The health impact of smokeless tobacco products: a systematic review

Introduction The objective was to systematically review studies on health outcomes from smokeless tobacco (SLT) products. Methods We analysed published literature on the health outcomes from SLT use between 01/01/2015 to 01/02/2020, following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol using PubMed, Embase, Scopus, and Google Scholar. Results Of 53 studies included, six were global, 32 from Asia, Middle East and Africa (AMEA), nine from USA and six from Europe. ‘Poor’-rated studies predominated (23;43%), in particular, for global (4;66%) and AMEA (16;50%). Health outcomes differed between SLT-products and regions; those in AMEA were associated with higher mortality (overall, cancer, Coronary heart disease (CHD), respiratory but not cardiovascular disease (CVD)), and morbidity (CVD, oral and head and neck cancers), with odds ratios up to 38.7. European studies showed no excess mortality (overall, CVD, from cancers) or morbidity (ischemic heart disease (IHD), stroke, oral, head and neck, pancreatic or colon cancers) from several meta-analyses; single studies reported elevated risk of rectal cancer and respiratory disorders. Pooled study data showed protection against developing Parkinson’s disease. US studies showed mixed results for mortality (raised overall, CHD, cancer and smoking-related cancer mortality; no excess risk of respiratory or CVD mortality). Morbidity outcomes were also mixed, with some evidence of increased IHD, stroke and cancer risk (oral, head and neck). No studies reported on switching from cigarettes to SLT-products. Conclusion Our review demonstrates stark differences between different SLT-products in different regions, ranging from zero harm from European snus to greatly increased health risks in AMEA. The literature on the safety profile for SLT-products for harm reduction is incomplete and potentially misinforming policy and regulation. Supplementary Information The online version contains supplementary material available at 10.1186/s12954-021-00557-6.


Introduction
The use of SLT-products exceeds that of all other forms of tobacco use in some parts of the world. The prevalence of SLT-product use in men is 30% in India, 6% in Iceland [1], and 20% in Sweden [2]. SLT is rising in parts of Europe and some have attributed its use to the concomitant reduction in smoking prevalence [3][4][5].
There are numerous types of SLT-products available globally which differ markedly in terms of their preparation, method of use and toxicity. [6] Key features of some of the most common SLT-products are detailed in "Appendix 1". Although there has been no clear consensus on safety profiles of SLT-products, it is generally accepted that they pose a lower health risk than cigarettes. Despite the many differences described above, SLT-products are often regarded together as a single product and safety concerns have resulted in varying regulations and bans on sales and use globally. The objective of this systematic review was to identify, narratively synthesize, assess the strength and quality of evidence, and critically appraise studies that report health outcomes associated with use of different SLT-products in different regions of the world.

Methods
We conducted a systematic review of published literature on the health impact of SLT-products between January 1, 2015, and February 1, 2020. SLT-products included all types including snus, chewing tobacco, snuff and other products included in Table 1 ("Appendix 1"). For the purpose of this review, we reported findings according to three geographical regions, which best align with different types of SLT-products consumed, namely Europe (EU), the Americas (USA), and SE Asia, Eastern Mediterranean and Africa (AMEA) regions. The study followed PRISMA guidelines for reporting systematic reviews [7]. We included health outcomes of new onset or control of disease end-points. We did not include other health outcomes such as short-term physiological changes which do not necessarily manifest as disease or quality of life or in vitro effects.

Search strategy and eligibility criteria
A literature search was conducted between October 1, 2019, and February 26, 2020, using the databases Pub-Med, Embase, Scopus, and Google Scholar using medical subject headings.

