Maternal smoking during pregnancy and birth defects in children : a systematic review with meta-analysis

This systematic review aimed to investigate the association between maternal smoking during pregnancy and birth defects in children. We performed an electronic search of observational studies in the databases ovid MEDLINE (1950 to April 2010), LILACS and SciELO. We included 188 studies with a total of 13,564,914 participants (192,655 cases). Significant positive associations were found between maternal smoking and birth defects in the following body systems: cardiovascular (OR: 1.11; 95%CI: 1.031.19), digestive (OR: 1.18; 95%CI: 1.07-1.30), musculoskeletal (OR: 1.27; 95%CI: 1.16-1.39) and face and neck (OR: 1.28; 95%CI: 1.19-1.37). The strength of association between maternal smoking and birth defects measured by the OR (95%CI) is significantly related to the amount of cigarettes smoked daily (χ2 = 12.1; df = 2; p = 0.002). In conclusion, maternal smoking during pregnancy is associated with congenital malformations in children and this association is dose-dependent. Smoking; Pregnancy; Congenital Abnormalities Resumo Esta revisão sistemática teve como objetivo investigar a associação entre fumo materno na gestação e as malformações congênitas em crianças. Uma busca eletrônica dos estudos observacionais foi realizada nas bases de dados ovid MEDLINE (1950 até abril de 2010), SciELO e LILACS. Foram incluídos nesta revisão 188 estudos com um total de 13.564.914 participantes (192.655 casos). Foram encontradas associações positivas significativas entre fumo materno e malformações dos sistemas: cardiovascular (OR: 1,11; IC95%: 1,03-1,19), digestivo (OR: 1,18; IC95%: 1,07-1,30), musculoesquelético (OR: 1,27; IC95%: 1,16-1,39) e face e pescoço (OR: 1,28; IC95%: 1,19-1,37). A força de associação entre fumo materno e malformações medida pelo OR (IC95%) está relacionada significativamente com a quantidade diária de cigarros consumidos (χ2 = 12,1; df = 2; p = 0,002). Concluímos que fumo materno na gestação está associado com maior risco de malformações congênitas em crianças e essa associação é dose-dependente. Hábito de Fumar; Gravidez; Anormalidades Congênitas 1 REVISÃO REVIEW http://dx.doi.org/10.1590/0102-311X00115813 Nicoletti D et al. 2 Cad. Saúde Pública, Rio de Janeiro, 30(12):1-40, dez, 2014 Introdução As malformações congênitas apresentam elevada mortalidade e morbidade nas crianças. Estimase que cerca de 5% dos nascidos vivos apresentem alguma anomalia do desenvolvimento 1. Nas últimas décadas, tem sido observada a crescente contribuição das malformações congênitas na mortalidade infantil 2,3. No Brasil, a proporção de óbitos infantis atribuíveis às malformações congênitas subiu de 9,7% em 1996, para 18,2% em 2008, com um aumento médio anual de 0,71% 3. Esse aumento pode ser por causa do melhor controle das doenças infectocontagiosas e carenciais e, consequentemente, à redução de óbitos infantis por essas doenças 1,3. A maioria das malformações congênitas tem etiologia multifatorial. Além de fator genético, sua ocorrência pode estar relacionada à exposição da criança, ainda antes do nascimento ou mesmo de seus pais, a substâncias tóxicas, entre elas o tabaco 4. Durante a realização do presente trabalho, uma revisão sistemática com 101 estudos observacionais foi publicada, demonstrando associação entre tabagismo materno na gestação e diversas anomalias congênitas em crianças 5. Esta revisão, no entanto, não incluiu um número considerável de estudos relevantes 6,7,8,9,10,11,12,13,14. Além disso, os defeitos da parede abdominal tais como hérnia diafragmática e inguinal, gastrosquise e onfalocele que deveriam ser considerados como anomalias do sistema musculoesquelético de acordo com a 10a revisão da Classificação Internacional de Doenças (CID-10) 15 foram classificados como malformações gastrointestinais. A presente revisão sistemática teve como objetivo investigar a associação entre fumo materno na gestação e as malformações congênitas em crianças, tendo sido estudada também a possível relação dose-resposta nessa associação.


This systematic review aimed to investigate the association between maternal smoking during pregnancy and birth defects in children.
