Cardiovascular effects of waterpipe smoking: a systematic review and meta-analysis

Waterpipe smoking has developed into a major and rapidly growing global tobacco epidemic affecting more than 100 million people worldwide. This study identifies and analyzes comprehensively all available data on the cardiovascular effects of waterpipe smoking. Databases PubMed, EMBASE, Web of Science, and the Cochrane Library were searched for studies published until December 2019 assessing cardiovascular effects of waterpipe smoking. We included experimental, cohort, cross-sectional and case-control studies and excluded systematic reviews, case reports/series and qualitative studies. Studies not conducted in humans or not distinguishing waterpipe smoking from other forms of smoking were also excluded. A total of 42 studies with 46 cardiovascular parameters were eligible for analysis. The meta-analysis included 31 studies with 38,037 individuals. Results showed that one waterpipe smoking session leads to immediate increases in heart rate and blood pressure (P < 0.001). Compared to non-smokers, waterpipe smokers had significantly lower high-density lipoprotein levels (P < 0.001), higher levels of low-density lipoprotein (P = 0.04), triglyceride (P < 0.001) and fasting blood glucose (P = 0.03) and higher heart rate (P = 0.04) with a tendency to have higher blood pressure. Mean heart rate, blood pressure and lipids levels did not differ between waterpipe and cigarette smokers, except for total cholesterol, being higher among waterpipe smokers (P < 0.001). Current level of evidence suggests that waterpipe smoking is associated with substantial adverse effects on cardiovascular system, which seem to be similar to those of cigarette smoking. Longitudinal studies are required to scrutinize the magnitude of these effects.


Introduction
Waterpipe smoking (WPS) has developed into a major and rapidly growing global tobacco epidemic worldwide . Several factors have hindered recognition of the harmful effects of WPS especially the flavored smoke and the fashionable aspect. In addition, the water gives the impression that it clears most of harmful substances in the smoke. A single WPS session lasts 30-90 minutes of uninterrupted smoking, producing a large volume of smoke which contains up to 80 times more of toxicants than those which found in the smoke of a single cigarette and carried through the water in the bubbles, so the assumption of users that the smoke is "filtered" by the water is incorrect (El-Zaatari et al., 2015;Shihadeh, 2003). Potential adverse cardiovascular effects of WPS have been reported in several scattered studies with varied results based on different estimation criteria. However, the overall clinical effect of WPS on the cardiovascular system is not clear yet. The few available systematic reviews on this topic were not specifically focused on cardiovascular outcomes or synthesis of the results were not performed, being of limited validity (El-Zaatari et al., 2015;Marshall et al., 2018;Rezk-Hanna and Benowitz, 2019;Waziry et al., 2018). We aimed to explore the clinical cardiovascular effects of WPS quantitatively and qualitatively and to compare them with those of cigarette smoking (CS) by combining all available related data in order to provide a better understanding of the relationship between WPS and cardiovascular disease (CVD) risk.

Study protocol and eligibility criteria
The analyses were conducted according to a prespecified, nonregistered protocol. A systematic literature search was performed for all published original studies on adults with no limitation on the number of participants. The exposure of interest was WPS and the outcomes of interest were any cardiovascular parameters. We included experimental studies assessing the acute effects of WPS and observational studies (i.e. cohort studies, cross-sectional studies and case-control studies) that compared waterpipe smokers to cigarette smokers or non-smokers. Systematic reviews, case reports/case series and qualitative studies were considered not appropriate for inclusion. Furthermore, studies not conducted in humans or not distinguishing WPS from other smoking forms were also excluded (Fig. 1).

Search and selection strategy
Until December 2019, two investigators (R.A., D.V.) independently searched PubMed, EMBASE, Web of Science, and the Cochrane Library using free search text terms combined with Medical Subject Heading (MeSH). The search was performed using variant terms and spellings for waterpipe used in different languages in the title or abstract, or as keywords. Eligibility assessment was performed independently by two investigators (R.A., L.K.). Disagreements were resolved by consulting a third investigator (M.B.).

Data extraction
One investigator extracted the following data from included studies and a second investigator checked the extracted data: 1) study characteristics including study design and settings and sample size; 2) characteristics of study participants including age, gender, health status, smoking status and settings; 3) data concerning outcome measurements for each cardiovascular parameter.

