Commentary

The review identified babies born with cleft defects, and looked into whether maternal active smoking increased the risks of the babies having cleft defects when compared to non-smoking mothers.

In conducting the meta-analysis, the authors looked at cohort and case control studies. These are observational-type studies, both effective in determining potential causality, and well positioned in the hierarchy of evidence. Although randomised control trials are generally considered to be the highest form of evidence, ethically, an interventional study design would be inappropriate in this scenario.

The authors used Cochrane, Pubmed and Ovid Medline databases for the review, including only English-language studies in the results, which is a possible source of bias (language bias). They did not perform a search of clinical trials. Pubmed and Science Direct searches produced nine other studies1, 2, 3, 4, 5, 6, 7, 8, 9 that may have been relevant to the meta-analysis. It is unclear why these studies were omitted. Two may have been omitted as they were written in Polish8 and Portuguese9. The case-control study by Little7 may have been omitted due to a referencing error, as the authors have included a meta-analysis10 by the same author. A study by Gunnerbeck1 may have been excluded because it also considered snuff use and not just cigarette smoking. Similarly, Van den Eeden's 1990 study6 may have been omitted as it covered multiple birth malformations and not just cleft defects. There may have been overlap between similar studies by the same author,2, 11 resulting in one of these studies being excluded.2 Three studies3, 4, 5 may all have been omitted as they focused more on the smoking/gene interface. However they did include a similar study by Krapels 2008.12

Within the materials and methods, the reader is advised that the meta-analysis would include any studies where mothers actively smoked during peri-conception, first trimester and throughout pregnancy. The 29 studies included in the review were categorised into total clefts, cleft lip+/−palate, cleft palate, isolated cleft and multiple clefts and smoking dose. The CONSORT diagram provides a useful flow of the papers reviewed. The authors state that the papers were divided into isolated and multiple groups, but the reader is not given further detail of these groupings. Some of the basic reasons for exclusion of the papers were provided, including duplicate resources and data, absent required information, insufficient sample size and using family members as controls.

The quality of the included studies is not discussed in the review. The authors state that most of the studies allowed for confounding factors, such as; mothers' race, age, education, previous gravidity, diabetes, infant weight, gender, alcohol consumption, folic acid, medication and vitamin use. However we are given very little indication as to which or how many of the studies this applies to or how these factors were weighted. We are also not told what sample sizes were used in the included studies. In a review such as this a validated tool such as the Newcastle Ottawa Scale13 is useful to illustrate to the reader the quality of included papers.

P values were calculated for the sub-groups, detailing levels of exposure, suggesting a weak but statistically significant correlation between clefts and maternal smoking.

After conducting both Begg's funnel plot and Egger's asymmetry tests, for each of the defect categories, the authors did not find any publication bias. However, they highlighted a possible reporters' bias on the part of the smoking rates of the mothers, speculating that they may not be entirely truthful about their smoking habits due to guilt.

The results of the review were inconsistent, with only a small statistical significance to support their hypothesis, that maternal smoking can increase the risk of oral clefts. Random effect models (to calculate the odds ratio) showed there was a moderate - high degree of heterogeneity, or difference in the data that were pooled together. Reasons put forward by the authors for this heterogeneity included inconsistent selection of controls14, 15 and small sample sizes. Such levels of heterogeneity would compromise the validity of any conclusions drawn from the meta-analysis.

Despite the inconsistent findings of the meta-analysis the study does show a statistically significant association between smoking and the risk of oral clefts. Dental practitioners should consider smoking as another risk for developing oral clefts, and support mothers with smoking cessation.