Participation rate: 72% for cases, 88% for controls
| OR compared with no smoking: ORs for pipe users show increasing trend with increased duration of use (test for trend statistically significant) |
Hsairi31 | Setting and period: Ariana, Tunis, December 1988 to May 1989 Cases: 110 lung cancer patients; epidermoid (56%), anaplastic small cell (17%), undifferentiated (13%), adenocarcinoma (12%), anaplastic large cell (3%), Controls: 110 residents of the same area
| Type: waterpipe, pipe, inhaled tobacco, cannabis Measurement tool: self-developed tool, no standardization reported Exposure levels of included subjects: not reported
| Lung cancer diagnosis Measurement tool: histologically confirmed in 70% of cases, ‘very probable’ in 30% of cases based on clinical, radiological and endoscopic suspicion Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases, controls were community based and selected from the same study base as cases. Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age, sex, consumption of cigarettes per day; adjustment for age, sex, cigarette consumption, occupational risks Participation rate: not reported
| OR compared with no smoking: |
Gupta28Study design: case–control study Funding: International Agency for Research on Cancer, Lyon France; and the International cancer related technology transfer fellowship award
| Setting and period: Northern India; recruitment, January 1995 to June 1997 Cases: 265 incident cases of lung cancer Controls: 525 visitors and attendants of the patients; two controls per case Participants were 85% male, of different religions, ages ranging from <50 to >70 living in rural or urban areas
| Type: cigarettes, bidis, hookah Measurement tool: self-developed tool, no standardization reported; cumulative consumption calculated as pack-year equivalent of cigarettes (1 hookah = 4 cigarettes) stratified in four groups (0–9; 10–19; 20–29; 30+) Exposure levels of included subjects: incompletely reported
| Lung cancer diagnosis | Selection bias: series of hospital based incident cases, community-based controls selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age and sex; adjustment for age, and education. No adjustment for any confounding factor including other forms of tobacco consumption Participation rate: not reported Missing data: 54.3% for exposure
| OR compared with no smoking: |
Hazelton32 | Setting and period: Gejiu city, Yunnan Province, China, 12 years follow-up (1976–88) Exposed: 1289 male waterpipe-only smokers and 2306 male waterpipe and cigarette smokers Non-exposed: 8416 males
| Type: Chinese waterpipe, cigarette Measurement tool: not reported, calculated as cumulative exposure and categorized into quartiles of liang/month year Measurement time points: every 5 years Exposure levels of included subjects: cumulative dose quartiles: 0 (n = 8416), 0.16–4.44 (n = 877), 4.44–6.25 (n = 888), 6.25–9.21 (n = 906), 9.21–82.19 (n = 924)
| Lung cancer mortality | Selection bias: smokers were representative of the study base, non-smokers selected from the same community as smokers Information bias: unlikely for mortality outcome; standardized exposure measurement: no Confounding: no adjustment (no adjusted RR reported; crude RR calculated from reported data) Participation rate: not reported 8% lost to follow-up
| |
Bedwani33 | Setting and period: Greater Alexandria, Egypt; study conducted, January 1994 to July 1996 Cases: 151 male incident cases of invasive bladder cancer with median age 61 years (range 31–74) admitted to a network of general and teaching hospitals. Controls: 157 males admitted for acute, non-neoplastic, non-urinary tract, non-smoking-related conditions, median age 50 years (range 32–74) 39 female cases and 30 female controls were excluded because ‘only one female case was a smoker’
| Type: waterpipe, cigarette, hashish Measurement tool: self-developed structured questionnaire; standardization not reported; participants categorized as ever smokers and never smokers. Age at starting, duration of habit, average daily consumption extracted from history but not reported Exposure levels of included subjects: not reported
| Bladder cancer diagnosis | Selection bias: hospital-based cases of bladder cancer confirmed within the year preceding interview, hospital-based controls selected from the same study base as cases. Selection related to exposure status (in the case of females and patients with non-smoking-related conditions) Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: adjustment for age, education, type of house, history of schistosomiasis, high-risk occupation, tobacco smoking Participation rate: >95% of eligible subjects participated in the interview
| OR compared with never smoking: 0.8 (95% CI 0.2–4.0) (waterpipe smoking) 5.4 (95% CI 2.4–12.1) (<20/day cigarette smoking) 7.6 (95% CI 3.4–16.8) for ≥20/day cigarette smoking
|
Nasrollahzadeh17 | Setting and period: Atrak Clinic in Gonbad City, eastern Golestan Province of Iran, December 2003 to June 2007 Cases: 300 oesophageal squamous cell carcinoma cases, 50% males, mean age 64.5 years Controls: 571 controls, two population-based matched control subjects per case for 90% of cases
| Type: hookah, cigarettes, nass chewing, opium Measurement tool: self-developed questionnaire, tested for reliability and validity, cumulative consumption calculated as hookah-years and categorized into never users, ≤32 hookah-years, >32 hookah-years Exposure levels of included subjects: median of 32 hookah-years
| Oesophageal squamous cell carcinoma diagnosis | Selection bias: series of incident cases (70% of incident cases registered with local cancer registry were referred to the Atrak Clinic), controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: yes Confounding: matching for age, sex, residence; adjustment for education, ethnicity, other types of tobacco use, total intake of fruit and vegetables Participation rate: not reported
| OR compared with no smoking: 1.85 (95% CI 0.95–3.58) (hookah smoking) 1.69 (95% CI 0.76–3.77) (hookah only smoking) 1.50 (95% CI 0.92–2.43) (cigarette only smoking) Test for trend significant for intensity (P = 0.03) but not for duration, total amount, or age started)
|
Feng34 | Setting and period: five hospitals in Algeria, Morocco, and Tunisia, January 2002 to March 2005 Cases: 636 incident cases of nasopharyngeal cancer Controls: 615 controls [patients hospitalized for non-cancer diseases (61%) and friends and family of non-cancer patients (39%)]
| Type: cigarette, shisha, snuff, cannabis Measurement tool: interviews, self-developed questionnaire, no standardization reported; participants categorized as ever smokers and never smokers. Ages of starting, and quitting daily consumption extracted from history but not reported Exposure levels of included subjects: not reported
| Nasopharyngeal carcinoma | Selection bias: series of hospital-based incident cases (2001–04), controls were hospital-based or friends and family of cases, and recruited from the same study base as cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: matching for hospital, age, sex, household type (urban/rural); adjustment for age Participation rate: >90%
| OR compared with never smoking: 0.49 (95% CI 0.20–1.23) (ever shisha smoking) |
Ali35 | Setting and period: Yemen; period not reported Population: Yemeni volunteers (27 men and six women) all chewers of al- qat, mean age 38.5 years (range 22–58), all of them free from any systematic diseases. 11 waterpipe smokers, 11 cigarette smokers, 11 non-smokers
| Type: cigarette smoking, waterpipe smoking Measurement tool: interviews, standardization not reported, subjects categorized as heavy cigarette smokers (>20/day), non-smokers and waterpipe smokers Exposure levels of included subjects: not reported
| Dysplasia of oral mucosa diagnosis | Selection bias: volunteer recruitment into the study Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: no matching or adjustment in the analysis reported Participation rate: not reported
| OR compared with no smoking: Results are restricted to the chewing side; no events in either group on non-chewing side. OR calculated from reported numbers |
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Qiao16 | Setting and period: Gejiu city, Yunnan Province, China, interviews conducted in 1985 Cases: 107 lung cancer cases in males 35–80 years old reported to Labor Protection Institute of the Yunnan Tin Corporation (YTC) during 1967–84 Controls: 107 controls chosen systematically by selecting every 20th person from the list of all living past or present workers of the YTC; one control per case
| Type: Chinese water pipes, cigarettes Measurement tool: self-developed questionnaire, no standardization reported; cumulative consumption calculated as pipe years Exposure levels: Cases: mean pipe years = 177 (range 0–560); Controls: mean pipe years = 122 (range 0–480)
| Lung cancer diagnosis | Selection bias: series of incident cases but unclear whether all incident cases reported, controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no. Surrogates were interviewed for 10% of cases and 6% of controls Confounding: matching for age; adjustment for age; no adjustment reported for radon Participation rate: not reported
| OR compared with never smoking: 1.9 (95% CI 0.4–9.4) (waterpipe only smoking) 2.1 (95% CI 0.5–9.1) (waterpipe and cigarette smoking) 0.9 (95% CI 0.1–5.4) (cigarette only smoking) Statistically significant dose response to water pipe smoking OR=3.4 (1.3–8.1) by quarter of pipe-years |
Lubin29Study design: case–control study Funding: China Science and Technology Commission; and the US National Cancer Institute Unclear whether this is a report on a subpopulation of Qiao18
| Setting and period: Gejiu city, Yunnan Province, China, study conducted in 1985 Cases: 74 lung cancer cases in males with mean age 62 years (range 35–80) alive at the time of the study reported to Labor Protection Institute of the YTC during 1981–84 Controls: 74 controls chosen from the list of all living past or present workers of the YTC; one control per case
| Type: Chinese water pipes, cigarettes Measurement tool: self-developed questionnaire, no standardization reported; cumulative consumption calculated as pipe years Exposure levels of included subjects: number of pipe years (cases/controls): 0 (6/16); 1–114 (18/23); 115–220 (21/15); ≥220 (29/20)
| Lung cancer diagnosis Measurement tool: cases confirmed by an independent panel of pathologists, clinicians and cytologists Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases but unclear whether all incident cases reported; excluded those who had died by the time of the study; controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no; validated confounder measurement (arsenic): no Confounding: matching for age (within 5-year age groups) Participation rate: not reported Data were sparse and authors were unable to adjust for important confounding factors
| OR compared with no tobacco smoking: |
Lubin30 | Setting and period: Gejiu city, Yunnan Province, China Cases: 427 male lung cancer patients, mean age 63 years (range 35–75); reported between 1984 and 1988 to the Cancer Registry of the Labor Protection Institute of YTC or to the Gejiu City Cancer Registry; 339 from Yunnan Tin Corporation and 88 from Gejiu City Controls: 1011 male controls, two controls per case, mean age 62 years (range 35–75), 770 from YTC and 241 from Gejiu city
| Type: Chinese waterpipe, cigarettes Measurement tool: self-developed standardized structured questionnaire, no standardization reported; consumption calculated as pipe years Exposure levels of included subjects: mean duration of pipe only smoking is 41 years
| Lung cancer diagnosis Measurement tool: confirmed by independent panel of pathologists, clinicians, radiologists and cytologists Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases, controls were community based and selected from the same study base as cases; Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age (within 5-year age groups); adjustment for age, number of years of work underground, source of subject, type of respondent Participation rate: 72% for cases, 88% for controls
| OR compared with no smoking: ORs for pipe users show increasing trend with increased duration of use (test for trend statistically significant) |
Hsairi31 | Setting and period: Ariana, Tunis, December 1988 to May 1989 Cases: 110 lung cancer patients; epidermoid (56%), anaplastic small cell (17%), undifferentiated (13%), adenocarcinoma (12%), anaplastic large cell (3%), Controls: 110 residents of the same area
| Type: waterpipe, pipe, inhaled tobacco, cannabis Measurement tool: self-developed tool, no standardization reported Exposure levels of included subjects: not reported
| Lung cancer diagnosis Measurement tool: histologically confirmed in 70% of cases, ‘very probable’ in 30% of cases based on clinical, radiological and endoscopic suspicion Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases, controls were community based and selected from the same study base as cases. Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age, sex, consumption of cigarettes per day; adjustment for age, sex, cigarette consumption, occupational risks Participation rate: not reported
| OR compared with no smoking: |
Gupta28Study design: case–control study Funding: International Agency for Research on Cancer, Lyon France; and the International cancer related technology transfer fellowship award
| Setting and period: Northern India; recruitment, January 1995 to June 1997 Cases: 265 incident cases of lung cancer Controls: 525 visitors and attendants of the patients; two controls per case Participants were 85% male, of different religions, ages ranging from <50 to >70 living in rural or urban areas
| Type: cigarettes, bidis, hookah Measurement tool: self-developed tool, no standardization reported; cumulative consumption calculated as pack-year equivalent of cigarettes (1 hookah = 4 cigarettes) stratified in four groups (0–9; 10–19; 20–29; 30+) Exposure levels of included subjects: incompletely reported
| Lung cancer diagnosis | Selection bias: series of hospital based incident cases, community-based controls selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age and sex; adjustment for age, and education. No adjustment for any confounding factor including other forms of tobacco consumption Participation rate: not reported Missing data: 54.3% for exposure
| OR compared with no smoking: |
Hazelton32 | Setting and period: Gejiu city, Yunnan Province, China, 12 years follow-up (1976–88) Exposed: 1289 male waterpipe-only smokers and 2306 male waterpipe and cigarette smokers Non-exposed: 8416 males
| Type: Chinese waterpipe, cigarette Measurement tool: not reported, calculated as cumulative exposure and categorized into quartiles of liang/month year Measurement time points: every 5 years Exposure levels of included subjects: cumulative dose quartiles: 0 (n = 8416), 0.16–4.44 (n = 877), 4.44–6.25 (n = 888), 6.25–9.21 (n = 906), 9.21–82.19 (n = 924)
| Lung cancer mortality | Selection bias: smokers were representative of the study base, non-smokers selected from the same community as smokers Information bias: unlikely for mortality outcome; standardized exposure measurement: no Confounding: no adjustment (no adjusted RR reported; crude RR calculated from reported data) Participation rate: not reported 8% lost to follow-up
| |
Bedwani33 | Setting and period: Greater Alexandria, Egypt; study conducted, January 1994 to July 1996 Cases: 151 male incident cases of invasive bladder cancer with median age 61 years (range 31–74) admitted to a network of general and teaching hospitals. Controls: 157 males admitted for acute, non-neoplastic, non-urinary tract, non-smoking-related conditions, median age 50 years (range 32–74) 39 female cases and 30 female controls were excluded because ‘only one female case was a smoker’
| Type: waterpipe, cigarette, hashish Measurement tool: self-developed structured questionnaire; standardization not reported; participants categorized as ever smokers and never smokers. Age at starting, duration of habit, average daily consumption extracted from history but not reported Exposure levels of included subjects: not reported
| Bladder cancer diagnosis | Selection bias: hospital-based cases of bladder cancer confirmed within the year preceding interview, hospital-based controls selected from the same study base as cases. Selection related to exposure status (in the case of females and patients with non-smoking-related conditions) Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: adjustment for age, education, type of house, history of schistosomiasis, high-risk occupation, tobacco smoking Participation rate: >95% of eligible subjects participated in the interview
| OR compared with never smoking: 0.8 (95% CI 0.2–4.0) (waterpipe smoking) 5.4 (95% CI 2.4–12.1) (<20/day cigarette smoking) 7.6 (95% CI 3.4–16.8) for ≥20/day cigarette smoking
|
Nasrollahzadeh17 | Setting and period: Atrak Clinic in Gonbad City, eastern Golestan Province of Iran, December 2003 to June 2007 Cases: 300 oesophageal squamous cell carcinoma cases, 50% males, mean age 64.5 years Controls: 571 controls, two population-based matched control subjects per case for 90% of cases