Table 1 Types of SLT products by World Health Organization region
African Region (AFR), Region of the Americas (AMR), South-East Asian Region (SEAR), European Region (EUR), Eastern Mediterranean Region (EMR), Western Pacific Region (WPR) There were two domains: one for SLT-products and one for health outcomes, specifically CVD, cancer, respiratory, mortality and 'other' health outcomes. Search terms included "Smokeless tobacco" OR "smokeless tobacco product" OR "chewing tobacco"OR "reduced risk tobacco"OR "non-cigarette tobacco" OR "snus" OR "snuff " AND "health outcome" OR "morbidity" OR "mortality" OR "cancer"OR "cardiovascular disease"OR "chronic obstruct pulmonary disease" OR "COPD" OR "CVD" OR "acute myocardial infarction"OR "stroke" OR "cardiovascular" OR "cerebrovascular"OR "health effects"OR "adverse" OR "effects" OR "respiratory".
Search results were filtered to include English language, human studies and studies published from 01/01/2015 until 01/02/2020, in order to capture current product types and their changing pattern of use. The health outcomes of interest such as mortality, cancer and CVD, can take many years to develop and manifest and would still have been captured from use of historical SLT products.
The references of relevant reviews were manually searched for additional eligible citations.
Titles, abstracts and full texts of the search results were sequentially screened by two reviewers independently for inclusion, using the eligibility criteria below, with disagreements resolved via blind review by a third reviewer. Figure 1 shows the inclusion and exclusion criteria used. Reasons for excluding studies are shown in Fig. 2.

Data extraction and quality assessment
For included studies, data were extracted including author, year, country, aim, study design, sample size, participants and relevant findings such as effect sizes and nature of impact on health outcomes. Studies were categorized by region including global, AMEA, USA and EU. A level of evidence category was assigned using the Oxford Centre for Evidence Based Medicine framework [8] and a similar approach used to categorise methodological quality as "good", "fair" or "poor" utilizing the National Institutes for Health (NIH) Quality Assessment Inclusion criteria were: Exclusion criteria were: • Meta-analyses/pooled data, randomized controlled trials, cohort, cross-sectional, ecological, casecontrol and case studies reporting primary or secondary quantitative data • Human in vivo studies • English language articles • Representative samples or clinical subgroups (e.g., patients with asthma, patients with high blood pressure) • Studies examining impact on health outcomes defined as a disease endpoint or impact on a disease end-point (e.g., disease control), including mortality, cardiovascular or respiratory disease, cancer or 'other' health outcomes • Studies not presenting novel data (e.g., commentaries, letters, reviews, consensus statements and institutional reports) as these may have led to biased selection from a handful of countries • Animal, in vitro and in silico studies, or studies examining constituents of SLT (e.g., carcinogens or toxins) because these do not necessarily translate to disease end-points in humans • Non-English language articles due to reasons of feasibility • Studies published before 1 st January 2015 • Studies examining biomarkers, intermediate markers, risk factors for disease or short term physiological changes (e.g., heart rate, blood pressure (BP), levels of carcinogens) rather than disease end-points, in non-disease situations, e.g. BP in non-diseased participants as they are not indicative of long-term disease outcomes. Only studies that reported on disease endpoints or control of a condition, such as hypertension, myocardial infarction etc., were included. Tools [9]. The NIH quality assessment tools include features to assess risk of bias, such as selection and reporting bias, with a "good" rating reflecting a low risk of bias, and a "poor" rating suggesting a high risk of bias. Data extraction and synthesis was performed by two reviewers independently with blind assessment by a third reviewer for cases with rater disagreement. Findings of all studies were independently reviewed, coded and compared between studies to identify relationships and themes. We considered a meta-analysis of studies included in our review to be inappropriate, partly due to the common methodological flaws and the vast heterogeneity between studies. As such no statistical testing was required, only narrative reporting of study findings.

Results
Of the 53 studies included, six included global data, 32 were exclusively from AMEA, nine exclusively from USA, and six exclusively from Europe. The number of studies by study design and health outcomes are shown in Table 2.
Two studies reported on benefits from SLT-product use; a cross-sectional study on hypertension and a meta-analysis on Parkinson's disease in Europe. Table 4 provides a summary of study design, key outcomes, level of evidence, and quality rating for the included studies by region. Additional file 1: Table 5 provides more detailed findings of each study.
A "poor"-rated meta-analysis of 32 global studies estimated 1.7 million disability-adjusted life years (DALYs) lost and 62,283 deaths in 2010 globally from cancers of the mouth, pharynx and oesophagus attributed to SLTproduct [14]. Most included studies adjusted for but didn't exclude smoking.