Risk of bias assessment
Risk of bias was assessed by two investigators independently (R.A., L.K.) using ROBINS-I (Sterne et al., 2016), a tool recommended from Cochrane Bias Methods Group for assessing the risk of bias in non-randomized interventions. In case of disagreements a third investigator (M.B.) was consulted.

Statistical analysis
Each analysis was conducted by pooling the appropriate data which could be extracted from at least three studies that we considered to be sufficiently similar in their design and comparison groups. Data from experimental studies were used to assess the acute effects of WPS. Combining data from observational studies and baseline data from multigroup experimental studies which have similar control groups were used to assess the non-acute effects of WPS by comparing waterpipe smokers to non-or cigarette smokers. Using a random-effects model, we pooled data from each study including sample size with both average value and standard deviation to estimate pooled mean difference or with prevalence rate to estimate pooled odds ratio. Results for tested parameters were presented using Forest plots along with their respective 95% confidence interval (CI). Relevant statistical heterogeneity was considered as Cochran's Q test P < 0.05 and I 2 > 50%. As sensitivity analyses, relevant statistical heterogeneity was removed using a previously described "sequential algorithm'' approach (Patsopoulos et al., 2008) based on repeating meta-analysis after dropping the outlying study that is responsible for the largest decrease in I2. Funnel plots for parameters reported in > 10 studies and eventual asymmetry was interpreted as the presence of publication bias. All P values were 2-sided, with P < 0.05 considered as significant. All statistical analyses were performed using Review Manager version 5.3 software. We presented summary and descriptive statistics of those parameters, where quantitative (meta-) analysis would not be appropriate.

Search results
After duplicates were removed, the literature search identified 2,141 studies, of which 149 studies investigated WPS effects. We excluded literature reviews, qualitative studies, case reports/series, and studies conducted among teenagers or on animals or machines. Finally, a total of 42 studies were eligible for inclusion in the systematic review. These included 23 observational (case-control and cross-sectional cohort) studies, which compared waterpipe to nonor cigarette smokers, and 19 experimental studies, all of which were non-randomized and investigated the acute effects of WPS, mostly using one-group pretest-posttest design (Fig. 1). One or more potential longitudinal effects of regular WPS were reported in three of these identified studies (Al Suwaidi et al., 2012;Wolfram et al., 2003;Wu et al., 2013).

Study characteristics
In all experimental studies each parameter was measured at least twice; at start and end of a 15-90-minute WPS session after 12-72-hour smoking abstinence. In the observational studies, waterpipe smokers were in general those who reported regular WPS, mostly twice a week at least. Time since the last WPS session before measurements was rarely reported. Most the included participants were healthy male adults especially in the experimental and case-control studies, except for 7 studies which conducted on patients with coronary heart disease (CHD) ( Table. 1). Generally, the search yielded 46 cardiovascular parameters.

Statistical analysis and study quality
Meta-analysis was conducted on 13 parameters including a total of 31 studies with 38037 participants. Most of this analysis showed a significant heterogeneity, which we considered by pooling the data using a random effects model. Based on ROBINS-I criteria we judged 29 studies with moderate and only 2 studies with serious risk of bias. According to the funnel plots there were no signs of publication bias. To assess the acute effect of WPS on each of heart rate (HR), systolic blood pressure (SBP) and diastolic blood pressure (DBP), meta-analysis was conducted for each of these parameters including 18, 14 and 13 experimental studies with 1814, 1460 and 1386 participants; respectively. Results showed that one WPS session led to acute increases in mean HR by 10.14 beat/min (95% CI: 8.41 to 11.88; P < 0.001), SBP by 7.70 mmHg (95% CI: 5.13 to 10.27; P < 0.001) and DBP by 4.86 mmHg (95% CI: 2.94 to 6.78; P < 0.001) (Fig. 2). After removing statistical heterogeneity in sensitivity analyses, the acute effect of WPS on these three hemodynamic parameters remained significant.

Vascular function
To investigate the potential acute effect of WPS on vascular function, different clinical methods were used in three experimental (Alomari et al., 2014;Bentur et al., 2014;Rezk-Hanna et al., 2019) and two case-control (Diab et al., 2015;Selim et al., 2013a) studies. One WPS session led to a significantly acute reduction in arterial blood flow by -8.8% (P = 0.035) and increase in arterial vascular resistance by 16% (P = 0.003) using strain-gauge plethysmography (Alomari et al., 2014), while no acute changes were observed in arterial pulse wave amplitude using "EndoPat'' device (Bentur et al., 2014). However, an acute reduction in flow-mediated dilatation (FMD) by 28% was observed after one WPS session (P < 0.001) in a recent study, where the effect of charcoal combustion was removed when the waterpipe-tobaccoproduct was heated electrically (Rezk-Hanna et al., 2019).

