The search yielded a total of 292 studies (176 from EMBASE, 62 from PUBMED, 51 from Scopus and 3 from CINAHL). After duplicates were removed and the title and abstract of studies were reviewed; 42 studies were considered potentially eligible for this review and assessed as full-texts based on predetermined inclusion and exclusion criteria. There were 11 articles excluded from this review, and 31 articles included, further details of which are shown in Fig. 1.
Characteristics of included studies
The characteristics of the 31 studies included in this review such as location, sample size, definitions of outcome, type of smokeless tobacco, and quantitative measures of outcome are shown in Table 1. There were eighteen studies from Asia (ten from India, three from Yemen, two from Taiwan, one each from Pakistan, Thailand and Indonesia), six studies from Europe (five from Sweden and one from Sweden and Norway), four studies from Oceania (two from Papua New Guinea, one from Palau and one from Australia), two studies from Africa (one from Ethiopia and one from South Africa), and one study from the United States of America. This review included a total sample size of 3,338,826 pregnant women. The largest study was from Sweden with a sample size of 1,070,013 women (17), while the smallest study was from Australia with a sample size of 50 pregnant women. (18) The predominant form of smokeless tobacco used in Ethiopian and Yemen studies was khat; whereas in other countries from Asia and Oceania it was betel/areca nut; and in India it was mishri. Snuff was used by both the European and South African subjects, while iqmik and pituri was used by subjects in the study from USA and Australia respectively. None of the studies reported a quantified measure of exposure, and most did not quantify frequency of smokeless tobacco use.
Table 1
Quality assessment of included studies as per the Newcastle-Ottawa Scale (referred to in line 252)
| Overall (maximum 9) | Selection | Comparability | Outcome |
| | Representativeness | Selection of non-exposed cohort | Ascertainment of exposure | Demonstration that outcome of interest was not present at start of study | | Assessment | Duration of follow up (adequate follow up minimum 9 months for outcome to occur) | Adequacy of follow up (> 80%) |
(Ghani et al, 1987) | 7 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 |
(Ali et al, 2021) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Baba et al, 2012) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Berger et al, 2016) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Chue et al, 2012) | 6 | 0 | 1 | 1 | 1 | 0 | 1 | 1 | 1 |
(Charlette et al, 2022) | 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(de Costa and Griew, 1982) | 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(Demelash et al, 2015) | 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(England et al, 2012) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(England et al, 2003) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Eriksson et al, 1991) | 6 | 1 | 1 | 0 | 1 | 0 | 1 | 1 | 1 |
(Ganganahalli et al, 2017) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Gupta and Subramoney, 2004) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Idris et al, 2020) | 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(Juarez and Merlo, 2013) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Kreyburg et al, 2019) | 8 | 1 | 1 | 0 | 1 | 2 | 1 | 1 | 1 |
(Krishna, 1978) | 7 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 |
(Krishnamurthy and Joshi, 1993) | 6 | 1 | 1 | 1 | 1 | 0 | 0 | 1 | 1 |
(Madley-Dowd et al, 2021) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Mallick, 2021) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Ome-Kaius et al, 2015) | 6 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 0 |
(Pratinidhi et al, 2010) | 7 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 |
(Pratinidhi et al, 2014) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Ratsch et al, 2021) | 7 | 0 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(Rygh et al, 2019) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Senn et al, 2009) | 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(Steyn et al, 2006) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Sulistiyani et al, 2019) | 6 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 0 |
(Verma et al, 1983) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Wang and Lee, 2012) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Yang et al, 2008) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
Table 1
Quality assessment of included studies as per the Newcastle-Ottawa Scale (referred to in line 252 of manuscript)
| Overall (maximum 9) | Selection | Comparability | Outcome |