Cardiovascular outcomes
One 'good'-rated meta-analyses of 19 studies on chewing tobacco, dip, snuff and snus from Sweden, North America, and Asia found no increase in IHD for combined regions (OR = 1.4; 95% CI 0.92-1.42) in studies that excluded former smokers [12].

Other cancers
A meta-analysis of 16 global studies found combined SLT products were associated with mortality due to cancers overall (HR = 1.31; 95% CI 1. 16 [13].

Table 3 Quality Ratings Assigned to Studies by Outcome
Totals may reflect duplicated studies that examined more than one health outcome. Quality assigned as "good", "fair" or "poor" utilizing NIH Quality Assessment Tools [8].

Health outcome
Good quality Fair quality Poor quality        [16].
A 'poor'-rated Indian meta-analysis of 25 studies reported increased oral cancer risk with combined SLTproducts (OR = 5.65; 95% CI 3.83-8.40) [22]. A 'poor'rated meta-analysis of Shammah use in Middle East and North Africa [23], in which only one of three studies adjusted for smoking, reported elevated risk of oral cancer (OR = 38.7; 95% CI 19.50-76.96).
A case-control study from Nepal reported an association between chewing tobacco and HNC (OR = 2.39; 95% CI 1.77-3.23), higher with heavy use (≥ 6 times per day) (OR = 2.91; 95% CI 2.06-4.12) and duration over 20 years (OR = 2.92; 95% CI 2.08-4.11) [36]. One study described the commonest sites for chewing tobacco related HNC cancer as the gingivobuccal complex [33]. An ecological analysis of regional population-based cancer registries in India found correlations for Khaini use and hypopharynx cancer (r = 0.48 males, r = 0.29 females), gutka use and mouth cancer in males (r = 0.54, r = − 0.19 for females) and oral tobacco and mouth cancer in males and females (r = 0.46 males, r = 0.17 females) [35] 'Other' types of SLT-product use (combined) correlated with hypopharynx cancer (r = 0.47). The study did not account for smoking.

Mortality
A large US study constituting a high level of evidence pooling two longitudinal studies found no increase in mortality overall or due to smoking-related cancers or CVD in never smoking SLT-product users compared with never-smoking never-SLT-product users [43]. Dual users of SLT-product and cigarettes had similar excess mortality (HR = 2.21; 95% CI 1.50-3.26-HR = 2.14; 95% CI 1.27-3.59) to exclusive smokers (non-SLT-product users) (HR = 2.10; 95% CI 1.99-2.22-HR = 1.88; 95% CI 1.75-2.02), compared with never tobacco users.

Oral cancer
A 'poor'-rated review that pooled 11 US studies found SLT-products (snuff and chewing tobacco) to be associated with cancers of oral cavity (OR = 1.81; 95% CI 1.04, 3.17) [49]. A large 'poor'-rated global meta-analysis found no association with oral cancer in North American data [14].

Head and neck cancer
The largest study investigating HNC, a 'poor'-rated global MA [14], reported for pharyngeal and oesophageal cancers, respectively: associations for all countries combined (OR = 2.23; 95% CI 1.55-3.20; OR = 2.17; 95% CI 1.70-2.78) but not for North America (single study only). A review that pooled 11 US studies found increased odds for HNC in snuff users (OR = 1.71; 95% CI 1.08-2.70) but not for ever-tobacco chewers, compared with never users [49]. with a dose-response effect with increasing duration of snuff use (p-value for trend = 0.007).

Other cancers
There were no exclusive US data on other cancers.

Other health outcomes
One US cross-sectional study reported no significant association between SLT-product use and a diagnosis of mental health disease or depression [50].

Oral cancer
There were no region-specific studies of oral cancer in Europe. A 'poor'-rated global meta-analysis showed no association between combined SLT-products oral cancer in Sweden or Norway [14]. A 'poor'-rated meta-analysis on 37 global case-control and cohort studies found no association between snus and moist snuff use and oral cancer in European data [53].

Head and neck cancer
A 'poor'-rated meta-analysis of combined SLT-product use showed no association with pharyngeal cancer but excess risk of oesophageal cancer (OR = 1.26; 95% CI 1.02-1.56) in Sweden and in a single study from Norway (OR = 1.40; 95% CI 0.61-3.21) [14].