Non-acute effects of waterpipe smoking 3.4.2.1 Hemodynamic parameters 3.4.2.1.1 Heart rate and blood pressure
Meta-analysis was conducted using data from 10 studies with 14909 participants for HR, 12 studies with 17386 participants for SBP and DBP and 5 studies with 5742 participants for having hypertension. Results showed an increased mean HR by 2.12 (95% CI: 0.11 to 4.13; P = 0.04) with a tendency to have higher blood pressure (BP) among waterpipe smokers compared to nonsmokers (Fig. 3). No significant association between WPS and any of these hemodynamic parameters was observed in meta-analysis after eliminating the statistical heterogeneity in sensitivity analyses. However, a large population-based cross-sectional study with 14310 participants (Al-Safi et al., 2009) showed a significant correlation between hemodynamic status and frequency of WPS. Compared to non-smokers, waterpipe smokers of 1-2 sessions/week had higher mean HR by < 1 beat/min (P < 0.01), SBP by 1 mmHg (P < 0.002) and DBP by 1 mmHg (P < 0.009); and those who smoked > 4 sessions/week had even higher HR by 3 beat/min, SBP by 7 mmHg and DBP by 4 mmHg (P < 0.001 for all comparisons).

Factor-VII and VIII
Factor-VII and VIII were investigated in one study (Muddathir et al., 2018), showing higher levels by about 25% and 50%, respectively, in waterpipe smokers than in non-smokers, especially in those who smoked waterpipe for more than 3 years (P < 0.01).

Other coagulation factors
No significant long-term effect has been reported on Ser-pine1/Plasminogen activator inhibitor-1 (Khan et al., 2020).

Oxidative status, inflammation, and vascular function 3.4.2.4.1. Oxidative status and inflammation
One study investigated the longitudinal effect of WPS on the oxidative status (Wolfram et al., 2003). Blood levels of malondialdehyde, 11-DH-TXB2 and 8-epi-PGF2a continued to rise over two weeks of daily WPS and remained significantly elevated even before the initiation of a new WPS session.

Vascular function
Two case-control studies evaluated the potential non-acute effect of regular WPS on FMD, where lower FMD values by about 30-60% were observed in waterpipe smokers than in non-smokers (P < 0.05) (Diab et al., 2015;Selim et al., 2013a), with a negative correlation between FMD and both smoking duration in years (P < 0.001) and number of daily sessions (P < 0.001) (Selim et al., 2013a).

Fasting blood glucose
Our meta-analysis showed that waterpipe smokers had a higher mean fasting blood glucose (FBG) of 4.66 mg/dl (95% CI: 0.53 to 8.80; P = 0.03) than non-smokers did, with no association between WPS and having diabetes mellitus (DM). After removing the statistical heterogeneity in sensitivity analyses, a significant correlation was revealed between WPS and both of DM (OR = 1.35; 95% CI: 1.16 to 1.57; P < 0.001) and mean FBG.

2015)
, but no significant correlation was observed in a hospital based cross-sectional study (Selim et al., 2013b).

Severity of coronary artery disease
Two cross-sectional studies assessed the correlation between WPS and the severity of coronary artery disease (CAD). Compared to non-smokers, waterpipe smokers had a higher mean Duke Jeopardy score (DJS) of anatomical extension of CAD (P < 0.05), with a positive correlation between DJS and the duration of smoking in years (r = 0.574, P < 0.001) (Selim et al., 2013b). Waterpipe smokers with > 40 waterpipe-years were more likely to have severe coronary stenosis (OR = 2.94; 95% CI: 1.04-8.33) with a higher Duke CAD prognostic index (ß = 7.835, P = 0.027) than non-smokers did (Sibai et al., 2014).

Lipoproteins
Pooled data from three observational studies with 5721 participants showed higher mean total cholesterol of 6.80 mg/dl (95% CI: 3.23 to 10.38; P < 0.001) in waterpipe than in cigarette smokers. No differences were found regarding levels of LDL-cholesterol, HDL-cholesterol or triglyceride or having dyslipidemia (Fig. 6). However, a significant correlation between WPS and triglyceride level was revealed after removing the statistical heterogeneity in sensitivity analyses.