We performed an electronic search of observational studies in the databases ovid MEDLINE  (1950 to April 2010), LILACS and SciELO.We included 188 studies with a total of 13,564,914  participants (192,655 cases).Significant positive associations were found between maternal smoking and birth defects in the following body systems: cardiovascular (OR: 1.11; 95%CI: 1.03-1.19), digestive (OR: 1.18; 95%CI: 1.07-1.30),musculoskeletal (OR: 1.27; 95%CI: 1. 16-1.39)  and face and neck (OR: 1.28; 95%CI: 1. 19-1.37).The strength of association between maternal smoking and birth defects measured by the OR (95%CI) is significantly related to the amount of cigarettes smoked daily (χ 2 = 12.1; df = 2; p = 0.002).In conclusion, maternal smoking during pregnancy is associated with congenital malformations in children and this association is dose-dependent.

Introduction
Birth defects are the cause of high mortality and morbidity in children.It is estimated that about 5% of live births present some abnormality in their development 1 .Over the past decades, birth defects have increasingly contributed to child mortality 2,3 .In Brazil, the rate of child deaths due to birth defects rose from 9.7% in 1996 to 18.2% in 2008, representing an annual average increase of 0.71% 3 .This increase may be due to a better management of infections and contagious, and nutrition-related diseases, which reduced child deaths from these conditions 1,3 .
Most birth defects are of multifactorial etiology.In addition to the genetic component, their occurrence may be related to exposure of the child, even before birth, or the parents to toxic substances, including tobacco 4 .While this investigation was being carried out, a systematic review with 101 observational studies was published, and showed an association between maternal smoking during pregnancy and different birth defects in children 5 .This review, however, did not include a considerable number of relevant studies 6,7,8,9,10,11,12,13,14 .Moreover, defects of the abdominal wall, such as congenital diaphragmatic and inguinal hernia, gastroschisis, and omphalocele, which should be considered musculoskeletal abnormalities, according to the 10 th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) 15 were classified as gastrointestinal defects.
The purpose of this systematic review is to investigate maternal smoking during pregnancy and birth defects in children.The possible doseresponse relation in that association was also studied.

Methods
A systematic review with meta-analysis was conducted.The procedures for the review and reporting of the results were based on the recommendations by MOOSE (Meta-analysis of Observational Studies in Epidemiology) 16 .The protocol for the review was assessed and approved by a panel that included two experts in Pediatrics and one in Epidemiology, in 2010.
Studies that investigated the association between maternal smoking during pregnancy and birth defects in children were considered eligible for the review.Studies that contemplated the association between maternal smoking and chromosomal abnormalities were ruled out.
The electronic search of the studies was made on databases Ovid MEDLINE (1950 until April 2010), SciELO, and LILACS.The strategy to search potentially relevant studies for the review on the databse Ovid MEDLINE is composed of two parts (Figure 1): the first (from line #1 to line #4) is the search strategy to identify studies on maternal smoking, and the secons part (from line #5 to line #20) is the strategy to find birth-defects-related studies.The bibliographic references of articles whose full text was obtained were reviewed, in order to identify additional studies.The Google Translator (https://translate.google.com.br/) was used to translate two articles, one in Lithuanian and other in French.
Study selection was independently made by four investigators (two teams of two).Selection process was made in two stages: in the first, the title and abstract of the articles identified during the electronic search were reviewed to select potential studies for this review.The full text of articles was obtained for which information from the title and the abstract met the inclusion criteria, or in cases where there was not enough information to decide about their inclusion.In the second stage, the articles were read in full for a final selection of the studies, with the inclusion and exclusion criteria being checked.Discrepancies among the investigators were resolved by consensus.Data extraction was performed by four investigators using a standard form.The extracted data were checked by the investigators.
Meta-analysis was performed using the software Stata, version 11.0 (Stata Corp., College Station, United States).A random effects model was applied.The association between maternal smoking during pregnancy and the presence of any kind of birth defects in children was evaluated by means of odds ratios (OR) and 95% confidence intervals (95%CI).When the original studies indicated the presence of more than one defect, the results of each defect were combined to obtain data of any type of defect.Whenever possible, adjusted OR was used.
The pre-defined sub-group analyses were performed to investigate the association between maternal smoking during pregnancy and birth defects in children, according to the organ systems involved.The classification of birth defects was based on the ICD-10.The pre-defined sub-group analyses were also used to assess the potential influence of the following methodological aspects in the results of the meta-analyses: design of the investigation (prospective vs. retrospective); size of the sample (cases) (≤ 200; 200-1,000; 1,000-5,000; > 5,000); adjustment/ matching of confounding factors, including age of the mother (yes vs. no).Two post hoc subgroup analyses were performed to assess the potential impact of exposure definition (maternal smok-Cad.Saúde Pública, Rio de Janeiro, 30 (12):2491-2529, dez, 2014  ing), and the period of exposure during pregnancy in the meta-analysis results.To investigate the dose-response relation between maternal smoking during pregnancy and birth defects in children, the analysis was stratified in three categories according to the number of cigarettes smoked per day (1-9, 10-19 and > 20).