Other cancer
A large 'fair'-rated review of nine Swedish cohort studiesfound no association with colorectal cancer for current (HR = 1.22; 95% CI 0.91, 1.64) or former exclusive snus users (HR = 1.12; 95% CI 0.75, 1.67); no association with colon cancer (HR = 1.02; 95% CI 0.81, 1.29) in current exclusive snus users but increased risk of rectal cancer in current snus users (HR = 1.38; 95% CI 1.07, 1.77) in never-smokers, with no dose-response effect for quantity or duration [51]. No association was found with pancreatic cancer pooling the same Swedish cohort studies [54].

Discussion
This is one of the first articles to systematically review health outcomes from SLT product use, and in particular, to differentiate between the different types of products used in Asia, Middle East and Africa, Sweden, other parts of Europe and the US.
Most studies were from AMEA and were less likely to be of rigorous study design than those from Europe and the USA. Two-thirds of global studies and a half of US studies evaluated mortality (66%; 50%), whereas AMEA studies mostly evaluated cancer (23; 72%). Meta-analyses made up 100% of global studies and 57% of Europe studies. Case-control represented 50% of AMEA studies.
Methodological flaws with the greatest impact included combining different SLT-products as seen in the global meta-analyses [10][11][12][13][14]53], and widespread failure to adequately account for dual and former cigarette smoking.

Health outcomes
Results indicate stark differences for health outcomes for different SLT-products and regions. There is overwhelming evidence that SLT-products in AMEA are associated with harmful health outcomes, including higher mortality: strongly for overall, cancer, CHD; less so for respiratory mortality and not shown to increase overall CVD mortality; increased CVD morbidity, with strong associations for IHD and stroke, and mixed evidence for hypertension and dyslipidaemia. Different SLT-products, even within the same region, have varied strengths of association with oral cancer, with odds ratios ranging from 29 to 39 for shammah; 23 for naswar, 11 for supari, 5.5 gutkha, 8.5 for chewing tobacco and 3.8 for tokomak dipping compared to nonuse. All types of SLT-products used in AMEA were associated with head and neck cancers albeit with lower odds than for oral cancer, of up to 3.2.
In stark contrast, the fewer but higher-quality studies in Europe, predominantly in Sweden, found snus and other SLT-products not to cause higher mortality or morbidity overall or from overall mortality, CVD or cancers. Two high quality meta-analyses showed no excess mortality, although one smaller cohort study contradicted this finding. Five meta-analyses found no excess IHD risk, and four found no excess stroke risk. There was no excess oral or head and neck cancers, pancreatic or colon cancer, but raised risk of rectal cancer in one study [51] and harms to respiratory disease from snus use [56]. There was robust evidence from pooled studies for a protective effect of snus against the development of Parkinson's disease (by more than 50%) [55]. The differences in detrimental health outcomes seen between snus users in Sweden and other parts of Europe compared to elsewhere may in part be attributable to the different chemical content [57].
US studies showed more mixed results from SLTproduct use with some evidence of harmful health outcomes. Meta-analyses and longitudinal studies showed mixed results for overall mortality, and mortality due to CHD, overall cancer and smoking-related cancers but no excess risk of respiratory or CVD mortality. Risk of nonfatal CVD were also mixed but the most rigorous study reported elevated risk for both IHD and stroke [47]. A single cross-sectional study reported reduced hypertension rates in SLT-product users. There were mixed results for oral and head and neck cancers ranging from no excess risk to a pooled odds ratio of 1.8 [49].
No studies of more novel products such as tobacco-free nicotine pouches were captured. Of the 53 studies, none reported on the health impact of switching from cigarettes to SLT-products.