Oxidative status, inflammation, and vascular function
Meta-analysis showed no difference regarding FMD between waterpipe and cigarette smokers. Furthermore, the three studies that compared CRP levels (Diab et al., 2015;Khan et al., 2020;Saffar Soflaei et al., 2018) and the two studies that compared levels of malondialdehyde and TAC (Khan et al., 2020;Koubaa et al., 2015a) observed no differences between waterpipe and cigarette smokers.

Coagulation factors
In two of the three case-control studies (Hashem Sezavar et al., 2004;Khan et al., 2020;Muddathir et al., 2018), waterpipe smokers had significantly higher levels of fibrinogen than cigarette smokers did (P < 0.05) (Hashem Sezavar et al., 2004;Muddathir et al., 2018). The one study which investigated the levels of factor-VII and VIII (Muddathir et al., 2018) found no significant difference between waterpipe and cigarette smokers.

Discussion
The main findings of our analyses are: i) WPS leads to an acute increase in HR, SBP and DBP; ii) waterpipe smokers have increased HR, higher triglyceride and LDL-Cholesterol and lower HDL-Cholesterol levels compared to non-smokers; iii) the cardiometabolic profile in waterpipe smokers is not less worse than in cigarette smokers.
It is well known that increased HR and BP, the two most widely used hemodynamic parameters in assessment of cardiovascular system, negatively affect cardiovascular outcome. (Ettehad et al., 2016;Nikolovska Vukadinović et al., 2017). According to our results one WPS session causes acute increase in HR by about 11 beats/minute, SBP by 7 mmHg und DBP by 5 mmHg. This itself may lead to increased oxygen demand of the heart, augment shear stress of the blood vessel, which in some cases may provoke ACS, thereby increasing morbidity and mortality. Based on available data, it is not known how long these acute hemodynamic effects of WPS might last. The answer to this question needs more studies with serial measurements of these parameters. As waterpipe is mostly consumed regularly for several times a week, it could be expected that accumulation of these acute adverse effects negatively impacts prognosis over time. However, our results show that waterpipe smokers had slightly increased HR in comparison with non-smokers for about 2 beats/minute, while SBD and DBP tend to be higher among waterpipe smokers but without reaching the statistical significance. These results differ somewhat unexpectedly from those observed with the acute effects of WPS, which could be partially explained through large heterogeneity across the studies. Furthermore, frequency and duration of WPS sessions and years of smoking were not adjusted across the studies, which might affect the results. A significant positive correlation was previously observed between the number of weekly use of waterpipe and each of SBP, SBP and HR (Al-Safi et al., 2009). The acute hemodynamic changes revealed in our analyses may be attributed to some extent to nicotine exposure, which augments the sympathetic nervous system activity, leading to increases in HR, myocardial contractility and cardiac output (Salahuddin et al., 2012). Such an effect has been reported in the three cross-over design studies (Blank et al., 2011;Cobb et al., 2012;Shishani et al., 2014) comparing flavor-matched tobacco-with tobacco-free-WPS. However, an acute cardiac autonomic dysregulation was observed after a WPS session independently of nicotine content (Cobb et al., 2012). In addition, the high levels of exposed CO during WPS (Eissenberg and Shihadeh, 2009) can lead to decreased oxygen supply to tissues including the heart due to the formation of CO-Hb (Benowitz, 2003). In turn, it has been well established that hypoxia is a potent stimulator of several autonomic mechanisms leading to increases in resting HR, BP and cardiac output (Vigo et al., 2010). These findings contradict harm reduction claims of so-called``herbal'' waterpipe-products and correspond to outcomes of non-clinical studies on such products using a human-mimic waterpipe-machine (Hammal et al., 2015). Owing to the lack of data from longitudinal studies, it is not possible to determine to what extent WPS can be hemodynamically harmful at the long-term.
Pooling data from available studies revealed a significant correlation between WPS and increased triglyceride and LDLcholesterol and decreased HDL-cholesterol, which are recognized as CVD risk factors that promote atherosclerosis (Pedro-Botet et al., 2020). As is well known for CS, the mechanisms responsible are not clearly elucidated. However, the triglyceride/high-density lipoprotein abnormalities have recently been suggested to be related to insulin resistance. (Ambrose and Barua, 2004). This can be supported with our results that showed a significant increase in FBG in waterpipe smokers compared to non-smokers.
The association between CS and increased activity of coagulation factors and thrombotic risk has been previously reported (Eliasson, 1995). Likewise, WPS also correlates with increased levels of fibrinogen and factors VII and VIII (Hashem Sezavar et al., 2004;Khan et al., 2020;Muddathir et al., 2018), which may increase thrombogenicity, thus increasing the risk for cardiovascular events. The harmful effects of WPS are reflected by the increased CAC score (Chami et al., 2019) and the acute (Alomari et al., 2014;Rezk-Hanna et al., 2019;Wolfram et al., 2003) and long-term (Al-Numair et al., 2007;Diab et al., 2015;Ghasemi et al., 2010;Koubaa et al., 2015a;Selim et al., 2013a) endothelial dysfunction, which were established among waterpipe smokers, providing clinical evidence for the potential contribution of WPS to vascular disease. Our findings on WPS effects on cardiovascular system explain and support results of studies which reported a significant correlation between WPS and each of incidence (Al-Amri et al., 2019;Islami et al., 2013;Jabbour et al., 2003;Platt et al., 2017), worse clinical outcomes (Al Suwaidi et al., 2012) and estimated prognoses (Selim et al., 2013b;Sibai et al., 2014) of CVD.
As cardiovascular effects of CS are well known (Ambrose and Barua, 2004;Leone, 2003;Mons et al., 2015), the comparison between WPS and CS is of a great importance. Unfortunately, fewer studies could be included for this comparison. The lack of studies reporting the rate of cardiovascular and cerebrovascular events in waterpipe vs cigarette smokers may be the main limitation. However, based on our results, the non-acute effects of WPS on the vast majority of cardiovascular parameters of interest seem similar to those produced by CS. The few available studies showed no clear difference in the CVD incidence between WPS and CS (Al Suwaidi et al., 2012;Saffar Soflaei et al., 2018). Furhtermore, CVD complications (Al Suwaidi et al., 2012) and mortality (Al Suwaidi et al., 2012) tend to be of higher rates in waterpipe than in cigarette smokers. This might be explained due to prolonged WPS and, therefore, cumulatively higher amounts of inhaled toxic substances and consequently adverse effects on the cardiovascular system (Sibai et al., 2014).