| Type: hookah, cigarettes, nass chewing, opium Measurement tool: self-developed questionnaire, tested for reliability and validity, cumulative consumption calculated as hookah-years and categorized into never users, ≤32 hookah-years, >32 hookah-years Exposure levels of included subjects: median of 32 hookah-years
| Oesophageal squamous cell carcinoma diagnosis | Selection bias: series of incident cases (70% of incident cases registered with local cancer registry were referred to the Atrak Clinic), controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: yes Confounding: matching for age, sex, residence; adjustment for education, ethnicity, other types of tobacco use, total intake of fruit and vegetables Participation rate: not reported
| OR compared with no smoking: 1.85 (95% CI 0.95–3.58) (hookah smoking) 1.69 (95% CI 0.76–3.77) (hookah only smoking) 1.50 (95% CI 0.92–2.43) (cigarette only smoking) Test for trend significant for intensity (P = 0.03) but not for duration, total amount, or age started)
|
Feng34 | Setting and period: five hospitals in Algeria, Morocco, and Tunisia, January 2002 to March 2005 Cases: 636 incident cases of nasopharyngeal cancer Controls: 615 controls [patients hospitalized for non-cancer diseases (61%) and friends and family of non-cancer patients (39%)]
| Type: cigarette, shisha, snuff, cannabis Measurement tool: interviews, self-developed questionnaire, no standardization reported; participants categorized as ever smokers and never smokers. Ages of starting, and quitting daily consumption extracted from history but not reported Exposure levels of included subjects: not reported
| Nasopharyngeal carcinoma | Selection bias: series of hospital-based incident cases (2001–04), controls were hospital-based or friends and family of cases, and recruited from the same study base as cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: matching for hospital, age, sex, household type (urban/rural); adjustment for age Participation rate: >90%
| OR compared with never smoking: 0.49 (95% CI 0.20–1.23) (ever shisha smoking) |
Ali35 | Setting and period: Yemen; period not reported Population: Yemeni volunteers (27 men and six women) all chewers of al- qat, mean age 38.5 years (range 22–58), all of them free from any systematic diseases. 11 waterpipe smokers, 11 cigarette smokers, 11 non-smokers
| Type: cigarette smoking, waterpipe smoking Measurement tool: interviews, standardization not reported, subjects categorized as heavy cigarette smokers (>20/day), non-smokers and waterpipe smokers Exposure levels of included subjects: not reported
| Dysplasia of oral mucosa diagnosis | Selection bias: volunteer recruitment into the study Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: no matching or adjustment in the analysis reported Participation rate: not reported
| OR compared with no smoking: Results are restricted to the chewing side; no events in either group on non-chewing side. OR calculated from reported numbers |
Table 1 Characteristics of included studies assessing the effect of waterpipe tobacco smoking on cancer outcomes
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Qiao16 | Setting and period: Gejiu city, Yunnan Province, China, interviews conducted in 1985 Cases: 107 lung cancer cases in males 35–80 years old reported to Labor Protection Institute of the Yunnan Tin Corporation (YTC) during 1967–84 Controls: 107 controls chosen systematically by selecting every 20th person from the list of all living past or present workers of the YTC; one control per case
| Type: Chinese water pipes, cigarettes Measurement tool: self-developed questionnaire, no standardization reported; cumulative consumption calculated as pipe years Exposure levels: Cases: mean pipe years = 177 (range 0–560); Controls: mean pipe years = 122 (range 0–480)
| Lung cancer diagnosis | Selection bias: series of incident cases but unclear whether all incident cases reported, controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no. Surrogates were interviewed for 10% of cases and 6% of controls Confounding: matching for age; adjustment for age; no adjustment reported for radon Participation rate: not reported
| OR compared with never smoking: 1.9 (95% CI 0.4–9.4) (waterpipe only smoking) 2.1 (95% CI 0.5–9.1) (waterpipe and cigarette smoking) 0.9 (95% CI 0.1–5.4) (cigarette only smoking) Statistically significant dose response to water pipe smoking OR=3.4 (1.3–8.1) by quarter of pipe-years |
Lubin29Study design: case–control study Funding: China Science and Technology Commission; and the US National Cancer Institute Unclear whether this is a report on a subpopulation of Qiao18
| Setting and period: Gejiu city, Yunnan Province, China, study conducted in 1985 Cases: 74 lung cancer cases in males with mean age 62 years (range 35–80) alive at the time of the study reported to Labor Protection Institute of the YTC during 1981–84 Controls: 74 controls chosen from the list of all living past or present workers of the YTC; one control per case
| Type: Chinese water pipes, cigarettes Measurement tool: self-developed questionnaire, no standardization reported; cumulative consumption calculated as pipe years Exposure levels of included subjects: number of pipe years (cases/controls): 0 (6/16); 1–114 (18/23); 115–220 (21/15); ≥220 (29/20)
| Lung cancer diagnosis Measurement tool: cases confirmed by an independent panel of pathologists, clinicians and cytologists Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases but unclear whether all incident cases reported; excluded those who had died by the time of the study; controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no; validated confounder measurement (arsenic): no Confounding: matching for age (within 5-year age groups) Participation rate: not reported Data were sparse and authors were unable to adjust for important confounding factors
| OR compared with no tobacco smoking: |
Lubin30 | Setting and period: Gejiu city, Yunnan Province, China Cases: 427 male lung cancer patients, mean age 63 years (range 35–75); reported between 1984 and 1988 to the Cancer Registry of the Labor Protection Institute of YTC or to the Gejiu City Cancer Registry; 339 from Yunnan Tin Corporation and 88 from Gejiu City Controls: 1011 male controls, two controls per case, mean age 62 years (range 35–75), 770 from YTC and 241 from Gejiu city
| Type: Chinese waterpipe, cigarettes Measurement tool: self-developed standardized structured questionnaire, no standardization reported; consumption calculated as pipe years Exposure levels of included subjects: mean duration of pipe only smoking is 41 years
| Lung cancer diagnosis Measurement tool: confirmed by independent panel of pathologists, clinicians, radiologists and cytologists Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases, controls were community based and selected from the same study base as cases; Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age (within 5-year age groups); adjustment for age, number of years of work underground, source of subject, type of respondent Participation rate: 72% for cases, 88% for controls
| OR compared with no smoking: ORs for pipe users show increasing trend with increased duration of use (test for trend statistically significant) |
Hsairi31 | Setting and period: Ariana, Tunis, December 1988 to May 1989 Cases: 110 lung cancer patients; epidermoid (56%), anaplastic small cell (17%), undifferentiated (13%), adenocarcinoma (12%), anaplastic large cell (3%), Controls: 110 residents of the same area
| Type: waterpipe, pipe, inhaled tobacco, cannabis Measurement tool: self-developed tool, no standardization reported Exposure levels of included subjects: not reported
| Lung cancer diagnosis Measurement tool: histologically confirmed in 70% of cases, ‘very probable’ in 30% of cases based on clinical, radiological and endoscopic suspicion Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases, controls were community based and selected from the same study base as cases. Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age, sex, consumption of cigarettes per day; adjustment for age, sex, cigarette consumption, occupational risks Participation rate: not reported
| OR compared with no smoking: |
Gupta28Study design: case–control study Funding: International Agency for Research on Cancer, Lyon France; and the International cancer related technology transfer fellowship award
| Setting and period: Northern India; recruitment, January 1995 to June 1997 Cases: 265 incident cases of lung cancer Controls: 525 visitors and attendants of the patients; two controls per case Participants were 85% male, of different religions, ages ranging from <50 to >70 living in rural or urban areas
| Type: cigarettes, bidis, hookah Measurement tool: self-developed tool, no standardization reported; cumulative consumption calculated as pack-year equivalent of cigarettes (1 hookah = 4 cigarettes) stratified in four groups (0–9; 10–19; 20–29; 30+) Exposure levels of included subjects: incompletely reported
| Lung cancer diagnosis | Selection bias: series of hospital based incident cases, community-based controls selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age and sex; adjustment for age, and education. No adjustment for any confounding factor including other forms of tobacco consumption Participation rate: not reported Missing data: 54.3% for exposure
| OR compared with no smoking: |
Hazelton32 | Setting and period: Gejiu city, Yunnan Province, China, 12 years follow-up (1976–88) Exposed: 1289 male waterpipe-only smokers and 2306 male waterpipe and cigarette smokers Non-exposed: 8416 males
| Type: Chinese waterpipe, cigarette Measurement tool: not reported, calculated as cumulative exposure and categorized into quartiles of liang/month year Measurement time points: every 5 years Exposure levels of included subjects: cumulative dose quartiles: 0 (n = 8416), 0.16–4.44 (n = 877), 4.44–6.25 (n = 888), 6.25–9.21 (n = 906), 9.21–82.19 (n = 924)
| Lung cancer mortality | Selection bias: smokers were representative of the study base, non-smokers selected from the same community as smokers Information bias: unlikely for mortality outcome; standardized exposure measurement: no Confounding: no adjustment (no adjusted RR reported; crude RR calculated from reported data) Participation rate: not reported 8% lost to follow-up
| |
Bedwani33 | Setting and period: Greater Alexandria, Egypt; study conducted, January 1994 to July 1996 Cases: 151 male incident cases of invasive bladder cancer with median age 61 years (range 31–74) admitted to a network of general and teaching hospitals. Controls: 157 males admitted for acute, non-neoplastic, non-urinary tract, non-smoking-related conditions, median age 50 years (range 32–74) 39 female cases and 30 female controls were excluded because ‘only one female case was a smoker’
| Type: waterpipe, cigarette, hashish Measurement tool: self-developed structured questionnaire; standardization not reported; participants categorized as ever smokers and never smokers. Age at starting, duration of habit, average daily consumption extracted from history but not reported Exposure levels of included subjects: not reported
| Bladder cancer diagnosis | Selection bias: hospital-based cases of bladder cancer confirmed within the year preceding interview, hospital-based controls selected from the same study base as cases. Selection related to exposure status (in the case of females and patients with non-smoking-related conditions) Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: adjustment for age, education, type of house, history of schistosomiasis, high-risk occupation, tobacco smoking Participation rate: >95% of eligible subjects participated in the interview
| OR compared with never smoking: 0.8 (95% CI 0.2–4.0) (waterpipe smoking) 5.4 (95% CI 2.4–12.1) (<20/day cigarette smoking) 7.6 (95% CI 3.4–16.8) for ≥20/day cigarette smoking
|
Nasrollahzadeh17 | Setting and period: Atrak Clinic in Gonbad City, eastern Golestan Province of Iran, December 2003 to June 2007 Cases: 300 oesophageal squamous cell carcinoma cases, 50% males, mean age 64.5 years Controls: 571 controls, two population-based matched control subjects per case for 90% of cases
| Type: hookah, cigarettes, nass chewing, opium Measurement tool: self-developed questionnaire, tested for reliability and validity, cumulative consumption calculated as hookah-years and categorized into never users, ≤32 hookah-years, >32 hookah-years Exposure levels of included subjects: median of 32 hookah-years
| Oesophageal squamous cell carcinoma diagnosis | Selection bias: series of incident cases (70% of incident cases registered with local cancer registry were referred to the Atrak Clinic), controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: yes Confounding: matching for age, sex, residence; adjustment for education, ethnicity, other types of tobacco use, total intake of fruit and vegetables Participation rate: not reported
| OR compared with no smoking: 1.85 (95% CI 0.95–3.58) (hookah smoking) 1.69 (95% CI 0.76–3.77) (hookah only smoking) 1.50 (95% CI 0.92–2.43) (cigarette only smoking) Test for trend significant for intensity (P = 0.03) but not for duration, total amount, or age started)
|
Feng34 | Setting and period: five hospitals in Algeria, Morocco, and Tunisia, January 2002 to March 2005 Cases: 636 incident cases of nasopharyngeal cancer Controls: 615 controls [patients hospitalized for non-cancer diseases (61%) and friends and family of non-cancer patients (39%)]
| Type: cigarette, shisha, snuff, cannabis Measurement tool: interviews, self-developed questionnaire, no standardization reported; participants categorized as ever smokers and never smokers. Ages of starting, and quitting daily consumption extracted from history but not reported Exposure levels of included subjects: not reported
| Nasopharyngeal carcinoma | Selection bias: series of hospital-based incident cases (2001–04), controls were hospital-based or friends and family of cases, and recruited from the same study base as cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: matching for hospital, age, sex, household type (urban/rural); adjustment for age Participation rate: >90%
| OR compared with never smoking: 0.49 (95% CI 0.20–1.23) (ever shisha smoking) |
Ali35 | Setting and period: Yemen; period not reported Population: Yemeni volunteers (27 men and six women) all chewers of al- qat, mean age 38.5 years (range 22–58), all of them free from any systematic diseases. 11 waterpipe smokers, 11 cigarette smokers, 11 non-smokers
| Type: cigarette smoking, waterpipe smoking Measurement tool: interviews, standardization not reported, subjects categorized as heavy cigarette smokers (>20/day), non-smokers and waterpipe smokers Exposure levels of included subjects: not reported
| Dysplasia of oral mucosa diagnosis | Selection bias: volunteer recruitment into the study Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: no matching or adjustment in the analysis reported Participation rate: not reported
| OR compared with no smoking: Results are restricted to the chewing side; no events in either group on non-chewing side. OR calculated from reported numbers |
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Qiao16 | Setting and period: Gejiu city, Yunnan Province, China, interviews conducted in 1985 Cases: 107 lung cancer cases in males 35–80 years old reported to Labor Protection Institute of the Yunnan Tin Corporation (YTC) during 1967–84 Controls: 107 controls chosen systematically by selecting every 20th person from the list of all living past or present workers of the YTC; one control per case
| Type: Chinese water pipes, cigarettes Measurement tool: self-developed questionnaire, no standardization reported; cumulative consumption calculated as pipe years Exposure levels: Cases: mean pipe years = 177 (range 0–560); Controls: mean pipe years = 122 (range 0–480)
| Lung cancer diagnosis | Selection bias: series of incident cases but unclear whether all incident cases reported, controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no. Surrogates were interviewed for 10% of cases and 6% of controls Confounding: matching for age; adjustment for age; no adjustment reported for radon Participation rate: not reported
| OR compared with never smoking: 1.9 (95% CI 0.4–9.4) (waterpipe only smoking) 2.1 (95% CI 0.5–9.1) (waterpipe and cigarette smoking) 0.9 (95% CI 0.1–5.4) (cigarette only smoking) Statistically significant dose response to water pipe smoking OR=3.4 (1.3–8.1) by quarter of pipe-years |
Lubin29Study design: case–control study Funding: China Science and Technology Commission; and the US National Cancer Institute Unclear whether this is a report on a subpopulation of Qiao18
| Setting and period: Gejiu city, Yunnan Province, China, study conducted in 1985 Cases: 74 lung cancer cases in males with mean age 62 years (range 35–80) alive at the time of the study reported to Labor Protection Institute of the YTC during 1981–84 Controls: 74 controls chosen from the list of all living past or present workers of the YTC; one control per case