| | Representativeness | Selection of non-exposed cohort | Ascertainment of exposure | Demonstration that outcome of interest was not present at start of study | | Assessment | Duration of follow up (adequate follow up minimum 9 months for outcome to occur) | Adequacy of follow up (> 80%) |
(Ghani et al, 1987) | 7 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 |
(Ali et al, 2021) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Baba et al, 2012) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Berger et al, 2016) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Chue et al, 2012) | 6 | 0 | 1 | 1 | 1 | 0 | 1 | 1 | 1 |
(Charlette et al, 2022) | 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(de Costa and Griew, 1982) | 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(Demelash et al, 2015) | 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(England et al, 2012) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(England et al, 2003) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Eriksson et al, 1991) | 6 | 1 | 1 | 0 | 1 | 0 | 1 | 1 | 1 |
(Ganganahalli et al, 2017) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Gupta and Subramoney, 2004) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Idris et al, 2020) | 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(Juarez and Merlo, 2013) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Kreyburg et al, 2019) | 8 | 1 | 1 | 0 | 1 | 2 | 1 | 1 | 1 |
(Krishna, 1978) | 7 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 |
(Krishnamurthy and Joshi, 1993) | 6 | 1 | 1 | 1 | 1 | 0 | 0 | 1 | 1 |
(Madley-Dowd et al, 2021) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Mallick, 2021) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Ome-Kaius et al, 2015) | 6 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 0 |
(Pratinidhi et al, 2010) | 7 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 |
(Pratinidhi et al, 2014) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Ratsch et al, 2021) | 7 | 0 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(Rygh et al, 2019) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Senn et al, 2009) | 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
(Steyn et al, 2006) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Sulistiyani et al, 2019) | 6 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 0 |
(Verma et al, 1983) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Wang and Lee, 2012) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
(Yang et al, 2008) | 9 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 |
Quality assessment
There was high variation in the methodological quality of included studies as summarised in Table 2. Self-reported data regarding the use of smokeless tobacco during pregnancy was collected via interviews and hospital records in all the included studies. Consequently, there a high risk of recall bias and performance bias in these studies since none of the studies reported any biochemical/objective measurements that were performed to ascertain tobacco use. Moreover, statistical adjustments for confounding factors were not made in fifteen of the included studies. (17, 19–34)
Table 2
Characteristics of studies which explored the association between maternal smokeless tobacco use during pregnancy and birth weight (referred to in line 270)
Reference | Region | Time frame | Study design | Definition of birth weight outcome | Smokeless tobacco type | Definition of exposure | Sample population | Number of exposed | Number of non-exposed | Odds ratio and 95% CI for low birth weight |
(Abdul Ghani et al, 1987) | Yemen Arab Republic | 1984 | cohort | Mean birth weight (g) | khat | self-reported khat use | 1181 | 414 | 295 | - |
(Ali et al, 2021) | Pakistan | 2013–2019 | cohort | Low birth weight (< 2500g) | gutkha | gutkha | 3554 | 505 | 197 | - |
(Baba et al, 2012) | Sweden | 1999–2010 | cohort | small for gestational age (birthweight > 2 standard deviations below the mean weight for gestational age) | snuff | used snuff 3 months before pregnancy | 945371 | 23,514 | 663649 | 1.26 (1.09–1.46) |
(Berger et al, 2016) | Palau | 2007–2013 | cohort | Low birth weight (< 2500g) | betel nut | self reported at least 1 use of chewing betel nut with tobacco during pregnancy | 1629 | 893 | 278 | 2.4 (1.0–6.0) |
(Charlette et al, 2022) | India | 2021 | cohort | Low birth weight (< 2500g) | unspecified | smokeless tobacco use during pregnancy | 463 | 23 | 440 | 2.58 (1.1–6.06) |
(Chue et al, 2012) | Thailand | 1997–2006 | cohort | Low birth weight (< 2500g) | betel nut | self reported areca nut use during pregnancy | 9264 | 4963 | 2722 | 0.9091 (0.7694–1.0742) |
(de Costa and Griew, 1982) | Papua New Guinea | 1981 | cohort | Low birth weight (< 2500g) | betel nut | self reported history of betel consumption throughout pregnancy | 800 | 400 | 400 | 1.4227 (0.418–2.4042) |