Levels of evidence, quality and study design
No studies were above 2a for level of evidence [8]. There were no meta-analyses, pooled studies, or indeed individual interventional studies, which perhaps reflects difficulty conducting these in real world settings. Metaanalyses comprised the most common study design (21 studies); despite being large, including over 30 studies [14,53] and 350,000 participants [54,55], only five of the 21 meta-analyses rated as 'good' [10,12,21,47,55]. Particularly problematic themes included pooling different SLT-products, failing to account for heterogeneity of studies, pooling studies despite variation in sampling methodologies, and failing to report country-specific results, even when these were available.
Case-control and cross-sectional studies also predominated, both which are problematic in terms of accounting for bias, such as failing to account for temporality of exposure and outcome, as well as former smoking status, rendering cross-sectional studies inappropriate for causal inferences. Two-thirds of global and half of AMEA region studies were rated as being of 'poor' quality; all studies exclusively from Europe and two-thirds of those from USA were rated as 'good' or 'fair' .

Definitions of exposures
Studies frequently failed to account for quantity and duration of SLT-product use, dual and former use of cigarettes, and in former smokers, duration since quitting. Standard definitions exist for smoking that consider both quantity and duration [58] and similar approaches should be used for SLT-products. Furthermore, a strong dose response effect has been demonstrated in several studies for both quantity and duration of SLT-product use in AMEA, which should form part of the measurement of exposure.

Accounting for smoking status
Indian SLT-product users often smoke concurrently [59,60] and it is essential for both dual and former cigarette use to be accounted for when investigating health outcomes. Of snus use in Sweden, 82% were former or dual users of cigarettes [61]. In our review, only nine studies accounted for both former and current smoking, four out of 11 studies in USA and Europe, and six out of 49 studies from global and AMEA regions.

Publication bias
No formal evidence for publication bias was found in many of the meta-analyses in our review. However, the small number of studies investigating SLT-products in Sweden, Europe and US suggests that this is an underresearched area and the preponderance of reporting on negative outcomes could indicate the presence of publication bias.

Role of SLT-products in reducing smoking rates
SLT-product use in India represents two-thirds of all global SLT use [62] with prevalence rates of 30% in men and 13% in women, exceeding those for cigarettes (7% men, 0.6% women) and bidis (14% men, 1.2% women).
The use of snus by smokers has been associated with decreased cigarette smoking and increased abstinence of smoking [63][64][65][66][67][68][69]. Other studies do not support some of these findings [68,70,71]. Some have postulated snus use in Sweden has led to low smoking prevalence rates through a "reverse gateway" effect [69]. The low prevalence of smoking in favour of snus use in Sweden compared to the rest of Europe may have contributed to its lower rates of tobacco-attributable deaths (72/100,000 Sweden, 128/100,000 EU) and cancer-specific deaths (14/100,000 Sweden, 36/100,000 EU) in men in 2019 [72]. This strengthens the argument for safer forms of SLT-products such as Swedish snus to be used as a form of tobacco harm reduction on the pathway to stopping smoking. Indeed, data from Swedish longitudinal studies show in primary smokers who started secondary snus use, 10.6% reduced to occasional smoking and 76.3% stopped smoking altogether [5]. Furthermore, between 40 and 50% of secondary snus users later also quit snus use (during 7 years of follow up) [5,74], Modelling has suggested switching from smoking to Swedish snus is likely to result in net health gains [74].

Informing Policy
The findings of our review have implications for policy makers. SLT-products are subject to regulations with regard to sales restrictions, advertising, packaging and labelling. [75] Sweden has demonstrated that through strong regulation of composition, SLT-product-related harm has been minimised [76]. The Tobacco Products Directive (TPD) in the European Union has issued a total ban on Swedish snus outside of Sweden whilst allowing South-East Asian SLT-products [77], a policy which is contradicted the findings of our review and previous scientific evidence. The findings of this review, together with growing evidence of their role in reducing smoking rates, do not support the continuation of a ban on Swedish snus and other tobacco harm reduction products as a safer alternative to cigarette smoking.