| Type: Chinese water pipes, cigarettes Measurement tool: self-developed questionnaire, no standardization reported; cumulative consumption calculated as pipe years Exposure levels of included subjects: number of pipe years (cases/controls): 0 (6/16); 1–114 (18/23); 115–220 (21/15); ≥220 (29/20)
| Lung cancer diagnosis Measurement tool: cases confirmed by an independent panel of pathologists, clinicians and cytologists Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases but unclear whether all incident cases reported; excluded those who had died by the time of the study; controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no; validated confounder measurement (arsenic): no Confounding: matching for age (within 5-year age groups) Participation rate: not reported Data were sparse and authors were unable to adjust for important confounding factors
| OR compared with no tobacco smoking: |
Lubin30 | Setting and period: Gejiu city, Yunnan Province, China Cases: 427 male lung cancer patients, mean age 63 years (range 35–75); reported between 1984 and 1988 to the Cancer Registry of the Labor Protection Institute of YTC or to the Gejiu City Cancer Registry; 339 from Yunnan Tin Corporation and 88 from Gejiu City Controls: 1011 male controls, two controls per case, mean age 62 years (range 35–75), 770 from YTC and 241 from Gejiu city
| Type: Chinese waterpipe, cigarettes Measurement tool: self-developed standardized structured questionnaire, no standardization reported; consumption calculated as pipe years Exposure levels of included subjects: mean duration of pipe only smoking is 41 years
| Lung cancer diagnosis Measurement tool: confirmed by independent panel of pathologists, clinicians, radiologists and cytologists Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases, controls were community based and selected from the same study base as cases; Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age (within 5-year age groups); adjustment for age, number of years of work underground, source of subject, type of respondent Participation rate: 72% for cases, 88% for controls
| OR compared with no smoking: ORs for pipe users show increasing trend with increased duration of use (test for trend statistically significant) |
Hsairi31 | Setting and period: Ariana, Tunis, December 1988 to May 1989 Cases: 110 lung cancer patients; epidermoid (56%), anaplastic small cell (17%), undifferentiated (13%), adenocarcinoma (12%), anaplastic large cell (3%), Controls: 110 residents of the same area
| Type: waterpipe, pipe, inhaled tobacco, cannabis Measurement tool: self-developed tool, no standardization reported Exposure levels of included subjects: not reported
| Lung cancer diagnosis Measurement tool: histologically confirmed in 70% of cases, ‘very probable’ in 30% of cases based on clinical, radiological and endoscopic suspicion Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases, controls were community based and selected from the same study base as cases. Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age, sex, consumption of cigarettes per day; adjustment for age, sex, cigarette consumption, occupational risks Participation rate: not reported
| OR compared with no smoking: |
Gupta28Study design: case–control study Funding: International Agency for Research on Cancer, Lyon France; and the International cancer related technology transfer fellowship award
| Setting and period: Northern India; recruitment, January 1995 to June 1997 Cases: 265 incident cases of lung cancer Controls: 525 visitors and attendants of the patients; two controls per case Participants were 85% male, of different religions, ages ranging from <50 to >70 living in rural or urban areas
| Type: cigarettes, bidis, hookah Measurement tool: self-developed tool, no standardization reported; cumulative consumption calculated as pack-year equivalent of cigarettes (1 hookah = 4 cigarettes) stratified in four groups (0–9; 10–19; 20–29; 30+) Exposure levels of included subjects: incompletely reported
| Lung cancer diagnosis | Selection bias: series of hospital based incident cases, community-based controls selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: matching for age and sex; adjustment for age, and education. No adjustment for any confounding factor including other forms of tobacco consumption Participation rate: not reported Missing data: 54.3% for exposure
| OR compared with no smoking: |
Hazelton32 | Setting and period: Gejiu city, Yunnan Province, China, 12 years follow-up (1976–88) Exposed: 1289 male waterpipe-only smokers and 2306 male waterpipe and cigarette smokers Non-exposed: 8416 males
| Type: Chinese waterpipe, cigarette Measurement tool: not reported, calculated as cumulative exposure and categorized into quartiles of liang/month year Measurement time points: every 5 years Exposure levels of included subjects: cumulative dose quartiles: 0 (n = 8416), 0.16–4.44 (n = 877), 4.44–6.25 (n = 888), 6.25–9.21 (n = 906), 9.21–82.19 (n = 924)
| Lung cancer mortality | Selection bias: smokers were representative of the study base, non-smokers selected from the same community as smokers Information bias: unlikely for mortality outcome; standardized exposure measurement: no Confounding: no adjustment (no adjusted RR reported; crude RR calculated from reported data) Participation rate: not reported 8% lost to follow-up
| |
Bedwani33 | Setting and period: Greater Alexandria, Egypt; study conducted, January 1994 to July 1996 Cases: 151 male incident cases of invasive bladder cancer with median age 61 years (range 31–74) admitted to a network of general and teaching hospitals. Controls: 157 males admitted for acute, non-neoplastic, non-urinary tract, non-smoking-related conditions, median age 50 years (range 32–74) 39 female cases and 30 female controls were excluded because ‘only one female case was a smoker’
| Type: waterpipe, cigarette, hashish Measurement tool: self-developed structured questionnaire; standardization not reported; participants categorized as ever smokers and never smokers. Age at starting, duration of habit, average daily consumption extracted from history but not reported Exposure levels of included subjects: not reported
| Bladder cancer diagnosis | Selection bias: hospital-based cases of bladder cancer confirmed within the year preceding interview, hospital-based controls selected from the same study base as cases. Selection related to exposure status (in the case of females and patients with non-smoking-related conditions) Information bias: objective outcome evaluation: yes; standardized exposure measurement: no Confounding: adjustment for age, education, type of house, history of schistosomiasis, high-risk occupation, tobacco smoking Participation rate: >95% of eligible subjects participated in the interview
| OR compared with never smoking: 0.8 (95% CI 0.2–4.0) (waterpipe smoking) 5.4 (95% CI 2.4–12.1) (<20/day cigarette smoking) 7.6 (95% CI 3.4–16.8) for ≥20/day cigarette smoking
|
Nasrollahzadeh17 | Setting and period: Atrak Clinic in Gonbad City, eastern Golestan Province of Iran, December 2003 to June 2007 Cases: 300 oesophageal squamous cell carcinoma cases, 50% males, mean age 64.5 years Controls: 571 controls, two population-based matched control subjects per case for 90% of cases
| Type: hookah, cigarettes, nass chewing, opium Measurement tool: self-developed questionnaire, tested for reliability and validity, cumulative consumption calculated as hookah-years and categorized into never users, ≤32 hookah-years, >32 hookah-years Exposure levels of included subjects: median of 32 hookah-years
| Oesophageal squamous cell carcinoma diagnosis | Selection bias: series of incident cases (70% of incident cases registered with local cancer registry were referred to the Atrak Clinic), controls were selected from the same study base as cases Information bias: objective outcome evaluation: yes; standardized exposure measurement: yes Confounding: matching for age, sex, residence; adjustment for education, ethnicity, other types of tobacco use, total intake of fruit and vegetables Participation rate: not reported
| OR compared with no smoking: 1.85 (95% CI 0.95–3.58) (hookah smoking) 1.69 (95% CI 0.76–3.77) (hookah only smoking) 1.50 (95% CI 0.92–2.43) (cigarette only smoking) Test for trend significant for intensity (P = 0.03) but not for duration, total amount, or age started)
|
Feng34 | Setting and period: five hospitals in Algeria, Morocco, and Tunisia, January 2002 to March 2005 Cases: 636 incident cases of nasopharyngeal cancer Controls: 615 controls [patients hospitalized for non-cancer diseases (61%) and friends and family of non-cancer patients (39%)]
| Type: cigarette, shisha, snuff, cannabis Measurement tool: interviews, self-developed questionnaire, no standardization reported; participants categorized as ever smokers and never smokers. Ages of starting, and quitting daily consumption extracted from history but not reported Exposure levels of included subjects: not reported
| Nasopharyngeal carcinoma | Selection bias: series of hospital-based incident cases (2001–04), controls were hospital-based or friends and family of cases, and recruited from the same study base as cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: matching for hospital, age, sex, household type (urban/rural); adjustment for age Participation rate: >90%
| OR compared with never smoking: 0.49 (95% CI 0.20–1.23) (ever shisha smoking) |
Ali35 | Setting and period: Yemen; period not reported Population: Yemeni volunteers (27 men and six women) all chewers of al- qat, mean age 38.5 years (range 22–58), all of them free from any systematic diseases. 11 waterpipe smokers, 11 cigarette smokers, 11 non-smokers
| Type: cigarette smoking, waterpipe smoking Measurement tool: interviews, standardization not reported, subjects categorized as heavy cigarette smokers (>20/day), non-smokers and waterpipe smokers Exposure levels of included subjects: not reported
| Dysplasia of oral mucosa diagnosis | Selection bias: volunteer recruitment into the study Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: no matching or adjustment in the analysis reported Participation rate: not reported
| OR compared with no smoking: Results are restricted to the chewing side; no events in either group on non-chewing side. OR calculated from reported numbers |
The pooled OR for the association of waterpipe tobacco smoking with lung cancer diagnosis was 2.12 [95% confidence interval (CI) 1.32–3.42; I2 = 0%] (Figure 3). We judged the overall quality of evidence to be very low ( Appendix 3). The calculated crude relative risk (RR) for the association with lung cancer mortality was 4.39 (3.82–5.04). We judged the overall quality of evidence to be very low ( Appendix 3). A sensitivity analysis restricted to studies with no major methodological limitations, included the study by Hsairi et al.31 for which the OR was 3.0 (95% CI 1.2–7.6).
Figure 3
Association between waterpipe tobacco smoking and lung cancer
Bladder cancer
The one included case–control study33 reported a potentially protective association between waterpipe tobacco smoking and bladder cancer diagnosis of 0.8 (95% CI 0.2–4.0) (Table 1). A stratified analysis by the amount of cigarette smoking suggested a potential interaction between waterpipe and cigarette smoking but the CI overlapped and no interaction test was reported. An interaction test we ran was statistically not significant (P = 0.55). We judged the overall quality of evidence to be very low ( Appendix 3).
Oesophageal cancer
The one study assessing oesophageal cancer diagnosis was a case–control study (Table 1).17 The reported OR for the association of waterpipe tobacco smoking with oesophageal cancer diagnosis was 1.85 (95% CI 0.95–3.58). We judged the overall quality of evidence to be low ( Appendix 3).
Nasopharyngeal cancer
The one included study was a case–control study (Table 1). The OR for the association of waterpipe tobacco smoking with nasopharyngeal cancer was 0.49 (95% CI 0.20–1.23). We judged the overall quality of evidence to be very low ( Appendix 3).
Oral dysplasia
The one included study was a cross-sectional study that recruited exclusively subjects who practiced ‘takhzeen al-qat’, i.e. a practice that is distinct from waterpipe smoking and consists of chewing a green-leaved plant for its stimulant effect (Table 1).35 The OR for the association of waterpipe tobacco smoking with dysplasia of the oral mucosa on chewing side was 8.33 (95% CI 0.78–9.47). There were no events observed in either group on non-chewing side. We judged the overall quality of evidence to be very low ( Appendix 3).
Pregnancy outcomes
Of the three included studies, two were retrospective cohort studies19,36 and one was a case–control study (Table 2).37 All three studies assessed low birth-weight. One study also reported on Apgar score, pulmonary problems, malformations and perinatal complications.19 The pooled OR for the association of waterpipe tobacco smoking with low birth-weight was 2.12 (95% CI 1.08–4.18; I2 = 0%) (Figure 4). The reported OR for the association of waterpipe tobacco smoking with newborn pulmonary problems was 3.65 (95% CI 1.52, 8.75). The associations were not significant for Apgar scores at 1 and 5 min, malformations or perinatal complications. We judged the overall quality of evidence for pregnancy outcomes to be low ( Appendix 3).
Figure 4
Association between waterpipe tobacco smoking and low birth-weight
Table 2 Characteristics of included studies assessing the effect of waterpipe tobacco smoking on pregnancy outcomes
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Nuwayhid19 | Setting and period: hospitals in Lebanon; 1993 and 1995 Participants: pregnant women delivering in hospitals. 106 narghile smokers (mean age 27.6 years), 277 cigarette smokers (mean age 29.1) and 512 non-smokers (mean age 28.5)
| Type of exposure: narghile, cigarettes Measurement tool: interviews, standardization not reported; categorized into <1 vs ≥1 per day; and started smoking during first vs second or third trimester Measurement time points: not reported Exposure levels of included subjects: <1 per day (n = 38); ≥1 per day (n = 67) per day; first trimester (n = 78); second or third trimester (n = 25)
| Low birth-weight, Apgar score <7 (1 and 5 min), pulmonary problems, malformations, perinatal complications | Selection bias: unclear whether sample is representative (data collection in a few number of hospitals) Information bias: objective outcome measurement: yes but based on medical records, standardized exposure measurement: no Confounding: adjustment for age, hospital location, gestational age, exposure to passive smoking, and previous low birth-weight Participation rate: not reported <10% missing data
| OR for narghile smoking compared with no smoking low birth-weight = 2.17 (95% CI 0.74–6.33) low birth-weight = 2.36 (95% CI 0.52–10.73) (<1/day) low birth-weight = 2.07 (95% CI 0.54–7.94) (≥1/day) low birth-weight = 2.93 (95% CI 0.97–8.83) (narghile smoking started in first trimester) Test for trend for low birth-weight not significant for intensity (P < 0.18) Apgar score <7 at 1 min = 1.73 (95% CI 0.73–4.14) Apgar score <7 at 5 min = 3.39 (95% CI 0.54–21.42) Pulmonary problems = 3.65 (95% CI 1.52–8.75) Malformations = 2.01 (95% CI 0.59–6.88) Perinatal complications =1.67 (95% CI 0.82–3.41) OR for cigarette smoking compared with no smoking Low birth-weight = 2.00 (95% CI 0.96–4.20) low birth-weight = 2.25 (95% CI 1.04–4.86) (narghile smoking started in first trimester) Apgar score <7 at 1 min = 1.59 (95% CI 0.82–3.07) Apgar score <7 at 5 min = 2.62 (95% CI 0.56–12.29) Pulmonary problems = 1.76 (0.80–3.87) Malformations = 1.36 (0.49–3.80) Perinatal complications = 1.50 (0.87–2.57)
|
Aghamolaei37 | Setting and period: Shariati Hospital of Bandar Abbas, South of Iran; period not reported Cases: 60 Intra-uterine growth retardation infants Controls: 60 normal birth infants Included subjects: all term infants with gestational age 37–42 weeks; 29 males and 31 females in each group. ‘None of included mothers smoked and used alcohol during pregnancy and none of them had diabetes’
| Type: hookah smoking by mother Measurement tool: structured questionnaire administered to mothers, standardization not reported Exposure levels of included subjects: not reported
| Intra uterine growth retardation (IUGR) Definition: term infants with a birth-weight <2500 g Measurement tool: birth weight determined up to 10 min of delivery using a digital baby scale Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases of IUGR in maternity wards, controls selected from the same study base as the cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: explored 18 potential risk factors and adjusted for maternal work, hypertension, antenatal care Participation rate: not reported
| |
Tamim36 | Setting and period: six major hospitals in greater Beirut, Lebanon, August 2000 to August 2003 Population: 378 singleton newborns to women exclusive narghile smokers, 929 singleton newborns to women exclusive cigarette smokers, 84 singleton newborns to women smoking both types of tobacco, 7201 singleton newborns to women non-smokers
| Type: narghile, cigarettes Measurement tool: interviews, questionnaire pilot tested; narghile smokers categorized into ≤1 vs >1 per day; categorized into <1 vs ≥1 per day; and started smoking during first vs second or third trimester Measurement time points: not reported Exposure levels of included subjects: ≤1 per day (n = 233); >1 per day (n = 145); first trimester (n = 309); second or third trimester (n = 69)