(Demelash et al, 2015) | Ethiopia | 2013 | case control | Low birth weight (< 2500g) | khat | self reported khat chewing during pregnancy | 387 | 129 cases | 258 controls | 6.4 (2.42–17.10) |
(England et al, 2012) | USA | 1997–2005 | cohort | Low birth weight (< 2500g) | iqmik | iqmik or commercial chew consumption, based on data extracted from clinic records | 4725 | 247 | 121 | - |
(England et al, 2003) | Sweden | 1999–2000 | cohort | small for gestational age (birthweight > 2 standard deviations below the mean weight for gestational age) | snuff | snuff use during pregnancy | 27762 | 789 | 11495 | 1.25 (0.72–2.17) |
(Eriksson et al, 1991) | Yemen Arab Republic | 1991 | cohort | Low birth weight (< 2500g) | khat | khat chewing during pregnancy | 1141 | 142 | 382 | 1.3302 (0.8495 to 2.0831) |
(Ganganahalli et al, 2017) | India | 2011 | cohort | Low birth weight (< 2500g) | mishri | used mishri during pregnancy- self reported and case records | 210 | 105 | 105 | - |
(Gupta and Subramoney, 2004) | India | 2002 | cohort | Low birth weight (< 2500g) | unspecified | used a smokeless tobacco product at least once a day for the past six months | 1217 | 206 | 768 | - |
(Idris et al, 2020) | Yemen | 2016 | case control | Low birth weight (< 2500g) | khat | khat chewing during pregnancy | 252 | 83 | 43 | 1.76 (1.06–2.92) |
(Juarez and Merlo, 2013) | Sweden | 2002–2010 | cohort | Mean birthweight (g) | snuff | Swedish snus use during pregnancy | 938932 | 8339 | 591690 | - |
(Kreyberg et al, 2019) | Norway and Sweden | 2014–2016 | cohort | Mean birthweight (g) | snuff | self reported snus use | 2697 | 150 | 2046 | - |
(Krishna, 1978) | India | 1971–1972 | cohort | Mean birthweight (g) | unspecified | keeping a wad of locally grown and cured tobacco in the mouth for 8 to 10 hours per day during pregnancy | 1148 | 209 | 1168 | - |
(Krishnamurthy and Joshi, 1993) | India | 1993 | cohort | Low birth weight (< 2500g) | unspecified | self reported maternal use of tobacco, mostly mishri, during pregnancy | 178 | 42 | 41 | 3.2(1.5–6.9) |
(Madley-Dowd et al, 2021) | Sweden | 1999–2010 | cohort | small for gestational age (birthweight > two standard deviations below the mean weight for gestational age according to the gender-specific Swedish fetal growth curves) | snuff | self reported snus use at any time during pregnancy | 1181264 | 14665 | 1E + 06 | 1.05(0.94–1.17) |
(Mallick, 2021) | India | 2015–2016 | cohort | Low birth weight (< 2500g) | unspecified | maternal tobacco chewing | 81869 | 3629 | 53953 | 1.163 (1.058–1.227] |
(Ome-Kaius et al, 2015) | Papua New Guinea | 2009–2013 | cohort | Low birth weight (< 2500g) | betel nut | self reported areca nut chewing during pregnancy | 2700 | 1459 | 310 | 0.94 (0.65–1.38) |
(Pratinidhi et al, 2010) | India | 2003–2005 | cohort | Low birth weight (< 2500g) | mishri | self reported mishri use during pregnancy | 705 | 218 | 487 | 2.4026 (1.5206–3.7963) |
(Pratinidhi et al, 2014) | India | 2011 | cohort | Low birth weight (< 2500g) | mishri | self reported mishri use during pregnancy | 2150 | 258 | 258 | 67.9016 (37.3926-123.3033) |
(Ratsch et al, 2021) | Australia | 2021 | cohort | Mean birthweight (g) | pituri | self reported pituri use during pregnancy | 73 | 19 | 31 | 1.75 (0.3152–9.7161) |
(Rygh et al, 2019) | Norway | 2012–2017 | cohort | Mean birthweight (g) | snuff | self reported snus use during the third trimester of pregnancy | 102011 | 201 | 9213 | - |
(Senn et al, 2009) | India | 2007–2008 | cross sectional | Low birth weight (< 2500g) | betel nut | betel nut chewing during pregnancy | 310 | 292 | 18 | 1.9 (0.4–17) |
(Steyn et al, 2006) | South Africa | 1990 | cohort | Mean birthweight (g) | snuff | self reported snuff use during pregnancy | 1593 | 120 | 1376 | - |
(Sulistiyani et al, 2019) | Indonesia | 2018 | cross sectional | Low birth weight (< 2500g) | unspecified | smokeless tobacco use during pregnancy | 99 | 18 | 81 | - |
(Verma et al, 1983) | India | 1978–1979 | cohort | Mean birthweight (g) | unspecified | tobacco chewing during pregnancy, minimum 150 mg/day | 140 | 70 | 70 | - |
(Wang and Lee, 2012) | Taiwan | 2005 | cohort | Low birth weight (< 2500g) | betel nut | betel quid chewing during pregnancy | 24200 | 19 | 8413 | 0.368 (0.048–2.841) |
(Yang et al, 2008) | Taiwan | 2003–2004 | cohort | Low birth weight (< 2500g) | betel nut | self reported betel quid chewing during pregnancy | 1264 | 464 | 800 | 2.40 (1.21–4.80) |
Effects of exposure
Thirty-one studies included in this review were clinically and methodologically diverse. There were twenty-seven cohort studies, two case-control and two cross-sectional studies