Strengths and limitations
A limitation of these analyses is the high heterogeneity of the available data. This can be attributed to the geographic diversity of the analyzed studies, the multiple study protocols, as well as different frequency and session duration of WPS. Furthermore, some parameters were reported in only few studies, which limits the explanatory power of this analysis. Therefore, some results should be interpreted with caution. As long-term follow-up studies on cardiovascular effects of WPS are scarce, chronic effect of WPS has been estimated in our analyses using data from available case-control and cross-sectional cohort studies that compared waterpipe smokers to non-smokers. Of note, the observed effects could be explained solely due to WPS, as limited number of waterpipe smokers may occasionally have also smoked cigarettes. Although the comparison between WPS and CS may be the most important part of systematic review, only few studies are available for this comparison, with the main limitation is the lack of studies reporting the rate of cardiovascular and cerebrovascular events. On the other hand, some waterpipe smokers could be ex-cigarette smokers. The time spent smoking cigarette are likely impacting, observed outcomes. Such information is missed in most observational studies. Thus, a meta-regression considering time from quitting of cigarette smoking in waterpipe smokers is not possible. This applies to the other results showed a worst cardiometabolic profile of waterpipe smokers compared to non-smokers, as many studies did not consider all potential confounders in the comparisons. To the best of our knowledge, our study provides the most comprehensive image of the potential cardiovascular effects of WPS based on scientific evidence and reflects the current magnitude of research efforts performed so far regarding this issue.

Conclusions
There is still a widespread believe that WPS is harmless and not real smoking. This wide-ranged systematic review and metaanalysis outlines the spectrum of acute and long-term cardiovascular effects of waterpipe smoking. Despite all the stated limitations, current level of evidence suggests that WPS is associated with substantial adverse effects on cardiovascular system, which seem to be similar to what reported for cigarette smoking. Further research is required especially longitudinal studies evaluating the long-term consequences of both waterpipe smoking and cessation to scrutinize the magnitude of these effects and to provide a strong evidence for a causal relationship.