| Low birth-weight Measurement tool: data collection from obstetric and nursery charts, low birth-weight defined as ≤2500 g Blinding of outcome adjudicator: not reported
| Selection bias: sample representative of populations of six major hospitals in Greater Beirut Information bias: objective outcome measurement: yes but based on medical records, standardized exposure measurement: no (pilot tested only) Confounding: adjustment for maternal/paternal education, mother’s working status, mother’s age, parity, diabetes, bleeding, pregnancy hypertensive disorders, gestational age, and passive smoking Participation rate: not reported
| OR for narghile only smoking compared with no smoking low birth-weight = 1.32 (95% CI 0.39–4.40) (OR pooled for ≤1 day and >1 day) low birth-weight = 0.7 (95% CI 0.3–1.6) (≤1 day) low birth-weight = 2.4 (95% CI 1.2–5.0) (>1 day) low birth-weight = 1.2 (95% CI 0.6–2.2) (narghile smoking started in first trimester) OR for cigarette only smoking compared with no smoking = 1.32 (95% CI 0.91–1.90)
|
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Nuwayhid19 | Setting and period: hospitals in Lebanon; 1993 and 1995 Participants: pregnant women delivering in hospitals. 106 narghile smokers (mean age 27.6 years), 277 cigarette smokers (mean age 29.1) and 512 non-smokers (mean age 28.5)
| Type of exposure: narghile, cigarettes Measurement tool: interviews, standardization not reported; categorized into <1 vs ≥1 per day; and started smoking during first vs second or third trimester Measurement time points: not reported Exposure levels of included subjects: <1 per day (n = 38); ≥1 per day (n = 67) per day; first trimester (n = 78); second or third trimester (n = 25)
| Low birth-weight, Apgar score <7 (1 and 5 min), pulmonary problems, malformations, perinatal complications | Selection bias: unclear whether sample is representative (data collection in a few number of hospitals) Information bias: objective outcome measurement: yes but based on medical records, standardized exposure measurement: no Confounding: adjustment for age, hospital location, gestational age, exposure to passive smoking, and previous low birth-weight Participation rate: not reported <10% missing data
| OR for narghile smoking compared with no smoking low birth-weight = 2.17 (95% CI 0.74–6.33) low birth-weight = 2.36 (95% CI 0.52–10.73) (<1/day) low birth-weight = 2.07 (95% CI 0.54–7.94) (≥1/day) low birth-weight = 2.93 (95% CI 0.97–8.83) (narghile smoking started in first trimester) Test for trend for low birth-weight not significant for intensity (P < 0.18) Apgar score <7 at 1 min = 1.73 (95% CI 0.73–4.14) Apgar score <7 at 5 min = 3.39 (95% CI 0.54–21.42) Pulmonary problems = 3.65 (95% CI 1.52–8.75) Malformations = 2.01 (95% CI 0.59–6.88) Perinatal complications =1.67 (95% CI 0.82–3.41) OR for cigarette smoking compared with no smoking Low birth-weight = 2.00 (95% CI 0.96–4.20) low birth-weight = 2.25 (95% CI 1.04–4.86) (narghile smoking started in first trimester) Apgar score <7 at 1 min = 1.59 (95% CI 0.82–3.07) Apgar score <7 at 5 min = 2.62 (95% CI 0.56–12.29) Pulmonary problems = 1.76 (0.80–3.87) Malformations = 1.36 (0.49–3.80) Perinatal complications = 1.50 (0.87–2.57)
|
Aghamolaei37 | Setting and period: Shariati Hospital of Bandar Abbas, South of Iran; period not reported Cases: 60 Intra-uterine growth retardation infants Controls: 60 normal birth infants Included subjects: all term infants with gestational age 37–42 weeks; 29 males and 31 females in each group. ‘None of included mothers smoked and used alcohol during pregnancy and none of them had diabetes’
| Type: hookah smoking by mother Measurement tool: structured questionnaire administered to mothers, standardization not reported Exposure levels of included subjects: not reported
| Intra uterine growth retardation (IUGR) Definition: term infants with a birth-weight <2500 g Measurement tool: birth weight determined up to 10 min of delivery using a digital baby scale Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases of IUGR in maternity wards, controls selected from the same study base as the cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: explored 18 potential risk factors and adjusted for maternal work, hypertension, antenatal care Participation rate: not reported
| |
Tamim36 | Setting and period: six major hospitals in greater Beirut, Lebanon, August 2000 to August 2003 Population: 378 singleton newborns to women exclusive narghile smokers, 929 singleton newborns to women exclusive cigarette smokers, 84 singleton newborns to women smoking both types of tobacco, 7201 singleton newborns to women non-smokers
| Type: narghile, cigarettes Measurement tool: interviews, questionnaire pilot tested; narghile smokers categorized into ≤1 vs >1 per day; categorized into <1 vs ≥1 per day; and started smoking during first vs second or third trimester Measurement time points: not reported Exposure levels of included subjects: ≤1 per day (n = 233); >1 per day (n = 145); first trimester (n = 309); second or third trimester (n = 69)
| Low birth-weight Measurement tool: data collection from obstetric and nursery charts, low birth-weight defined as ≤2500 g Blinding of outcome adjudicator: not reported
| Selection bias: sample representative of populations of six major hospitals in Greater Beirut Information bias: objective outcome measurement: yes but based on medical records, standardized exposure measurement: no (pilot tested only) Confounding: adjustment for maternal/paternal education, mother’s working status, mother’s age, parity, diabetes, bleeding, pregnancy hypertensive disorders, gestational age, and passive smoking Participation rate: not reported
| OR for narghile only smoking compared with no smoking low birth-weight = 1.32 (95% CI 0.39–4.40) (OR pooled for ≤1 day and >1 day) low birth-weight = 0.7 (95% CI 0.3–1.6) (≤1 day) low birth-weight = 2.4 (95% CI 1.2–5.0) (>1 day) low birth-weight = 1.2 (95% CI 0.6–2.2) (narghile smoking started in first trimester) OR for cigarette only smoking compared with no smoking = 1.32 (95% CI 0.91–1.90)
|
Table 2 Characteristics of included studies assessing the effect of waterpipe tobacco smoking on pregnancy outcomes
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Nuwayhid19 | Setting and period: hospitals in Lebanon; 1993 and 1995 Participants: pregnant women delivering in hospitals. 106 narghile smokers (mean age 27.6 years), 277 cigarette smokers (mean age 29.1) and 512 non-smokers (mean age 28.5)
| Type of exposure: narghile, cigarettes Measurement tool: interviews, standardization not reported; categorized into <1 vs ≥1 per day; and started smoking during first vs second or third trimester Measurement time points: not reported Exposure levels of included subjects: <1 per day (n = 38); ≥1 per day (n = 67) per day; first trimester (n = 78); second or third trimester (n = 25)
| Low birth-weight, Apgar score <7 (1 and 5 min), pulmonary problems, malformations, perinatal complications | Selection bias: unclear whether sample is representative (data collection in a few number of hospitals) Information bias: objective outcome measurement: yes but based on medical records, standardized exposure measurement: no Confounding: adjustment for age, hospital location, gestational age, exposure to passive smoking, and previous low birth-weight Participation rate: not reported <10% missing data
| OR for narghile smoking compared with no smoking low birth-weight = 2.17 (95% CI 0.74–6.33) low birth-weight = 2.36 (95% CI 0.52–10.73) (<1/day) low birth-weight = 2.07 (95% CI 0.54–7.94) (≥1/day) low birth-weight = 2.93 (95% CI 0.97–8.83) (narghile smoking started in first trimester) Test for trend for low birth-weight not significant for intensity (P < 0.18) Apgar score <7 at 1 min = 1.73 (95% CI 0.73–4.14) Apgar score <7 at 5 min = 3.39 (95% CI 0.54–21.42) Pulmonary problems = 3.65 (95% CI 1.52–8.75) Malformations = 2.01 (95% CI 0.59–6.88) Perinatal complications =1.67 (95% CI 0.82–3.41) OR for cigarette smoking compared with no smoking Low birth-weight = 2.00 (95% CI 0.96–4.20) low birth-weight = 2.25 (95% CI 1.04–4.86) (narghile smoking started in first trimester) Apgar score <7 at 1 min = 1.59 (95% CI 0.82–3.07) Apgar score <7 at 5 min = 2.62 (95% CI 0.56–12.29) Pulmonary problems = 1.76 (0.80–3.87) Malformations = 1.36 (0.49–3.80) Perinatal complications = 1.50 (0.87–2.57)
|
Aghamolaei37 | Setting and period: Shariati Hospital of Bandar Abbas, South of Iran; period not reported Cases: 60 Intra-uterine growth retardation infants Controls: 60 normal birth infants Included subjects: all term infants with gestational age 37–42 weeks; 29 males and 31 females in each group. ‘None of included mothers smoked and used alcohol during pregnancy and none of them had diabetes’
| Type: hookah smoking by mother Measurement tool: structured questionnaire administered to mothers, standardization not reported Exposure levels of included subjects: not reported
| Intra uterine growth retardation (IUGR) Definition: term infants with a birth-weight <2500 g Measurement tool: birth weight determined up to 10 min of delivery using a digital baby scale Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases of IUGR in maternity wards, controls selected from the same study base as the cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: explored 18 potential risk factors and adjusted for maternal work, hypertension, antenatal care Participation rate: not reported
| |
Tamim36 | Setting and period: six major hospitals in greater Beirut, Lebanon, August 2000 to August 2003 Population: 378 singleton newborns to women exclusive narghile smokers, 929 singleton newborns to women exclusive cigarette smokers, 84 singleton newborns to women smoking both types of tobacco, 7201 singleton newborns to women non-smokers
| Type: narghile, cigarettes Measurement tool: interviews, questionnaire pilot tested; narghile smokers categorized into ≤1 vs >1 per day; categorized into <1 vs ≥1 per day; and started smoking during first vs second or third trimester Measurement time points: not reported Exposure levels of included subjects: ≤1 per day (n = 233); >1 per day (n = 145); first trimester (n = 309); second or third trimester (n = 69)
| Low birth-weight Measurement tool: data collection from obstetric and nursery charts, low birth-weight defined as ≤2500 g Blinding of outcome adjudicator: not reported
| Selection bias: sample representative of populations of six major hospitals in Greater Beirut Information bias: objective outcome measurement: yes but based on medical records, standardized exposure measurement: no (pilot tested only) Confounding: adjustment for maternal/paternal education, mother’s working status, mother’s age, parity, diabetes, bleeding, pregnancy hypertensive disorders, gestational age, and passive smoking Participation rate: not reported
| OR for narghile only smoking compared with no smoking low birth-weight = 1.32 (95% CI 0.39–4.40) (OR pooled for ≤1 day and >1 day) low birth-weight = 0.7 (95% CI 0.3–1.6) (≤1 day) low birth-weight = 2.4 (95% CI 1.2–5.0) (>1 day) low birth-weight = 1.2 (95% CI 0.6–2.2) (narghile smoking started in first trimester) OR for cigarette only smoking compared with no smoking = 1.32 (95% CI 0.91–1.90)
|
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Nuwayhid19 | Setting and period: hospitals in Lebanon; 1993 and 1995 Participants: pregnant women delivering in hospitals. 106 narghile smokers (mean age 27.6 years), 277 cigarette smokers (mean age 29.1) and 512 non-smokers (mean age 28.5)
| Type of exposure: narghile, cigarettes Measurement tool: interviews, standardization not reported; categorized into <1 vs ≥1 per day; and started smoking during first vs second or third trimester Measurement time points: not reported Exposure levels of included subjects: <1 per day (n = 38); ≥1 per day (n = 67) per day; first trimester (n = 78); second or third trimester (n = 25)
| Low birth-weight, Apgar score <7 (1 and 5 min), pulmonary problems, malformations, perinatal complications | Selection bias: unclear whether sample is representative (data collection in a few number of hospitals) Information bias: objective outcome measurement: yes but based on medical records, standardized exposure measurement: no Confounding: adjustment for age, hospital location, gestational age, exposure to passive smoking, and previous low birth-weight Participation rate: not reported <10% missing data
| OR for narghile smoking compared with no smoking low birth-weight = 2.17 (95% CI 0.74–6.33) low birth-weight = 2.36 (95% CI 0.52–10.73) (<1/day) low birth-weight = 2.07 (95% CI 0.54–7.94) (≥1/day) low birth-weight = 2.93 (95% CI 0.97–8.83) (narghile smoking started in first trimester) Test for trend for low birth-weight not significant for intensity (P < 0.18) Apgar score <7 at 1 min = 1.73 (95% CI 0.73–4.14) Apgar score <7 at 5 min = 3.39 (95% CI 0.54–21.42) Pulmonary problems = 3.65 (95% CI 1.52–8.75) Malformations = 2.01 (95% CI 0.59–6.88) Perinatal complications =1.67 (95% CI 0.82–3.41) OR for cigarette smoking compared with no smoking Low birth-weight = 2.00 (95% CI 0.96–4.20) low birth-weight = 2.25 (95% CI 1.04–4.86) (narghile smoking started in first trimester) Apgar score <7 at 1 min = 1.59 (95% CI 0.82–3.07) Apgar score <7 at 5 min = 2.62 (95% CI 0.56–12.29) Pulmonary problems = 1.76 (0.80–3.87) Malformations = 1.36 (0.49–3.80) Perinatal complications = 1.50 (0.87–2.57)
|
Aghamolaei37 | Setting and period: Shariati Hospital of Bandar Abbas, South of Iran; period not reported Cases: 60 Intra-uterine growth retardation infants Controls: 60 normal birth infants Included subjects: all term infants with gestational age 37–42 weeks; 29 males and 31 females in each group. ‘None of included mothers smoked and used alcohol during pregnancy and none of them had diabetes’
| Type: hookah smoking by mother Measurement tool: structured questionnaire administered to mothers, standardization not reported Exposure levels of included subjects: not reported
| Intra uterine growth retardation (IUGR) Definition: term infants with a birth-weight <2500 g Measurement tool: birth weight determined up to 10 min of delivery using a digital baby scale Blinding of outcome adjudicator: not reported
| Selection bias: series of incident cases of IUGR in maternity wards, controls selected from the same study base as the cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: explored 18 potential risk factors and adjusted for maternal work, hypertension, antenatal care Participation rate: not reported
| |
Tamim36 | Setting and period: six major hospitals in greater Beirut, Lebanon, August 2000 to August 2003 Population: 378 singleton newborns to women exclusive narghile smokers, 929 singleton newborns to women exclusive cigarette smokers, 84 singleton newborns to women smoking both types of tobacco, 7201 singleton newborns to women non-smokers
| Type: narghile, cigarettes Measurement tool: interviews, questionnaire pilot tested; narghile smokers categorized into ≤1 vs >1 per day; categorized into <1 vs ≥1 per day; and started smoking during first vs second or third trimester Measurement time points: not reported Exposure levels of included subjects: ≤1 per day (n = 233); >1 per day (n = 145); first trimester (n = 309); second or third trimester (n = 69)
| Low birth-weight Measurement tool: data collection from obstetric and nursery charts, low birth-weight defined as ≤2500 g Blinding of outcome adjudicator: not reported
| Selection bias: sample representative of populations of six major hospitals in Greater Beirut Information bias: objective outcome measurement: yes but based on medical records, standardized exposure measurement: no (pilot tested only) Confounding: adjustment for maternal/paternal education, mother’s working status, mother’s age, parity, diabetes, bleeding, pregnancy hypertensive disorders, gestational age, and passive smoking Participation rate: not reported
| OR for narghile only smoking compared with no smoking low birth-weight = 1.32 (95% CI 0.39–4.40) (OR pooled for ≤1 day and >1 day) low birth-weight = 0.7 (95% CI 0.3–1.6) (≤1 day) low birth-weight = 2.4 (95% CI 1.2–5.0) (>1 day) low birth-weight = 1.2 (95% CI 0.6–2.2) (narghile smoking started in first trimester) OR for cigarette only smoking compared with no smoking = 1.32 (95% CI 0.91–1.90)
|
Periodontal disease
Of the five included studies,38–42 four were conducted in the same (or in a subgroup of the same) group of participants (Table 3).39–42 These four studies were cross-sectional and evaluated the same outcome (i.e. periodontal disease) using different outcome measurement (periodontal bone height loss, plaque index and gingivitis, deepening of the sulci or pockets, vertical periodontal bone loss). We did not pool the four related studies evaluating periodontal disease as their data were derived from the same participants. Their results were consistent in showing a statistically significant association of waterpipe tobacco smoking with periodontal disease (OR = 3–5). We judged the overall quality of evidence to be low ( Appendix 3).