Strengths and limitations
It's a challenge to estimate the risk of disease attributable to such a heterogeneous risk factor such as SLT-products [13]. Any review involving SLT-products will be limited by these issues, unless a single product is studied such as the European snus or the Asian naswar [13]. The output of our systematic review is thus limited due to its reliance on studies which have reported on heterogenous SLTproducts. Furthemore, a meta-analysis of included studies could not be undertaken due to the methodological flaws and vast heterogeneity between studies.
We summarized findings by region and reported on different products as the best 'fit' for categorization of SLT-product use. However, this is not perfect due to the changing landscape and product variation within regions. This issue will only be resolved by future studies carefully documenting and reporting separately for each type of SLT-product.
We sought to identify only those articles where the main research question was on health outcomes from use of SLT-products. The key health outcomes under investigation were mortality, CVD, respiratory and cancer as these make up the major health concerns from SLT-products. We also searched for general health outcomes to identify the breadth of health outcomes being reported.
Finally, the search strategy results were limited to English language reports, and there is a risk that potentially relevant studies reporting health outcomes with ENDS use were subsequently not included.

Conclusion
Our review found studies on SLT-product use focus predominantly on negative health impacts and no studies were found on the health impact from switching from cigarettes to SLT-products. The strength of evidence and quality of the published studies are generally poor, particularly for global studies and those from Asia, Middle East and Africa.
Our review found large differences on the impact on health outcomes between different SLT-products in different regions. Use of SLT-products in Asia, Middle East and Africa region is associated with harmful health outcomes including higher overall and cancer mortality, CVD morbidity, and greatly increased morbidity from most smoking-related cancers, in particular oral cancer. In stark contrast, SLT-products used in Sweden and other parts of Europe such as snus have not been shown on the whole to cause higher mortality or morbidity from CVD or most cancers with evidence for a protective effect against the development of Parkinson's disease. SLT-product use in the US shows more mixed results for mortality, CVD and cancer outcomes with a higher risk than for Europe but substantially lower than those from SE Asia, Middle East and Africa.
Further studies are required to investigate health outcomes from switching from cigarettes to SLT-products and to investigate the full breadth of health outcomes. The wider impacts from SLT-product use on society, such as new uptake in never smokers and nicotine addiction as must also be considered.
Considering the widespread and increasing use of SLT-products in certain parts of the world, there is far less evidence base for their impact on health outcomes compared with cigarette smoking, which is in part due to their predominant use in developing countries. However, the emergence of SLT-products as a driver for reduced smoking rates in Sweden and other parts of Europe warrant further clarification of risk from specific and novel SLT-products.

Appendix 1
There are numerous types of SLT-products available globally which differ markedly in terms of their preparation, method of use and toxicity [6].
Even in western countries, SLT-products are not a homogeneous category [89]. In the US, three traditional types of SLT-products are used: powdered dry snuff, loose leaf chewing tobacco and moist snuff although use of the former two has rapidly declined [90]. In Scandinavia, especially in Sweden, there is a long tradition of moist snuff use, where 'snus' (the generic term for moist snuff in Swedish, pronounced 'snoose') is essentially the only type of SLT-product in use [91].
In the US, dry snuff is made from fermented, fire-cured tobacco that is pulverized into powder. Loose-leaf chewing tobacco consists of air-cured leaf tobacco. Moist snuff usually consists of fire-cured dark tobaccos and is used by a 'pinch' between the thumb and forefinger and placing inside the lip [90].
The fermentation of traditional American products results in higher concentrations of unwanted bacterially mediated by-products, especially TSNAs and nitrite. In Swedish manufacturing of snus air cured tobacco leaves are subjected to pasteurization, yielding virtually sterile products containing very low levels of TSNAs. The manufacturing of Swdish snus does not involve fermentation. Instead, the air cured tobacco leaves are subjected to a heating process (pasteurization), yielding virtually sterile products containing very low levels of TSNAs. Further, in Sweden's manufacturing of snus the tobacco leaves are processed according to Swedish legal regulations for food products and the rigorous, industry standard "Gothi-aTek". Therefore, physical and chemical characteristics of US and Swedish snus products can vary considerably and should not be considered "equivalent" [57].
Other SLT-products also exist, for example, traditional tobacco pouches may contain moist or dry snuff, or small pieces of leaf tobacco and pellets of compressed tobacco. Nicotine pouches contain either tobacco-derived nicotine or synthetic nicotine, but no tobacco leaf, dust, or stem, and are described as either similar to or being a tobacco-free version of snus.
The commonest SLT-products globally are shown in Table 1 (Copied from: IARC Monographs on the Evaluation of Carcinogenic Risks to Humans [6].