Heterogeneity of results among the studies was assessed through the I 2 statistic.I 2 > 75% indicates significant heterogeneity 17 .Possible causes for heterogeneity were examined through the above mentioned sub-group analyses.The publication bias was investigated with the use of the funnel plot and the Egger test 17 .
In the sub-group analyses, according to the organ systems involved, there were significant positive associations between maternal smoking and defects in the cardiovascular system (OR: 1.11; 95%CI: 1.03-1.19),digestive system (OR: 1.18; 95%CI: 1.07-1.30),musculoskeletal system (OR: 1.27; 95%CI: 1.16-139), and face and neck (OR: 1.28; 95%CI: 1.19-1.37)(Figure 3).Other subgroup analyses showed that retrospective studies and those with smaller sample size (≤ 1.000 cases) has higher combined OR values.Using or not adjustment/matching in the original studies to control confounding factors, particularly the age of the mother did not significantly affect the meta-analysis results (Table 3).Two post hoc sub-group analyses were performed to assess the potential impact of the definition of maternal smoking, and the period of pregnancy the pregnant mother was exposed to smoking in the meta-analysis results.There was no statistically significant difference between studies in which maternal smoking during pregnancy was explicitly defined as daily smoking (n = 91; OR: 1.21; 95%CI: 1.16-1.26)and those studies with no clear definition (n = 97; OR: 1.17; 95%CI: 1.11-1.23)(χ 2 = 1.0; p = 0.32).In addition, there was no statistically significant difference between studies in which exposure to smoking occurred in the first quarter of the pregnancy (n = 80; OR: 1.22; 95%CI: 1.17-1.29)and those studies with no clear definition (n = 108; OR: 1.16; 95%CI: 1.10-1.21)(χ 2 = 2.1; p = 0.15).
(χ² = 12.1; p = 0.002).Post hoc sub-group analyses were performed according to the design of the investigation, control of confounders, and size of the sample (cases).The statistically significant dose-response relation was seen in sub-groups of studies that had controlled confounder factors and in studies where the number of cases ranged between 200 and 5,000 (Table 4).The design of the investigation did not substantially affect the results of the dose-response relation.
The cumulative meta-analysis showed a statistically significant association between maternal smoking during pregnancy and birth defects in children when 40 studies published until 1990, with a total of 26,827 were included in the analysis (OR: 1.09; 95%CI: 1.001-1.19;p = 0.035).The  OR (95%CI) and the p value were respectively 1.16 (1.10-1.23)and < 0.001, when 87 studies published until 2000, with a total of 95,556 cases were included in the meta-analysis.The result of the meta-analysis remained almost unchanged when 101 studies (97,099 cases) published between 2001 and 2010 were included (Figure 5).In the funnel plot (Figure 6), a slight asymmetry in the lower left corner was observed due to lack of studies, which suggested that studies with small samples demonstrating protective effects of maternal smoking against defects in children had not been published.The Egger test also showed evidence of the "small studies" effect, which suggests the presence of publication bias (p < 0.001).Test for differences among sub-groups (χ 2 = 12; df = 2; p = 0.002).95%CI: 95% confidence interval; ES: effect size; OR: odds ratio.

Table 4
Post hoc subgroup analysis amout the dose-response relation between maternal smoking during pregnancy and birth defects in children.* p-value < 0.10 rather than 0.05 was considered statistically significant in the χ 2 test due to the low statistic value, as there are only 3 dose groups.

Discussion
This systematic review with meta-analysis has shown that children of mothers who smoked during pregnancy are at a higher risk of presenting birth defects of any type.Significant associations between maternal smoking during preg-nancy and birth defects of the cardiovascular, digestive, musculoskeletal systems and of the face and neck were evidenced.Positive associations were also observed between maternal smoking and birth defects of the respiratory, nervous, and urogenital systems; however, these associations were not statistically significant.Cumulative meta-analysis about the association between maternal smoking during pregnancy and birth defects in children.
Note: weights are of random effect analysis.
In this systematic review a statistically significant dose-response relation was alsou found between maternal smoking during pregnancy and the risk of birth defects in children; this means, the higher the number of cigarettes a day smoked by the mother, the higher the risk of having babies with some type of birth defects.It was also observed that all the three daily doses of Cad.Saúde Pública, Rio de Janeiro, 30 (12):2491-2529, dez, 2014 cigarette-smoking were significantly associated with higher risk of birth defects compared to non-smoking, suggesting that the regular use of cigarettes by the pregnant woman, even in small amounts, may cause adverse impact in the development of the fetus.