Table 3 Characteristics of included studies assessing the effect of waterpipe tobacco smoking on periodontal disease
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Natto39 | Setting and period: Jeddah, Saudi Arabia, period not reported Participants: 355 volunteers, 28% females, mean age 36.9 years (range 17–60), having 20 teeth and not pregnant
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 56.8 run- years for waterpipe smokers
| Periodontal disease measured as periodontal bone height loss Measurement tool: digital panoramic radiographs. Bone loss defined as bone height ≤70%. Mean bone height per individual is the ratio of the periodontal bone height to the root length Blinding of outcome adjudicator: yes
| Selection bias: volunteers recruited by newspaper announcements Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: not reported
| OR compared with no smoking 3.5 (95% CI 1.6–7.6) (waterpipe only smoking) 1.0 (95% CI 0.3–3.1) (light waterpipe only smoking) 7.5 (95% CI 3.0–18.3) (heavy waterpipe only smoking) The association between waterpipe smoking and bone height remained statistically significant after controlling for ‘education as a surrogate for socio-economic standard and other variables’ Differences between light and heavy exposures were statistically significant (P < 0.001)
|
Natto40 | Setting and period: Western part of Saudi Arabia, period not reported Participants: 244 volunteers, 34% females, mean age 37.4 years (range 25–70), having 20 teeth and not pregnant
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt) Exposure levels of included subject: not reported
| Periodontal disease measured as plaque index and gingivitis Measurement tool: Plaque: clinical examination, presence of visible dental plaque was recorded according to the criteria of Silness and Loe Blinding of outcome adjudicator: not reported
| Selection bias: 244 of the 355 volunteers participating in another study38 volunteered for a clinical examination Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, dental care habit Participation rate: 68%
| |
Natto41 | Setting and period: Jeddah, Saudi Arabia, period not reported Population: 262 volunteers, 35% females, mean age 36.5 (range 17–60). Participants were required to have 20 teeth and not be pregnant. 51 citizens with periodontal disease and 211 citizens without periodontal disease
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 35.8 run- years for waterpipe smokers
| Periodontal disease measured as deepening of the sulci or pockets Measurement tool: clinical examination probing the depth of the sulci or pockets with a 2-mm graduated periodontal probe; periodontal disease defined as ≥10 sites with a probing depth ≥5 mm Blinding of outcome adjudicator: not reported
| Selection bias: 262 of the 355 volunteers participating in another study18 volunteered for a clinical examination Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: 74%
| OR compared with no smoking 5.1 (95% CI 2.0–13.5 (waterpipe only smoking) 8.2 (95% CI 2.9–22.9) (heavy waterpipe smoking) 2.9 (95% CI 1.7–4.8) (light waterpipe smoking) 3.8 (95% CI 1.3–11.0) (cigarette only smoking) 5.0 (95% CI 1.6–16.2), P < 0.001 (heavy cigarette only smoking) 1.5 (95% CI 1.1–2.0), P < 0.001 (light cigarette only smoking) Differences between light and heavy exposure were statistically significant
|
Baljoon42 | Setting and period: Jeddah, Saudi Arabia, period not reported Population: 103 volunteers with vertical periodontal bone loss and 159 individuals without vertical periodontal bone loss
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 56.8 run- years for waterpipe smokers
| Periodontal disease measured as vertical periodontal bone loss Measurement tool: intra-oral radiographs; vertical periodontal bone loss defined as one-sided bone resorption of the interdental marginal bone ≥2 mm that had a typical angulation towards either the mesial or distal aspect of the root Blinding of outcome adjudicator: yes
| Selection bias: 262 of the 355 volunteers participating in another study,51,52 volunteered to participate in this study Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: 74%
| OR compared with no smoking 2.9 (95% CI 1.2–7.0) (waterpipe only smoking) 43.3 (95% CI 12.1–71.6) (heavy waterpipe smoking) 0.6 (95% CI 0.3–1.4) (light waterpipe smoking) 6.6 (95% CI 2.6–17.1) (cigarette only smoking) 18.3 (95% CI 6.2–53.9), P < 0.001 (heavy cigarette only smoking) 1.3 (95% CI 0.5–3.4), P < 0.001 (light cigarette only smoking) Differences between light and heavy exposures were statistically significant (P < 0.001)
|
Al-Belasy38 | Setting and period: Mansoura Egypt, January 2000 to February 2002 Exposed: 100 male shisha smokers, mean age 28.7 years (range 22–39), 100 male cigarette smokers, mean age 28.7 years (range 20–38) Non-exposed: 100 males non-smokers, mean age 27.7 years (range 20–37) Included subjects were required to be healthy not taking medications at the time of the study and with unilateral high mesoangular impactions with exposed occlusal surfaces. Patients with recent antibiotic use or the medial need for antibiotic prophylaxis were excluded
| Type: shisha, cigarettes Measurement tool: self-report, standardization not reported Measurement time points: on day of the surgery, on the first and second post-operative days Exposure levels of included subjects: not reported
| Dry socket Measurement tool: clinical diagnosis on the basis of constant radiating pain not relieved by the analgesic, accompanied by a denuded socket or necrotic clot and a fetid smell Blinding of outcome adjudicator: not reported
| Selection bias: only healthy patients not taking medications were included, non-smokers selected from the same community as smokers Information bias: objective outcome measurement: yes, standardized exposure measurement: no; same duration follow-up for both groups Confounding: no matching or adjustment in the analysis reported Participation rate: not reported Attrition rate: not reported
| ‘Shisha smokers had three times the risk of non-smokers for developing dry socket’ (RR = 3.7; P = 0.001); trend of developing dry socket with frequency of shisha smoking was statistically significant (P = 0.001) Compared with non-smokers, shisha smokers who smoked the day of surgery or the first day after surgery had a significantly increased incidence of dry socket (day 0, P = 0.001; day 1, P = 0.005)
|
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Natto39 | Setting and period: Jeddah, Saudi Arabia, period not reported Participants: 355 volunteers, 28% females, mean age 36.9 years (range 17–60), having 20 teeth and not pregnant
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 56.8 run- years for waterpipe smokers
| Periodontal disease measured as periodontal bone height loss Measurement tool: digital panoramic radiographs. Bone loss defined as bone height ≤70%. Mean bone height per individual is the ratio of the periodontal bone height to the root length Blinding of outcome adjudicator: yes
| Selection bias: volunteers recruited by newspaper announcements Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: not reported
| OR compared with no smoking 3.5 (95% CI 1.6–7.6) (waterpipe only smoking) 1.0 (95% CI 0.3–3.1) (light waterpipe only smoking) 7.5 (95% CI 3.0–18.3) (heavy waterpipe only smoking) The association between waterpipe smoking and bone height remained statistically significant after controlling for ‘education as a surrogate for socio-economic standard and other variables’ Differences between light and heavy exposures were statistically significant (P < 0.001)
|
Natto40 | Setting and period: Western part of Saudi Arabia, period not reported Participants: 244 volunteers, 34% females, mean age 37.4 years (range 25–70), having 20 teeth and not pregnant
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt) Exposure levels of included subject: not reported
| Periodontal disease measured as plaque index and gingivitis Measurement tool: Plaque: clinical examination, presence of visible dental plaque was recorded according to the criteria of Silness and Loe Blinding of outcome adjudicator: not reported
| Selection bias: 244 of the 355 volunteers participating in another study38 volunteered for a clinical examination Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, dental care habit Participation rate: 68%
| |
Natto41 | Setting and period: Jeddah, Saudi Arabia, period not reported Population: 262 volunteers, 35% females, mean age 36.5 (range 17–60). Participants were required to have 20 teeth and not be pregnant. 51 citizens with periodontal disease and 211 citizens without periodontal disease
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 35.8 run- years for waterpipe smokers
| Periodontal disease measured as deepening of the sulci or pockets Measurement tool: clinical examination probing the depth of the sulci or pockets with a 2-mm graduated periodontal probe; periodontal disease defined as ≥10 sites with a probing depth ≥5 mm Blinding of outcome adjudicator: not reported
| Selection bias: 262 of the 355 volunteers participating in another study18 volunteered for a clinical examination Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: 74%
| OR compared with no smoking 5.1 (95% CI 2.0–13.5 (waterpipe only smoking) 8.2 (95% CI 2.9–22.9) (heavy waterpipe smoking) 2.9 (95% CI 1.7–4.8) (light waterpipe smoking) 3.8 (95% CI 1.3–11.0) (cigarette only smoking) 5.0 (95% CI 1.6–16.2), P < 0.001 (heavy cigarette only smoking) 1.5 (95% CI 1.1–2.0), P < 0.001 (light cigarette only smoking) Differences between light and heavy exposure were statistically significant
|
Baljoon42 | Setting and period: Jeddah, Saudi Arabia, period not reported Population: 103 volunteers with vertical periodontal bone loss and 159 individuals without vertical periodontal bone loss
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 56.8 run- years for waterpipe smokers
| Periodontal disease measured as vertical periodontal bone loss Measurement tool: intra-oral radiographs; vertical periodontal bone loss defined as one-sided bone resorption of the interdental marginal bone ≥2 mm that had a typical angulation towards either the mesial or distal aspect of the root Blinding of outcome adjudicator: yes
| Selection bias: 262 of the 355 volunteers participating in another study,51,52 volunteered to participate in this study Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: 74%
| OR compared with no smoking 2.9 (95% CI 1.2–7.0) (waterpipe only smoking) 43.3 (95% CI 12.1–71.6) (heavy waterpipe smoking) 0.6 (95% CI 0.3–1.4) (light waterpipe smoking) 6.6 (95% CI 2.6–17.1) (cigarette only smoking) 18.3 (95% CI 6.2–53.9), P < 0.001 (heavy cigarette only smoking) 1.3 (95% CI 0.5–3.4), P < 0.001 (light cigarette only smoking) Differences between light and heavy exposures were statistically significant (P < 0.001)
|
Al-Belasy38 | Setting and period: Mansoura Egypt, January 2000 to February 2002 Exposed: 100 male shisha smokers, mean age 28.7 years (range 22–39), 100 male cigarette smokers, mean age 28.7 years (range 20–38) Non-exposed: 100 males non-smokers, mean age 27.7 years (range 20–37) Included subjects were required to be healthy not taking medications at the time of the study and with unilateral high mesoangular impactions with exposed occlusal surfaces. Patients with recent antibiotic use or the medial need for antibiotic prophylaxis were excluded
| Type: shisha, cigarettes Measurement tool: self-report, standardization not reported Measurement time points: on day of the surgery, on the first and second post-operative days Exposure levels of included subjects: not reported
| Dry socket Measurement tool: clinical diagnosis on the basis of constant radiating pain not relieved by the analgesic, accompanied by a denuded socket or necrotic clot and a fetid smell Blinding of outcome adjudicator: not reported
| Selection bias: only healthy patients not taking medications were included, non-smokers selected from the same community as smokers Information bias: objective outcome measurement: yes, standardized exposure measurement: no; same duration follow-up for both groups Confounding: no matching or adjustment in the analysis reported Participation rate: not reported Attrition rate: not reported
| ‘Shisha smokers had three times the risk of non-smokers for developing dry socket’ (RR = 3.7; P = 0.001); trend of developing dry socket with frequency of shisha smoking was statistically significant (P = 0.001) Compared with non-smokers, shisha smokers who smoked the day of surgery or the first day after surgery had a significantly increased incidence of dry socket (day 0, P = 0.001; day 1, P = 0.005)
|
Table 3 Characteristics of included studies assessing the effect of waterpipe tobacco smoking on periodontal disease
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Natto39 | Setting and period: Jeddah, Saudi Arabia, period not reported Participants: 355 volunteers, 28% females, mean age 36.9 years (range 17–60), having 20 teeth and not pregnant
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 56.8 run- years for waterpipe smokers
| Periodontal disease measured as periodontal bone height loss Measurement tool: digital panoramic radiographs. Bone loss defined as bone height ≤70%. Mean bone height per individual is the ratio of the periodontal bone height to the root length Blinding of outcome adjudicator: yes
| Selection bias: volunteers recruited by newspaper announcements Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: not reported
| OR compared with no smoking 3.5 (95% CI 1.6–7.6) (waterpipe only smoking) 1.0 (95% CI 0.3–3.1) (light waterpipe only smoking) 7.5 (95% CI 3.0–18.3) (heavy waterpipe only smoking) The association between waterpipe smoking and bone height remained statistically significant after controlling for ‘education as a surrogate for socio-economic standard and other variables’ Differences between light and heavy exposures were statistically significant (P < 0.001)
|
Natto40 | Setting and period: Western part of Saudi Arabia, period not reported Participants: 244 volunteers, 34% females, mean age 37.4 years (range 25–70), having 20 teeth and not pregnant
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt) Exposure levels of included subject: not reported
| Periodontal disease measured as plaque index and gingivitis Measurement tool: Plaque: clinical examination, presence of visible dental plaque was recorded according to the criteria of Silness and Loe Blinding of outcome adjudicator: not reported
| Selection bias: 244 of the 355 volunteers participating in another study38 volunteered for a clinical examination Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, dental care habit Participation rate: 68%
| |
Natto41 | Setting and period: Jeddah, Saudi Arabia, period not reported Population: 262 volunteers, 35% females, mean age 36.5 (range 17–60). Participants were required to have 20 teeth and not be pregnant. 51 citizens with periodontal disease and 211 citizens without periodontal disease
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 35.8 run- years for waterpipe smokers
| Periodontal disease measured as deepening of the sulci or pockets Measurement tool: clinical examination probing the depth of the sulci or pockets with a 2-mm graduated periodontal probe; periodontal disease defined as ≥10 sites with a probing depth ≥5 mm Blinding of outcome adjudicator: not reported
| Selection bias: 262 of the 355 volunteers participating in another study18 volunteered for a clinical examination Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: 74%
| OR compared with no smoking 5.1 (95% CI 2.0–13.5 (waterpipe only smoking) 8.2 (95% CI 2.9–22.9) (heavy waterpipe smoking) 2.9 (95% CI 1.7–4.8) (light waterpipe smoking) 3.8 (95% CI 1.3–11.0) (cigarette only smoking) 5.0 (95% CI 1.6–16.2), P < 0.001 (heavy cigarette only smoking) 1.5 (95% CI 1.1–2.0), P < 0.001 (light cigarette only smoking) Differences between light and heavy exposure were statistically significant
|
Baljoon42 | Setting and period: Jeddah, Saudi Arabia, period not reported Population: 103 volunteers with vertical periodontal bone loss and 159 individuals without vertical periodontal bone loss
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 56.8 run- years for waterpipe smokers
| Periodontal disease measured as vertical periodontal bone loss Measurement tool: intra-oral radiographs; vertical periodontal bone loss defined as one-sided bone resorption of the interdental marginal bone ≥2 mm that had a typical angulation towards either the mesial or distal aspect of the root Blinding of outcome adjudicator: yes
| Selection bias: 262 of the 355 volunteers participating in another study,51,52 volunteered to participate in this study Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: 74%
| OR compared with no smoking 2.9 (95% CI 1.2–7.0) (waterpipe only smoking) 43.3 (95% CI 12.1–71.6) (heavy waterpipe smoking) 0.6 (95% CI 0.3–1.4) (light waterpipe smoking) 6.6 (95% CI 2.6–17.1) (cigarette only smoking) 18.3 (95% CI 6.2–53.9), P < 0.001 (heavy cigarette only smoking) 1.3 (95% CI 0.5–3.4), P < 0.001 (light cigarette only smoking) Differences between light and heavy exposures were statistically significant (P < 0.001)
|
Al-Belasy38 | Setting and period: Mansoura Egypt, January 2000 to February 2002 Exposed: 100 male shisha smokers, mean age 28.7 years (range 22–39), 100 male cigarette smokers, mean age 28.7 years (range 20–38) Non-exposed: 100 males non-smokers, mean age 27.7 years (range 20–37) Included subjects were required to be healthy not taking medications at the time of the study and with unilateral high mesoangular impactions with exposed occlusal surfaces. Patients with recent antibiotic use or the medial need for antibiotic prophylaxis were excluded
| Type: shisha, cigarettes Measurement tool: self-report, standardization not reported Measurement time points: on day of the surgery, on the first and second post-operative days Exposure levels of included subjects: not reported
| Dry socket Measurement tool: clinical diagnosis on the basis of constant radiating pain not relieved by the analgesic, accompanied by a denuded socket or necrotic clot and a fetid smell Blinding of outcome adjudicator: not reported
| Selection bias: only healthy patients not taking medications were included, non-smokers selected from the same community as smokers Information bias: objective outcome measurement: yes, standardized exposure measurement: no; same duration follow-up for both groups Confounding: no matching or adjustment in the analysis reported Participation rate: not reported Attrition rate: not reported
| ‘Shisha smokers had three times the risk of non-smokers for developing dry socket’ (RR = 3.7; P = 0.001); trend of developing dry socket with frequency of shisha smoking was statistically significant (P = 0.001) Compared with non-smokers, shisha smokers who smoked the day of surgery or the first day after surgery had a significantly increased incidence of dry socket (day 0, P = 0.001; day 1, P = 0.005)
|
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Natto39 | Setting and period: Jeddah, Saudi Arabia, period not reported Participants: 355 volunteers, 28% females, mean age 36.9 years (range 17–60), having 20 teeth and not pregnant
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 56.8 run- years for waterpipe smokers
| Periodontal disease measured as periodontal bone height loss Measurement tool: digital panoramic radiographs. Bone loss defined as bone height ≤70%. Mean bone height per individual is the ratio of the periodontal bone height to the root length Blinding of outcome adjudicator: yes
| Selection bias: volunteers recruited by newspaper announcements Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: not reported
| OR compared with no smoking 3.5 (95% CI 1.6–7.6) (waterpipe only smoking) 1.0 (95% CI 0.3–3.1) (light waterpipe only smoking) 7.5 (95% CI 3.0–18.3) (heavy waterpipe only smoking) The association between waterpipe smoking and bone height remained statistically significant after controlling for ‘education as a surrogate for socio-economic standard and other variables’ Differences between light and heavy exposures were statistically significant (P < 0.001)
|
Natto40 | Setting and period: Western part of Saudi Arabia, period not reported Participants: 244 volunteers, 34% females, mean age 37.4 years (range 25–70), having 20 teeth and not pregnant
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt) Exposure levels of included subject: not reported
| Periodontal disease measured as plaque index and gingivitis Measurement tool: Plaque: clinical examination, presence of visible dental plaque was recorded according to the criteria of Silness and Loe Blinding of outcome adjudicator: not reported
| Selection bias: 244 of the 355 volunteers participating in another study38 volunteered for a clinical examination Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, dental care habit Participation rate: 68%
| |
Natto41 | Setting and period: Jeddah, Saudi Arabia, period not reported Population: 262 volunteers, 35% females, mean age 36.5 (range 17–60). Participants were required to have 20 teeth and not be pregnant. 51 citizens with periodontal disease and 211 citizens without periodontal disease
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 35.8 run- years for waterpipe smokers
| Periodontal disease measured as deepening of the sulci or pockets Measurement tool: clinical examination probing the depth of the sulci or pockets with a 2-mm graduated periodontal probe; periodontal disease defined as ≥10 sites with a probing depth ≥5 mm Blinding of outcome adjudicator: not reported
| Selection bias: 262 of the 355 volunteers participating in another study18 volunteered for a clinical examination Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: 74%
| OR compared with no smoking 5.1 (95% CI 2.0–13.5 (waterpipe only smoking) 8.2 (95% CI 2.9–22.9) (heavy waterpipe smoking) 2.9 (95% CI 1.7–4.8) (light waterpipe smoking) 3.8 (95% CI 1.3–11.0) (cigarette only smoking) 5.0 (95% CI 1.6–16.2), P < 0.001 (heavy cigarette only smoking) 1.5 (95% CI 1.1–2.0), P < 0.001 (light cigarette only smoking) Differences between light and heavy exposure were statistically significant
|
Baljoon42 | Setting and period: Jeddah, Saudi Arabia, period not reported Population: 103 volunteers with vertical periodontal bone loss and 159 individuals without vertical periodontal bone loss
| Type: waterpipe, cigarettes Measurement tool: standardized questionnaire administered in interviews, standardization not reported; calculated as run-years (product of waterpipe runs per day with years of smoking; a run is the completion of the waterpipe smoking until the tobacco is burnt); heavy exposure was defined as ≥40 run-years Exposure levels of included subject: mean exposure of 56.8 run- years for waterpipe smokers
| Periodontal disease measured as vertical periodontal bone loss Measurement tool: intra-oral radiographs; vertical periodontal bone loss defined as one-sided bone resorption of the interdental marginal bone ≥2 mm that had a typical angulation towards either the mesial or distal aspect of the root Blinding of outcome adjudicator: yes
| Selection bias: 262 of the 355 volunteers participating in another study,51,52 volunteered to participate in this study Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age only Participation rate: 74%
| OR compared with no smoking 2.9 (95% CI 1.2–7.0) (waterpipe only smoking) 43.3 (95% CI 12.1–71.6) (heavy waterpipe smoking) 0.6 (95% CI 0.3–1.4) (light waterpipe smoking) 6.6 (95% CI 2.6–17.1) (cigarette only smoking) 18.3 (95% CI 6.2–53.9), P < 0.001 (heavy cigarette only smoking) 1.3 (95% CI 0.5–3.4), P < 0.001 (light cigarette only smoking) Differences between light and heavy exposures were statistically significant (P < 0.001)
|
Al-Belasy38 | Setting and period: Mansoura Egypt, January 2000 to February 2002 Exposed: 100 male shisha smokers, mean age 28.7 years (range 22–39), 100 male cigarette smokers, mean age 28.7 years (range 20–38) Non-exposed: 100 males non-smokers, mean age 27.7 years (range 20–37) Included subjects were required to be healthy not taking medications at the time of the study and with unilateral high mesoangular impactions with exposed occlusal surfaces. Patients with recent antibiotic use or the medial need for antibiotic prophylaxis were excluded
| Type: shisha, cigarettes Measurement tool: self-report, standardization not reported Measurement time points: on day of the surgery, on the first and second post-operative days Exposure levels of included subjects: not reported
| Dry socket Measurement tool: clinical diagnosis on the basis of constant radiating pain not relieved by the analgesic, accompanied by a denuded socket or necrotic clot and a fetid smell Blinding of outcome adjudicator: not reported
| Selection bias: only healthy patients not taking medications were included, non-smokers selected from the same community as smokers Information bias: objective outcome measurement: yes, standardized exposure measurement: no; same duration follow-up for both groups Confounding: no matching or adjustment in the analysis reported Participation rate: not reported Attrition rate: not reported
| ‘Shisha smokers had three times the risk of non-smokers for developing dry socket’ (RR = 3.7; P = 0.001); trend of developing dry socket with frequency of shisha smoking was statistically significant (P = 0.001) Compared with non-smokers, shisha smokers who smoked the day of surgery or the first day after surgery had a significantly increased incidence of dry socket (day 0, P = 0.001; day 1, P = 0.005)
|
The fifth included study was a cohort study with 7 days follow-up after surgical removal of mandibular third molars and evaluated the outcome of dry socket.38 Dry socket, or alveolar osteitis, is the most common complication following tooth extractions. It is caused by the dislodgement of the blood clot at the site of the tooth extraction, exposing underlying bone and nerves and causing increasing pain. The reported RR for the association of waterpipe tobacco smoking with dry socket was 3.7 (P = 0.001). We judged the overall quality of evidence to be low ( Appendix 3).