included in this systematic review. Crude odds ratios were presented in only eleven (17, 23, 25–27, 30, 31, 33–37) of the included studies, and could be calculated by the authors using the data published for a further six studies. (18, 22, 38–41) Relative risk was presented in two of the included studies. (19, 26) A statistically significant association between use of smokeless tobacco and low birth weight was reported in twelve of the studies. (20, 21, 23, 26, 27, 30, 33–35, 40–42) All of these studies defined low birth weight as < 2500 grams. The highest odds ratio [67.9016 (37.3926-123.3033)] was reported in the study conducted in 2014 by Pratinidhi et al. (41) Meanwhile, the lowest odds ratio [0.368 (0.048–2.841)] was reported by Wang and Lee, although their results were not statistically significant. (37)
A statistically significant reduction in mean birth weight was reported in maternal smokeless tobacco users in a further eleven studies. (20, 22, 26, 33, 36, 38, 40, 41, 43–45) On the contrary, four studies reported that there was no statistically significant difference in birth weight in infants born to mothers who used smokeless tobacco during pregnancy. (18, 24, 29, 32)
Out of all the studies included in this review, seven studies reported on snuff users (17, 20, 25, 28, 29, 32, 44), seven reported on betel nut users (21, 22, 31, 33, 36–38), four reported on khat users (23, 27, 39, 43), three on mishri (40–42), one reported on gutka (19), iqmik (24) and pituri (18) users respectively, and seven did not specify the type of smokeless tobacco used by participants (26, 30, 34, 35, 45–47).
Two studies on khat (23, 27), two on mishri (40, 41), one on snuff (20), one on betel quid (33) and none on areca nut, betel nut, pituri, iqmik or gutka reported a statistically significant association between maternal smokeless tobacco use and low birth weight. One study on khat (43), three on snuff (25, 28, 44), one on mishri (42), two on betel nut (33, 36) and none on iqmik or pituri reported a significant reduction in mean birthweight in infants born to mothers who used smokeless tobacco during pregnancy.
Meta analysis
Seventeen studies involving 732689 participants were included in the meta-analysis, which is shown in Fig. 2. (20–23, 27, 30, 31, 33–42) Other studies from this review were not included due to the provision of insufficient data for meta-analysis to be performed. The probability of low birth weight infants was greater in those women who consumed smokeless tobacco during pregnancy as compared to those who did not consume any tobacco products during pregnancy. The overall pooled estimate under the random effects model showed that there was a statistically significant association (OR = 1.91 [1.38, 2.65], P < 0.0001);) between maternal smokeless tobacco use during pregnancy and low birth weight. Smokeless tobacco users were almost twice as likely to have low birth weight infants as compared to non tobacco users. The test for heterogeneity produced Tau square of 0.37, I2 = 95%, test for overall effect z = 3.91, (P < 0.0001)). This demonstrates a high level of heterogeneity between studies. The highest risk estimates observed were (OR = 67.90 [37.39, 123.30]), in a study conducted in India, which evaluated the effects of maternal mishri consumption during pregnancy on birth weight. (41) However, the wide CIs suggest that this effect estimate may be due to factors such as the small sample size of the study which decrease its reliability.
Subgroup analysis was performed for five studies for betel nut (21, 22, 31, 36, 38), two studies for betel quid (33, 37), three studies for khat (23, 27, 39), and four studies for mishri users, as shown in Fig. 3. (35, 40–42) Subgroup analysis could not be performed for snuff users. (17, 20, 25, 28, 29, 32, 44) This is because only one study provided sufficient data on snuff use and low birth weight to be included in the meta analysis. (20) The results of the subgroup analysis varied based on type of smokeless tobacco. Our analysis found nearly eleven times greater probability of low birth weight in infants of mothers who consumed mishri during pregnancy (OR = 10.98 [2.03, 59.34], P < 0.05). However, no statistically significant association with low birth weight was found for betel nut (OR = 1.02 [0.84, 1.25], P > 0.05), betel quid (OR = 1.51 [0.47, 4.89], P > 0.05) or khat (OR = 1.04 [0.49, 2.21], P < 0.001) users. Although subgroup analysis for khat users showed that P < 0.001, the results are not statistically significant. This is likely due to inadequate pooled sample size of the included studies and the influence of confounding factors.
Most of the studies are aggregated towards the centre of the funnel plot (see Fig. 4 and Fig. 5), which is symmetrical on visual inspection, indicating that there was no publication bias in this meta analysis.