Infectious disease
We identified three studies on hepatitis C. The three cross-sectional studies were conducted in Egypt and restricted to male participants exposed to group waterpipe tobacco smoking (Table 4).22,43,44 The pooled OR for the association of group waterpipe tobacco with hepatitis C was 0.98 (95% CI 0.80–1.21; I2 = 0%) (Figure 5). We judged the overall quality of evidence to be very low ( Appendix 3).
Figure 5
Association between waterpipe tobacco smoking and hepatitis C infection
Table 4Characteristics of included studies assessing the effect of waterpipe tobacco smoking on hepatitis C infection
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Habib22 | Setting and period: Aghour El Soughra, a rural village in Nile Delta, in 1997 Population: male village inhabitants screened for Hepatitis C. Four-hundred and fifty-five tested positive for Hepatitis C and 1372 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: Group goza smoking (assessed in men only) Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Hepatitis C infection | Selection bias: sample was ‘systematically selected’ Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, sex, marital status, education, history of invasive medical procedures and dental procedures Participation rate: half of households selected 0.15% missing data Two subjects (goza smokers) excluded for age <20 years It is not clear how female subjects were dealt with in the regression analyses
| |
Medhat44 | Setting and period: Community in Upper Egypt, period not reported. Population: male village inhabitants screened for Hepatitis C. 308 tested positive for Hepatitis C and 2409 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: group goza smoking (assessed in men only) Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Hepatitis C infection Measurement tool: second generation Enzyme Immunoassay (Abbott HCV EIA 2.0). Of the 523 participants, 514 with positive anti-HCV test results were tested for HCV RNA by a one-step reverse transcriptase-polymerase chain reaction method Blinding of outcome adjudicator: not reported
| Selection bias: sample represents 62.8% of village inhabitants Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, injection history, hospital or clinical experiences, dental treatments, obstetric exposures (women ≥20 years), sharing razors within family (males only), circumcision, shaving at community barber (men only), ear piercing (females only), tattoo, cautery Participation rate: 6033 (53.7%) of the 11 227 village inhabitants were included in the study It is not clear how female subjects were dealt with in the regression analyses
| OR for group goza smoking compared with non-group goza smoking = 0.9 (0.4– 2.0) (males ≤30 years old) OR for group goza smoking compared with non-group goza smoking = 0.8 (0.5–1.2) (males >30 years old)
|
El-Sadawy43 | Setting and period: Urban and rural areas of Sharkia Governorate, Egypt; period not reported Population: male village inhabitants screened for hepatitis C. 217 tested positive for Hepatitis C and 565 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: Group goza smoking (assessed in men only) Measurement tool: specifically designed questionnaire; pilot tested before survey for reliability Exposure levels of included subjects: not reported
| Hepatitis C infection | Selection bias: used ‘stratified random sampling, which included systematic sampling Information bias: objective outcome measurement: yes; however, while 25.8% tested positive by MEIA, 7.66% test positive by PCR; standardized exposure measurement: no (pilot tested only) Confounding: unclear whether reported results were adjusted for potential confounders Participation rate not reported Missing data not reported
| |
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Habib22 | Setting and period: Aghour El Soughra, a rural village in Nile Delta, in 1997 Population: male village inhabitants screened for Hepatitis C. Four-hundred and fifty-five tested positive for Hepatitis C and 1372 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: Group goza smoking (assessed in men only) Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Hepatitis C infection | Selection bias: sample was ‘systematically selected’ Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, sex, marital status, education, history of invasive medical procedures and dental procedures Participation rate: half of households selected 0.15% missing data Two subjects (goza smokers) excluded for age <20 years It is not clear how female subjects were dealt with in the regression analyses
| |
Medhat44 | Setting and period: Community in Upper Egypt, period not reported. Population: male village inhabitants screened for Hepatitis C. 308 tested positive for Hepatitis C and 2409 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: group goza smoking (assessed in men only) Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Hepatitis C infection Measurement tool: second generation Enzyme Immunoassay (Abbott HCV EIA 2.0). Of the 523 participants, 514 with positive anti-HCV test results were tested for HCV RNA by a one-step reverse transcriptase-polymerase chain reaction method Blinding of outcome adjudicator: not reported
| Selection bias: sample represents 62.8% of village inhabitants Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, injection history, hospital or clinical experiences, dental treatments, obstetric exposures (women ≥20 years), sharing razors within family (males only), circumcision, shaving at community barber (men only), ear piercing (females only), tattoo, cautery Participation rate: 6033 (53.7%) of the 11 227 village inhabitants were included in the study It is not clear how female subjects were dealt with in the regression analyses
| OR for group goza smoking compared with non-group goza smoking = 0.9 (0.4– 2.0) (males ≤30 years old) OR for group goza smoking compared with non-group goza smoking = 0.8 (0.5–1.2) (males >30 years old)
|
El-Sadawy43 | Setting and period: Urban and rural areas of Sharkia Governorate, Egypt; period not reported Population: male village inhabitants screened for hepatitis C. 217 tested positive for Hepatitis C and 565 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: Group goza smoking (assessed in men only) Measurement tool: specifically designed questionnaire; pilot tested before survey for reliability Exposure levels of included subjects: not reported
| Hepatitis C infection | Selection bias: used ‘stratified random sampling, which included systematic sampling Information bias: objective outcome measurement: yes; however, while 25.8% tested positive by MEIA, 7.66% test positive by PCR; standardized exposure measurement: no (pilot tested only) Confounding: unclear whether reported results were adjusted for potential confounders Participation rate not reported Missing data not reported
| |
Table 4Characteristics of included studies assessing the effect of waterpipe tobacco smoking on hepatitis C infection
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Habib22 | Setting and period: Aghour El Soughra, a rural village in Nile Delta, in 1997 Population: male village inhabitants screened for Hepatitis C. Four-hundred and fifty-five tested positive for Hepatitis C and 1372 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: Group goza smoking (assessed in men only) Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Hepatitis C infection | Selection bias: sample was ‘systematically selected’ Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, sex, marital status, education, history of invasive medical procedures and dental procedures Participation rate: half of households selected 0.15% missing data Two subjects (goza smokers) excluded for age <20 years It is not clear how female subjects were dealt with in the regression analyses
| |
Medhat44 | Setting and period: Community in Upper Egypt, period not reported. Population: male village inhabitants screened for Hepatitis C. 308 tested positive for Hepatitis C and 2409 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: group goza smoking (assessed in men only) Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Hepatitis C infection Measurement tool: second generation Enzyme Immunoassay (Abbott HCV EIA 2.0). Of the 523 participants, 514 with positive anti-HCV test results were tested for HCV RNA by a one-step reverse transcriptase-polymerase chain reaction method Blinding of outcome adjudicator: not reported
| Selection bias: sample represents 62.8% of village inhabitants Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, injection history, hospital or clinical experiences, dental treatments, obstetric exposures (women ≥20 years), sharing razors within family (males only), circumcision, shaving at community barber (men only), ear piercing (females only), tattoo, cautery Participation rate: 6033 (53.7%) of the 11 227 village inhabitants were included in the study It is not clear how female subjects were dealt with in the regression analyses
| OR for group goza smoking compared with non-group goza smoking = 0.9 (0.4– 2.0) (males ≤30 years old) OR for group goza smoking compared with non-group goza smoking = 0.8 (0.5–1.2) (males >30 years old)
|
El-Sadawy43 | Setting and period: Urban and rural areas of Sharkia Governorate, Egypt; period not reported Population: male village inhabitants screened for hepatitis C. 217 tested positive for Hepatitis C and 565 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: Group goza smoking (assessed in men only) Measurement tool: specifically designed questionnaire; pilot tested before survey for reliability Exposure levels of included subjects: not reported
| Hepatitis C infection | Selection bias: used ‘stratified random sampling, which included systematic sampling Information bias: objective outcome measurement: yes; however, while 25.8% tested positive by MEIA, 7.66% test positive by PCR; standardized exposure measurement: no (pilot tested only) Confounding: unclear whether reported results were adjusted for potential confounders Participation rate not reported Missing data not reported
| |
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Habib22 | Setting and period: Aghour El Soughra, a rural village in Nile Delta, in 1997 Population: male village inhabitants screened for Hepatitis C. Four-hundred and fifty-five tested positive for Hepatitis C and 1372 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: Group goza smoking (assessed in men only) Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Hepatitis C infection | Selection bias: sample was ‘systematically selected’ Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, sex, marital status, education, history of invasive medical procedures and dental procedures Participation rate: half of households selected 0.15% missing data Two subjects (goza smokers) excluded for age <20 years It is not clear how female subjects were dealt with in the regression analyses
| |
Medhat44 | Setting and period: Community in Upper Egypt, period not reported. Population: male village inhabitants screened for Hepatitis C. 308 tested positive for Hepatitis C and 2409 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: group goza smoking (assessed in men only) Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Hepatitis C infection Measurement tool: second generation Enzyme Immunoassay (Abbott HCV EIA 2.0). Of the 523 participants, 514 with positive anti-HCV test results were tested for HCV RNA by a one-step reverse transcriptase-polymerase chain reaction method Blinding of outcome adjudicator: not reported
| Selection bias: sample represents 62.8% of village inhabitants Information bias: objective outcome measurement: yes, standardized exposure measurement: no Confounding: adjustment for age, injection history, hospital or clinical experiences, dental treatments, obstetric exposures (women ≥20 years), sharing razors within family (males only), circumcision, shaving at community barber (men only), ear piercing (females only), tattoo, cautery Participation rate: 6033 (53.7%) of the 11 227 village inhabitants were included in the study It is not clear how female subjects were dealt with in the regression analyses
| OR for group goza smoking compared with non-group goza smoking = 0.9 (0.4– 2.0) (males ≤30 years old) OR for group goza smoking compared with non-group goza smoking = 0.8 (0.5–1.2) (males >30 years old)
|
El-Sadawy43 | Setting and period: Urban and rural areas of Sharkia Governorate, Egypt; period not reported Population: male village inhabitants screened for hepatitis C. 217 tested positive for Hepatitis C and 565 tested negative for Hepatitis C (study also recruited females but restricted goza analysis to males)
| Type: Group goza smoking (assessed in men only) Measurement tool: specifically designed questionnaire; pilot tested before survey for reliability Exposure levels of included subjects: not reported
| Hepatitis C infection | Selection bias: used ‘stratified random sampling, which included systematic sampling Information bias: objective outcome measurement: yes; however, while 25.8% tested positive by MEIA, 7.66% test positive by PCR; standardized exposure measurement: no (pilot tested only) Confounding: unclear whether reported results were adjusted for potential confounders Participation rate not reported Missing data not reported
| |
We did not identify any eligible study assessing the association between waterpipe smoking and the transmission of tuberculosis. We did identify two reports of outbreak investigations suggesting an association between tuberculosis and sharing tobacco waterpipe23 and marijuana waterpipe.47
Respiratory illness
One cross-sectional study evaluated the association between waterpipe smoking and respiratory illness (defined as perennial rhinitis and including nasal congestion and wheezing) (Table 5).45 The study reported significant association of waterpipe tobacco smoking (OR = 2.3; 95% CI 1.1–5.1) and of waterpipe and/or cigarette smoking (OR = 2.5; 95% CI 1.6–3.8) with respiratory illness. We judged the overall quality of evidence to be very low ( Appendix 3).
Table 5Characteristics of included studies assessing the effect of waterpipe tobacco smoking on other outcomes
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Tamim45 | Setting and period: students in the second and third intermediate classes from five primary schools in Greater Beirut area, in April 2000 Population: 625 students with mean age 13 years (range 10–15). 143 students (85 males) reported respiratory tract symptoms, 482 students (237 males) reported no symptoms
| Type: second-hand exposure to cigarettes, narghile Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Respiratory illness throughout the year (not seasonal) including nasal congestion or wheezing | Selection bias: sample representative of populations of the five primary schools Information bias: objective outcome measurement: no, standardized exposure measurement: no Confounding: no matching or adjustment in the analysis reported Participation rate: not reported
| OR compared with no exposure at home 2.3 95% CI (1.1–5.1) (narghile only exposure) 2.5 (1.6–3.8) (narghile and/or cigarette exposure) 3.2 (1.9–5.4) (cigarette only exposure)
|
Inhortn46 Funding: National Science Foundation, the Fulbright Institute for International Education, the US Department of Education Fulbright-Hays Doctoral Dissertation Research Abroad Program and the Soroptimists International Founder Region Fellowship Program | Setting and period: infertility clinic of the University of Alexandria’s Shatby Hospital, the public obstetrics/gynaecology teaching hospital, rural areas of the northwestern Nile Delta region, Egypt; October 1988 to December 1989 Cases: husbands of 40 female patients of the university infertility clinic who were diagnosed of male-factor infertility (problems of semen and sperm) Controls: husbands of 90 fertile female patients of the university hospital Participants: 15–45 years old, lower and lower-middle socio-economic classes
| Type: husbands’ waterpipe and cigarette smoking behaviours Measurement tool: interviews, self-developed semi-structured questionnaire, no standardization reported; participants categorized as regular smokers and never smokers Exposure levels of included subjects: not reported
| Couple infertility associated with male-factor infertility | Selection bias: prevalent cases of infertility, controls were hospital-based and recruited from the same study base as cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no (second-hand information) Confounding: matching for age group, socio-economic class; adjustment for cigarette smoking, tea drinking, marital duration, husband’s age, husband’s education Participation rate: >98%
| OR compared with never smoking: 2.5 (1.0–6.3) (regular waterpipe smoking) |
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Tamim45 | Setting and period: students in the second and third intermediate classes from five primary schools in Greater Beirut area, in April 2000 Population: 625 students with mean age 13 years (range 10–15). 143 students (85 males) reported respiratory tract symptoms, 482 students (237 males) reported no symptoms
| Type: second-hand exposure to cigarettes, narghile Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Respiratory illness throughout the year (not seasonal) including nasal congestion or wheezing | Selection bias: sample representative of populations of the five primary schools Information bias: objective outcome measurement: no, standardized exposure measurement: no Confounding: no matching or adjustment in the analysis reported Participation rate: not reported
| OR compared with no exposure at home 2.3 95% CI (1.1–5.1) (narghile only exposure) 2.5 (1.6–3.8) (narghile and/or cigarette exposure) 3.2 (1.9–5.4) (cigarette only exposure)
|
Inhortn46 Funding: National Science Foundation, the Fulbright Institute for International Education, the US Department of Education Fulbright-Hays Doctoral Dissertation Research Abroad Program and the Soroptimists International Founder Region Fellowship Program | Setting and period: infertility clinic of the University of Alexandria’s Shatby Hospital, the public obstetrics/gynaecology teaching hospital, rural areas of the northwestern Nile Delta region, Egypt; October 1988 to December 1989 Cases: husbands of 40 female patients of the university infertility clinic who were diagnosed of male-factor infertility (problems of semen and sperm) Controls: husbands of 90 fertile female patients of the university hospital Participants: 15–45 years old, lower and lower-middle socio-economic classes
| Type: husbands’ waterpipe and cigarette smoking behaviours Measurement tool: interviews, self-developed semi-structured questionnaire, no standardization reported; participants categorized as regular smokers and never smokers Exposure levels of included subjects: not reported
| Couple infertility associated with male-factor infertility | Selection bias: prevalent cases of infertility, controls were hospital-based and recruited from the same study base as cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no (second-hand information) Confounding: matching for age group, socio-economic class; adjustment for cigarette smoking, tea drinking, marital duration, husband’s age, husband’s education Participation rate: >98%
| OR compared with never smoking: 2.5 (1.0–6.3) (regular waterpipe smoking) |
Table 5Characteristics of included studies assessing the effect of waterpipe tobacco smoking on other outcomes
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Tamim45 | Setting and period: students in the second and third intermediate classes from five primary schools in Greater Beirut area, in April 2000 Population: 625 students with mean age 13 years (range 10–15). 143 students (85 males) reported respiratory tract symptoms, 482 students (237 males) reported no symptoms
| Type: second-hand exposure to cigarettes, narghile Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Respiratory illness throughout the year (not seasonal) including nasal congestion or wheezing | Selection bias: sample representative of populations of the five primary schools Information bias: objective outcome measurement: no, standardized exposure measurement: no Confounding: no matching or adjustment in the analysis reported Participation rate: not reported
| OR compared with no exposure at home 2.3 95% CI (1.1–5.1) (narghile only exposure) 2.5 (1.6–3.8) (narghile and/or cigarette exposure) 3.2 (1.9–5.4) (cigarette only exposure)
|
Inhortn46 Funding: National Science Foundation, the Fulbright Institute for International Education, the US Department of Education Fulbright-Hays Doctoral Dissertation Research Abroad Program and the Soroptimists International Founder Region Fellowship Program | Setting and period: infertility clinic of the University of Alexandria’s Shatby Hospital, the public obstetrics/gynaecology teaching hospital, rural areas of the northwestern Nile Delta region, Egypt; October 1988 to December 1989 Cases: husbands of 40 female patients of the university infertility clinic who were diagnosed of male-factor infertility (problems of semen and sperm) Controls: husbands of 90 fertile female patients of the university hospital Participants: 15–45 years old, lower and lower-middle socio-economic classes
| Type: husbands’ waterpipe and cigarette smoking behaviours Measurement tool: interviews, self-developed semi-structured questionnaire, no standardization reported; participants categorized as regular smokers and never smokers Exposure levels of included subjects: not reported
| Couple infertility associated with male-factor infertility | Selection bias: prevalent cases of infertility, controls were hospital-based and recruited from the same study base as cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no (second-hand information) Confounding: matching for age group, socio-economic class; adjustment for cigarette smoking, tea drinking, marital duration, husband’s age, husband’s education Participation rate: >98%
| OR compared with never smoking: 2.5 (1.0–6.3) (regular waterpipe smoking) |
Study
. | Population
. | Exposure
. | Outcomes
. | Methodological features
. | Results
. |
---|
Tamim45 | Setting and period: students in the second and third intermediate classes from five primary schools in Greater Beirut area, in April 2000 Population: 625 students with mean age 13 years (range 10–15). 143 students (85 males) reported respiratory tract symptoms, 482 students (237 males) reported no symptoms
| Type: second-hand exposure to cigarettes, narghile Measurement tool: questionnaire, standardization not reported Exposure levels of included subjects: not reported
| Respiratory illness throughout the year (not seasonal) including nasal congestion or wheezing | Selection bias: sample representative of populations of the five primary schools Information bias: objective outcome measurement: no, standardized exposure measurement: no Confounding: no matching or adjustment in the analysis reported Participation rate: not reported
| OR compared with no exposure at home 2.3 95% CI (1.1–5.1) (narghile only exposure) 2.5 (1.6–3.8) (narghile and/or cigarette exposure) 3.2 (1.9–5.4) (cigarette only exposure)
|
Inhortn46 Funding: National Science Foundation, the Fulbright Institute for International Education, the US Department of Education Fulbright-Hays Doctoral Dissertation Research Abroad Program and the Soroptimists International Founder Region Fellowship Program | Setting and period: infertility clinic of the University of Alexandria’s Shatby Hospital, the public obstetrics/gynaecology teaching hospital, rural areas of the northwestern Nile Delta region, Egypt; October 1988 to December 1989 Cases: husbands of 40 female patients of the university infertility clinic who were diagnosed of male-factor infertility (problems of semen and sperm) Controls: husbands of 90 fertile female patients of the university hospital Participants: 15–45 years old, lower and lower-middle socio-economic classes
| Type: husbands’ waterpipe and cigarette smoking behaviours Measurement tool: interviews, self-developed semi-structured questionnaire, no standardization reported; participants categorized as regular smokers and never smokers Exposure levels of included subjects: not reported
| Couple infertility associated with male-factor infertility | Selection bias: prevalent cases of infertility, controls were hospital-based and recruited from the same study base as cases Information bias: objective outcome measurement: yes, standardized exposure measurement: no (second-hand information) Confounding: matching for age group, socio-economic class; adjustment for cigarette smoking, tea drinking, marital duration, husband’s age, husband’s education Participation rate: >98%
| OR compared with never smoking: 2.5 (1.0–6.3) (regular waterpipe smoking) |
Infertility
One case–control study evaluated the association between waterpipe smoking and male factor infertility (based on semen analysis) (Table 5).46 The reported OR for the association of waterpipe tobacco smoking with male factor infertility was not statistically significant (OR = 2.5; 95% CI 1.0–6.3). We judged the overall quality of evidence to be very low ( Appendix 3).
Discussion
We systematically reviewed the medical literature for the effects of waterpipe tobacco smoking on health outcomes. Based on the available evidence, waterpipe tobacco smoking was significantly associated with lung cancer, respiratory illness, low birth-weight and periodontal disease. It was not significantly associated with bladder cancer, nasopharyngeal cancer, oesophageal cancer, oral dysplasia or infertility but the CIs did not exclude important associations. Smoking waterpipe in groups was not significantly associated with hepatitis C infection. The overall quality of evidence varied from very low to low.
To our knowledge, this is the first meta-analysis of studies assessing the association of waterpipe smoking with health outcomes. While Knishkowy et al.24 reviewed the health effects of waterpipe tobacco smoking, they did not use systematic methods, formally evaluate the quality of the evidence or use statistical methods to pool the results across studies. The main strength of our study is the use of the Cochrane Collaboration methodology for conducting systematic reviews, i.e. using a very sensitive and comprehensive search strategy, a duplicate and independent selection process, a duplicate and independent data abstraction process, and a rigorous appraisal of the methodological quality of included studies. We also used the GRADE approach to rate the overall quality of evidence for each outcome.
For most of the outcomes of interest the evidence was either lacking, indirect or of lower quality. In terms of indirectness, lung cancer studies were conducted mostly in China and India where tobacco is typically unprocessed and burned directly by charcoal. The practice of waterpipe smoking that is involved in the recent global epidemic involves tobacco that is processed and flavoured and indirectly heated by the charcoal. In terms of quality, the one methodological study limitation that affected our rating of the quality of evidence for most outcomes was measurement bias, which has been reported in similar systematic reviews.48 In fact, only one study used a standardized exposure measurement tool in spite of the fact that the practice of waterpipe smoking can vary widely. Variables include the quantity of tobacco used, the type of tobacco used, the concomitant use of other substances, the frequency of smoking sessions, and the length of sessions, the number of years of smoking. In addition, no study reported using a standardized measurement tool for other forms of tobacco smoking in spite of the variety of these forms and the need to account for passive smoking and past smoking history.
The other methodological study limitation that affected our rating of the quality of evidence for many outcomes was the inappropriate handling of confounding, particularly for other forms of tobacco smoking and for factors such as radon exposure among miners as a risk factor for lung cancer. Furthermore, although many studies reported separate ORs for waterpipe and cigarette smoking, none reported tests for interactions between these two forms of smoking. This type of information is needed given that a significant proportion of waterpipe smokers also smoke cigarettes.11,49
In spite of the many methodological study limitations for lung cancer and pregnancy outcomes, the findings showed consistency (i.e. the low heterogeneity) across studies. Similarly, the lack of association between smoking waterpipe in groups and hepatitis C infection was characterized by the consistency across studies and a relatively narrow CI. These findings are consistent with current pathophysiological knowledge of the spread of hepatitis C by blood-to-blood contact. While the lack of evidence for an association between waterpipe smoking with bladder cancer, nasopharyngeal cancer, oral dysplasia and infertility could reflect a true absence of association, it could also be explained by lack of power to detect any existing association given their relatively wide CIs. Indeed, the lack of studies and of insufficient data on such long-term health effects is related to the short epidemiological time frame of the recent waterpipe epidemic.
The one study that reported on coronary heart disease was excluded because of its publication exclusively as an abstract.18 The quality of the evidence from that study was very low and the OR was 2.2 (95% CI 0.9–5.4) for ever smokers and 0.7 (95% CI 0.3–1.9) for current smokers compared with never smokers.
This study indicates the need for more research, as the World Health Organization (WHO) has already advised.50 The overall quality of evidence for the different outcomes was mostly low, limiting our confidence in the results. There is a need for high-quality cohort studies to identify and quantify with confidence all the health effects of waterpipe smoking and to explore its interaction with other forms of smoking. The high quality of future studies will depend on the use of standardized exposure measurement tools,49 another area of research need.
The study also has important implications for public health practice. Public health policy makers have to deal with suggestive but generally weak evidence of the association of waterpipe tobacco smoking with deleterious health outcomes. One approach could be to prioritize public health resources to deal with other public health crises (such as youth cigarette smoking) for which the evidence is much stronger, while awaiting further evidence regarding waterpipe smoking. Another approach could be to devote the necessary resources to control at a relatively early stage the rapid growth of the waterpipe trend. Whatever the decision policy makers make, it should be based on an objective consideration of the current available evidence in the proper public health context.
Acknowledgements
The authors thank Ms Ann Grifasi for her administrative assistance and Dr Ranime Saliba for her assistance. Those acknowledged have confirmed their agreement.
Conflict of interest: None declared.
Waterpipe tobacco smoking is possibly associated with a number of deleterious health outcomes including 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 of waterpipe tobacco smoking
References
1. The medical consequences of narghile (hookah, shisha) use in the world
,
Revue d Epidemiologie et de Sante Publique
,
2007
, vol.
55
(pg.
165
-
70
)
2American Lung Association
,
An Emerging Deadly Trend; Waterpipe Tobacco Use
,
2007
American Lung Association: Washington, DC
3, . Elements in hair and nails of residents from a village adjacent to New-Delhi—influence of place of occupation and smoking-habits
,
Biological Trace Element Res
,
1992
, vol.
34
(pg.
99
-
105
)
4, , , . Tobacco smoking using a waterpipe: a re-emerging strain in a global epidemic
,
Tobacco Control
,
2004
, vol.
13
(pg.
327
-
33
)
5, , , , , . Argileh smoking among university students: a tobacco epidemic
,
Nicotine Tobacco Res
,
2004
, vol.
6
(pg.
457
-
63
)
6, . Smoking habits among medical students in Central Saudi Arabia
,
Saudi Med J
,
2006
, vol.
27
(pg.
700
-
3
)
7, . Smoking among secondary-school boys in Bahrain: prevalence and risk factors
,
East Mediterr Health J
,
2003
, vol.
9
(pg.
78
-
86
)
8, , , et al. Smoking behavior and its influencing factors in a national-representative sample of Iranian adolescents: CASPIAN study
,
Preventive Med
,
2006
, vol.
42
(pg.
423
-
6
)
9, , , , , . Cigarette and waterpipe smoking among Lebanese adolescents, a cross-sectional study, 2003–2004
,
Nicotine Tobacco Res
,
2008
, vol.
10
(pg.
309
-
14
)
10, , . Is concern about waterpipe tobacco smoking warranted?
,
Aust N Z Publ Health
,
2008
, vol.
32
(pg.
181
-
2
)
11, . Waterpipe smoking in students: prevalence, risk factors, symptoms of addiction, and smoke intake
,
Evidence from one British university. BMC Public Health
,
2008
, vol.
8
pg.
174
12, . Canadian and English students' beliefs about waterpipe smoking: a qualitative study
,
BMC Public Health
,
2009
, vol.
9
pg.
10
13, , , . Waterpipe tobacco smoking on a US college campus: Prevalence and correlates
,
J Adolesc Health
,
2008
, vol.
42
(pg.
526
-
9
)
14, , , , . Comparison of cigarette and water-pipe smoking by Arab and non-Arab-American youth
,
Am J Prev Med
,
2008
, vol.
35
(pg.
334
-
9
)
15, , , et al. Tobacco use patterns among high school students: do Arab American youth differ?
,
Ethn Dis
,
2007
, vol.
17(2 Suppl 3)
(pg.
S3-22
-
S3-4
)
16, , , et al. Relation of radon exposure and tobacco use to lung cancer among tin miners in Yunnan Province, China
,
Am J Ind Med
,
1989
, vol.
16
(pg.
511
-
21
)
17, , , et al. Opium, tobacco, and alcohol use in relation to oesophageal squamous cell carcinoma in a high-risk area of Iran
,
Br J Cancer
,
2008
, vol.
98
(pg.
1857
-
63
)
18, , . Nargileh (Water-Pipe) smoking and incident coronary heart disease: a case-control study
,
Ann Epidemiol
,
2003
, vol.
13
pg.
570
19, , , et al. Narghile (hubble-bubble) smoking, low birth weight, and other pregnancy outcomes
,
Am J Epidemiol
,
1998
, vol.
148
(pg.
375
-
83
)
20, . Polycyclic aromatic hydrocarbons, carbon monoxide, “tar”, and nicotine in the mainstream smoke aerosol of the narghile water pipe
,
Food & Chem Toxicol
,
2005
, vol.
43
(pg.
655
-
61
)
21. Investigation of mainstream smoke aerosol of the argileh water pipe
,
Food Chem Toxicol
,
2003
, vol.
41
(pg.
143
-
52
)
22, , , et al. Hepatitis C virus infection in a community in the Nile Delta: risk factors for seropositivity
,
Hepatology
,
2001
, vol.
33
(pg.
248
-
53
)
23, , , . Tuberculosis and water pipes as source of infection
,
Ugeskr Laeger
,
2006
, vol.
168
(pg.
904
-
7
)
24, , , . Water-pipe (narghile) smoking: an emerging health risk behavior
,
Pediatrics
,
2005
, vol.
116
(pg.
e113
-
9
)
25, , . Interventions for waterpipe smoking cessation
,
Cochrane Database of Systematic Rev
,
2007
, vol.
4
pg.
CD005549
26, , , et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations
,
BMJ
,
2008
, vol.
336
(pg.
924
-
6
)
27, , , . Measuring inconsistency in meta-analysis
,
BMJ
,
2003
, vol.
327
(pg.
557
-
60
)
28, , , et al. Risk factors of lung cancer in Chandigarh, India
,
Indian J Med Res
,
2001
, vol.
113
(pg.
142
-
50
)
29, , , et al. Quantitative evaluation of the radon and lung cancer association in a case control study of Chinese tin miners
,
Cancer Res
,
1990
, vol.
50
(pg.
174
-
80
)
30, , , et al. Risk of lung cancer among cigarette and pipe smokers in southern China
,
Int J Cancer
,
1992
, vol.
51
(pg.
390
-
5
)
31, , . Facteurs etiologiques du cancer bronchique primitif en Tunisie
,
La Tunisie Medicale
,
1993
, vol.
71
(pg.
265
-
8
)
32, , , . Analysis of a historical cohort of Chinese tin miners with arsenic, radon, cigarette smoke, and pipe smoke exposures using the biologically based two-stage clonal expansion model
,
Radiat Res
,
2001
, vol.
156
(pg.
78
-
94
)
33, , , et al. Epidemiology of bladder cancer in Alexandria, Egypt: Tobacco smoking
,
Int J Cancer
,
1997
, vol.
73
(pg.
64
-
7
)
34, , , et al. Cannabis, tobacco and domestic fumes intake are associated with nasopharyngeal carcinoma in North Africa
,
Br J Cancer
,
2009
, vol.
101
(pg.
1207
-
12
)
35. Histopathologic changes in oral mucosa of Yemenis addicted to water-pipe and cigarette smoking in addition to takhzeen al-qat
,
Oral Surg Oral Med Oral Pathol Oral Radiol Endod
,
2007
, vol.
103
(pg.
e55
-
9
)
36, , , et al. , National Collaborative Perinatal Neonatal Network Beirut L
. Effect of narghile and cigarette smoking on newborn birthweight
,
BJOG
,
2008
, vol.
115
(pg.
91
-
7
)
37, , . Risk factors associated with Intrauterine Growth Retardation (IUGR) in Bandar Abbas
,
J Med Sci
,
2007
, vol.
7
(pg.
665
-
9
)
38. The relationship of “shisha” (water pipe) smoking to postextraction dry socket
,
J Oral Maxillofac Surg
,
2004
, vol.
62
(pg.
10
-
4
)
39, , . Tobacco smoking and periodontal bone height in a Saudi Arabian population
,
J Clin Periodontol
,
2005
, vol.
32
(pg.
1000
-
6
)
40, , , et al. Tobacco smoking and gingival health in a Saudi Arabian population
,
Oral Health Prev
,
2004
, vol.
2
(pg.
351
-
7
)
41, , . Tobacco smoking and periodontal health in a Saudi Arabian population
,
J Periodontol
,
2005
, vol.
76
(pg.
1919
-
26
)
42, , , et al. Smoking and vertical bone defects in a Saudi Arabian population
,
Oral Health Prev
,
2005
, vol.
3
(pg.
173
-
82
)
43, , , et al. Hepatitis C virus infection at Sharkia Governorate, Egypt: seroprevalence and associated risk factors
,
J Egypt Soc Parasitol
,
2004
, vol.
34(Suppl 1)
(pg.
367
-
84
)
44, , , et al. Hepatitis C in a community in Upper Egypt: risk factors for infection
,
Am J Trop Med Hyg
,
2002
, vol.
66
(pg.
633
-
8
)
45, , , et al. Exposure of children to environmental tobacco smoke (ETS) and its association with respiratory ailments
,
J Asthma
,
2003
, vol.
40
(pg.
571
-
6
)
46, . Ethnography, epidemiology and infertility in Egypt
,
Soc Sci Med
,
1994
, vol.
39
(pg.
671
-
86
)
47, , , et al. A cluster of tuberculosis associated with use of a marijuana water pipe
,
Int J Tuberc Lung Dis
,
2003
, vol.
7
(pg.
860
-
5
)
48, , , et al. Passive smoking and breast cancer in never smokers: prospective study and meta-analysis
,
Int J Epidemiol
,
2008
, vol.
37
(pg.
1069
-
79
)
49, , , . Standardizing questionnaire items for the assessment of waterpipe tobacco use in epidemiological studies
,
Public Health
,
2005
, vol.
119
(pg.
400
-
4
)
50WHO Study Group on Tobacco Product Regualation (TobReg)
Advisory Note
,
Waterpipe Tobacco Smoking: Health Effects, Research Needs and Recommended Actions by Regulators
,
2005
Geneva
WHO
51, , . Nargile (Hubble-Bubble) smoking-induced hand eczema
,
Int J Dermatol
,
2002
, vol.
41
(pg.
771
-
2
)
52, , , . Early invasive pulmonary aspergillosis in a leukemia patient linked to aspergillus contaminated marijuana smoking
,
Leukemia Lymphoma
,
2001
, vol.
42
(pg.
1433
-
7
)
Appendix 1 Electronic search strategies
MEDLINE (1950 onward)
Waterpipe*.mp.
“water pipe*”.mp.
shisha*.mp.
sheesha*.mp.
hooka*.mp.
huqqa*.mp.
guza*.mp.
goza*.mp.
narghil*.mp.
nargil*.mp.
arghil*.mp
argil*.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*.mp
argil*.mp
(hubbl* adj3 bubbl*).mp.
or/1-13
ISI the Web of Science
(waterpipe* OR “water pipe*” OR shisha* OR sheesha* OR hooka* OR huqqa* OR guza* OR goza* OR narghil* OR nargil* OR argil* OR arghil* OR (hubbl* SAME bubbl*)) AND (smoking OR smoke OR health OR disease OR cancer* OR malignan* OR lung* OR pulmonary OR heart OR cardiac OR vascular OR stroke) (in Title or Topic)
Appendix 2 GRADE approach for rating the quality of evidence
Quality is first assigned based on study design
. | Quality of evidence
. | Quality is lowered ifa
. | Quality is raised ifa
. |
---|
Randomized trials → | 4 = High | Limitations of design | Large effect |
−1 Serious | + 1 Large |
−2 Very serious | + 2 Very large |
| 3 = Moderate | Inconsistency | Dose response |
−1 Serious | + 1 Evidence of a gradient |
−2 Very serious |
Indirectness | All plausible confounding |
Observational study → | 2 = Low | −1 Serious | + 1 Would reduce a demonstrated effect, or |
−2 Very serious |
Imprecision | + 1 Would suggest a spurious effect when results show no effect |
−1 Serious |
| 1 = Very low | −2 Very serious |
Publication bias |
−1 Likely |
−2 Very likely |
Quality is first assigned based on study design
. | Quality of evidence
. | Quality is lowered ifa
. | Quality is raised ifa
. |
---|
Randomized trials → | 4 = High | Limitations of design | Large effect |
−1 Serious | + 1 Large |
−2 Very serious | + 2 Very large |
| 3 = Moderate | Inconsistency | Dose response |
−1 Serious | + 1 Evidence of a gradient |
−2 Very serious |
Indirectness | All plausible confounding |
Observational study → | 2 = Low | −1 Serious | + 1 Would reduce a demonstrated effect, or |
−2 Very serious |
Imprecision | + 1 Would suggest a spurious effect when results show no effect |
−1 Serious |
| 1 = Very low | −2 Very serious |
Publication bias |
−1 Likely |
−2 Very likely |
Quality is first assigned based on study design
. | Quality of evidence
. | Quality is lowered ifa
. | Quality is raised ifa
. |
---|
Randomized trials → | 4 = High | Limitations of design | Large effect |
−1 Serious | + 1 Large |
−2 Very serious | + 2 Very large |
| 3 = Moderate | Inconsistency | Dose response |
−1 Serious | + 1 Evidence of a gradient |
−2 Very serious |
Indirectness | All plausible confounding |
Observational study → | 2 = Low | −1 Serious | + 1 Would reduce a demonstrated effect, or |
−2 Very serious |
Imprecision | + 1 Would suggest a spurious effect when results show no effect |
−1 Serious |
| 1 = Very low | −2 Very serious |
Publication bias |
−1 Likely |
−2 Very likely |
Quality is first assigned based on study design
. | Quality of evidence
. | Quality is lowered ifa
. | Quality is raised ifa
. |
---|
Randomized trials → | 4 = High | Limitations of design | Large effect |
−1 Serious | + 1 Large |
−2 Very serious | + 2 Very large |
| 3 = Moderate | Inconsistency | Dose response |
−1 Serious | + 1 Evidence of a gradient |
−2 Very serious |
Indirectness | All plausible confounding |
Observational study → | 2 = Low | −1 Serious | + 1 Would reduce a demonstrated effect, or |
−2 Very serious |
Imprecision | + 1 Would suggest a spurious effect when results show no effect |
−1 Serious |
| 1 = Very low | −2 Very serious |
Publication bias |
−1 Likely |
−2 Very likely |
Factors that lower the quality of evidence
Limitations of design
Selection bias
Measurement bias for risk factor (validity of risk factor measurement instrument, blinding of risk status data collector and blinding of risk status adjudicator)
Measurement bias for outcome (validity of outcome measurement instrument, blinding of outcome data collector and blinding outcome adjudicator)
Inappropriate handling of confounding (stratification, adjustment)
Attrition (for cohort studies)
Inconsistency
Inconsistency in the estimates of the association across studies (i.e. heterogeneity) suggests true differences in underlying cause–effect relationship that may arise from differences in:
populations (e.g. different association in populations with differing other specific exposures or other genetic profiles);
risk factor (e.g. larger association with higher exposure or different associations with varying forms of the same exposure);
outcomes (e.g. different associations with different length of follow-up because of diminishing or increasing association with time).
When heterogeneity exists, but investigators fail to identify a plausible explanation, the quality of evidence should be downgraded by one or two levels, depending on the magnitude of the inconsistency in the results.Indirectness
The question being addressed by the authors of a systematic review is different from the available evidence regarding:
population (e.g. general population vs population of miners);
risk factor (e.g. cigarette use vs pipe use);
outcome (e.g. intra-uterine growth vs low birth-weight)
Imprecision
Results are imprecise when studies include relatively few patients and few events and thus have wide CIs around the estimate of the effect. The 95% CI (or alternative estimate of precision) around the pooled or best estimate of association includes both negligible association and appreciable benefit or appreciable harm.
Publication bias
Publication bias is a systematic underestimate or an overestimate of the underlying beneficial or harmful effect due to the selective publication of studies (publication bias). That is, investigators fail to report studies they have undertaken (typically those that show no effect).
Factors that increase the quality of evidence
Large magnitude of effect
When methodologically strong observational studies yield large or very large and consistent estimates of the magnitude of association, we may be confident about the results. The larger the magnitude of effect, the stronger becomes the evidence.
Dose–response gradient
A gradient dose–response may increase our confidence in the findings and thus enhance the assigned quality of evidence.
Plausible confounding, which would reduce a demonstrated effect
On occasion, all plausible confounders or biases from observational studies would result in an underestimate of the association estimate. If, for instance, only sick patients have an exposure, yet they still fare better, it is likely that the exposure effect is even larger than the data suggest.
Appendix 3 Rating the quality of evidence for different association using the GRADE framework
Outcome
. | Factors lowering the rating of quality
. | Factors raising the rating of quality
. | Final rating
. |
---|
Lung cancer diagnosis | Limitations of design: four of five studies did not appropriately deal with confounding | Dose response: two studies reported statistically significant dose–response relationships in terms of the total amount of exposure and duration of exposure | Very low |
Indirectness: most studies likely used unprocessed tobacco burned directly by charcoal |
Lung cancer mortality | Limitations of design: included study did not appropriately deal with confounding | None | Very low |
Indirectness: study likely used unprocessed tobacco burned directly by charcoal |
Bladder cancer diagnosis | Limitations of design: use of prevalent cases | None | Very low |
Imprecision |
Oesophageal cancer | None | Dose response: the included study reported statistically significant dose response relationship in terms of intensity of use | Low |
Imprecision |
Nasopharyngeal cancer | Limitations of design: investigators used matching and adjusted the analysis for a number of confounding factors but not for other types of smoking | None | Very low |
Imprecision |
Oral dysplasia | Limitations of design: not reporting dealing with confounding factors; also the analysis was stratified by chewing side (i.e. analysis of outcome occurring on the chewing side separate from the analysis of outcome occurring on the non-chewing side) | None | Very low |
Imprecision |
Pregnancy outcomes | None | None | Low |
Periodontal disease | Limitations of design: included studies partially dealt with confounding factors (three adjusted for age only and one adjusted for age and dental care habit) | Dose response: two studies found a statistically significant dose response relationship in terms of total amount of exposure | Low |
Dry socket | Limitations in design: study did not adequately deal with confounding factors | Dose response: statistically significant dose response relationship in terms of total amount of exposure | Low |
Hepatitis C infection | Limitations in design: study reported discrepancies between the results of MEIA and PCR | None | Very low |
Respiratory illness | Limitations in design: study did not adequately deal with confounding factors | None | Very low |
Infertility | None Imprecision | None | Very low |
Outcome
. | Factors lowering the rating of quality
. | Factors raising the rating of quality
. | Final rating
. |
---|
Lung cancer diagnosis | Limitations of design: four of five studies did not appropriately deal with confounding | Dose response: two studies reported statistically significant dose–response relationships in terms of the total amount of exposure and duration of exposure | Very low |
Indirectness: most studies likely used unprocessed tobacco burned directly by charcoal |
Lung cancer mortality | Limitations of design: included study did not appropriately deal with confounding | None | Very low |
Indirectness: study likely used unprocessed tobacco burned directly by charcoal |
Bladder cancer diagnosis | Limitations of design: use of prevalent cases | None | Very low |
Imprecision |
Oesophageal cancer | None | Dose response: the included study reported statistically significant dose response relationship in terms of intensity of use | Low |
Imprecision |
Nasopharyngeal cancer | Limitations of design: investigators used matching and adjusted the analysis for a number of confounding factors but not for other types of smoking | None | Very low |
Imprecision |
Oral dysplasia | Limitations of design: not reporting dealing with confounding factors; also the analysis was stratified by chewing side (i.e. analysis of outcome occurring on the chewing side separate from the analysis of outcome occurring on the non-chewing side) | None | Very low |
Imprecision |
Pregnancy outcomes | None | None | Low |
Periodontal disease | Limitations of design: included studies partially dealt with confounding factors (three adjusted for age only and one adjusted for age and dental care habit) | Dose response: two studies found a statistically significant dose response relationship in terms of total amount of exposure | Low |
Dry socket | Limitations in design: study did not adequately deal with confounding factors | Dose response: statistically significant dose response relationship in terms of total amount of exposure | Low |
Hepatitis C infection | Limitations in design: study reported discrepancies between the results of MEIA and PCR | None | Very low |
Respiratory illness | Limitations in design: study did not adequately deal with confounding factors | None | Very low |
Infertility | None Imprecision | None | Very low |
Outcome
. | Factors lowering the rating of quality
. | Factors raising the rating of quality
. | Final rating
. |
---|
Lung cancer diagnosis | Limitations of design: four of five studies did not appropriately deal with confounding | Dose response: two studies reported statistically significant dose–response relationships in terms of the total amount of exposure and duration of exposure | Very low |
Indirectness: most studies likely used unprocessed tobacco burned directly by charcoal |
Lung cancer mortality | Limitations of design: included study did not appropriately deal with confounding | None | Very low |
Indirectness: study likely used unprocessed tobacco burned directly by charcoal |
Bladder cancer diagnosis | Limitations of design: use of prevalent cases | None | Very low |
Imprecision |
Oesophageal cancer | None | Dose response: the included study reported statistically significant dose response relationship in terms of intensity of use | Low |
Imprecision |
Nasopharyngeal cancer | Limitations of design: investigators used matching and adjusted the analysis for a number of confounding factors but not for other types of smoking | None | Very low |
Imprecision |
Oral dysplasia | Limitations of design: not reporting dealing with confounding factors; also the analysis was stratified by chewing side (i.e. analysis of outcome occurring on the chewing side separate from the analysis of outcome occurring on the non-chewing side) | None | Very low |
Imprecision |
Pregnancy outcomes | None | None | Low |
Periodontal disease | Limitations of design: included studies partially dealt with confounding factors (three adjusted for age only and one adjusted for age and dental care habit) | Dose response: two studies found a statistically significant dose response relationship in terms of total amount of exposure | Low |
Dry socket | Limitations in design: study did not adequately deal with confounding factors | Dose response: statistically significant dose response relationship in terms of total amount of exposure | Low |
Hepatitis C infection | Limitations in design: study reported discrepancies between the results of MEIA and PCR | None | Very low |
Respiratory illness | Limitations in design: study did not adequately deal with confounding factors | None | Very low |
Infertility | None Imprecision | None | Very low |
Outcome
. | Factors lowering the rating of quality
. | Factors raising the rating of quality
. | Final rating
. |
---|
Lung cancer diagnosis | Limitations of design: four of five studies did not appropriately deal with confounding | Dose response: two studies reported statistically significant dose–response relationships in terms of the total amount of exposure and duration of exposure | Very low |
Indirectness: most studies likely used unprocessed tobacco burned directly by charcoal |
Lung cancer mortality | Limitations of design: included study did not appropriately deal with confounding | None | Very low |
Indirectness: study likely used unprocessed tobacco burned directly by charcoal |
Bladder cancer diagnosis | Limitations of design: use of prevalent cases | None | Very low |
Imprecision |
Oesophageal cancer | None | Dose response: the included study reported statistically significant dose response relationship in terms of intensity of use | Low |
Imprecision |
Nasopharyngeal cancer | Limitations of design: investigators used matching and adjusted the analysis for a number of confounding factors but not for other types of smoking | None | Very low |
Imprecision |
Oral dysplasia | Limitations of design: not reporting dealing with confounding factors; also the analysis was stratified by chewing side (i.e. analysis of outcome occurring on the chewing side separate from the analysis of outcome occurring on the non-chewing side) | None | Very low |
Imprecision |
Pregnancy outcomes | None | None | Low |
Periodontal disease | Limitations of design: included studies partially dealt with confounding factors (three adjusted for age only and one adjusted for age and dental care habit) | Dose response: two studies found a statistically significant dose response relationship in terms of total amount of exposure | Low |
Dry socket | Limitations in design: study did not adequately deal with confounding factors | Dose response: statistically significant dose response relationship in terms of total amount of exposure | Low |
Hepatitis C infection | Limitations in design: study reported discrepancies between the results of MEIA and PCR | None | Very low |
Respiratory illness | Limitations in design: study did not adequately deal with confounding factors | None | Very low |
Infertility | None Imprecision | None | Very low |
Published by Oxford University Press on behalf of the International Epidemiological Association © The Author 2010; all rights reserved.