The mechanisms of action of tobacco in the increase of abnormalities in babies are not accurately understood.It is believed that the vasoconstrictor effect of nicotine may reduce the uterine and placental blood flow 197 .Carbon monoxide binds to the hemoglobin in such a way that less oxygen is available for the placenta.In addition, the endothelial injury caused by tobacco increases the rupture of blood vessels from neovascularization of the placenta, leading to a decrease in the blood flow to the fetus, causing hypoxia which will likely result in abnormal fetal morphogenesis 198 .Therefore, exposure to toxins in tandem with hypoxia and cellular ischemia results in abnormal cellular proliferation.
Approximately one third of Brazilian adults were smokers by the end of the 1990s; there was, however, a reduction of about 50% (from 34% to 18.2%) in the prevalence of smokers in this population between 1989 and 2008 199 .A number of factors account for this reduction, including antitobacco policies and availability of smoking-cessation treatments.Smoking during pregnancy is of particular concern, as it is associated with many maternal-fetal outcomes, such as low-weight at birth, premature deliveries, perinatal deaths, and birth defects 200,201 .In countries such as the United States and Canada, where anti-tobacco governmental policies are aggressive, and strong investments are made to control smoking during pregnancy, the prevalence of maternal smoking during pregnancy is currently around 10 to 12% 202,203 .A recent study carried out in nine countries, including Latin America (Argentina, Brazil, Ecuador, Guatemala and Uruguay), Asia (India and Pakistan), and Africa (Democratic Republic of Congo and Zambia) showed higher prevalence of maternal smoking during pregnancy in Uruguay (18.3%), followed by Argentina (10.3%) and Brazil (6.1%) 204 .However, some local studies made in Brazil have shown a prevalence of active smoking of around 20% among pregnant women 201,205 , a proportion much higher than the reported in this international multicentric study.These data point the need for yet stronger actions against tobacco-use during pregnancy in Latin America, including Brazil.
There are a number of resources available to facilitate smoking cessation, such as anti-smoking patches, and anti-anxiety agents like bupropion 197 .These may be used prior to the patient become pregnant.For this reason, we stress the importance of pre-pregnancy counseling.
A systematic review has also shown an association between maternal smoking during pregnancy and birth defects in children 5 .Compared to that review, this one has included 20 additional studies 6,7,8,9,10,11,12,13,14,24,43,79,83,95,99,102,103,122,125,188 that have added about 10,000 cases of defects, and 800,000 of controls.Another difference between the two reviews is that 19 studies about abdominal wall defects were included in the meta-analysis of the gastro-intestinal system in the previous review, whereas these defects were classified as pertaining to the musculoskeletal system in this review.Despite these methodological differences, the results of these two reviews were similar in regards to the association between maternal smoking during pregnancy and defects of the cardiovascular, respiratory, digestive, nervous, urogenital and musculoskeletal systems.The meta-analysis from the previous review included 38 studies in which all defects were combined together did not show significant association between maternal smoking and birth defects (OR: 1.01; 95%CI: 0.96-1.07).The metaanalysis of the current review has included all the 188 studies in which the defects were both combined or of a particular type, and evidenced a statistically significant association between maternal smoking during pregnancy and the risk of any type of birth defect in children (OR: 1.18; 95%CI: 1.14-1.22).
The cumulative meta-analysis of this current review shows that there was already evidence of the association between maternal smoking during pregnancy and birth defects in children by analyzing the results of 40 studies published until 1990 that included a total of 26,827 cases of defects (OR = 1.09; p = 0.035).The evidence of the association became more robust with the results of 87 studies published until 2000, with a total of 95,556 cases (OR = 1.16; p < 0.0001).Between 2000 and 2010, more than 100 studies were carried out with some 100,000 cases of defects; the inclusion of these studies, however, did not change significantly the results of the meta-analysis.These data indicate that findings about the association between maternal smoking during pregnancy and birth defects in children are convincing, and there is no need of further epidemiological studies to investigate this association.
Some methodological studies should be considered in interpreting the results of this systematic review.The heterogeneity of the results of the studies included in this review is to be expected, considering the differences in the research design, type of defect and method used for diagnosis, definition of maternal smoking and control of the effect of confounders.Some of the confounding factors were investigated through subgroup analyses, whose results suggest that the type of defect, the design of the research and the size of the sample are possible causes of heterogeneity.The quality of the studies included was not assessed individually due to limitations of the tools currently available 206 ; however, the potential influences of the methodological aspects of the studies (research design, sample size, control of the effect of confounders, and definition of exposure) in the results of the meta-analysis were investigated through the sub-group analyses.The influence of passive smoking in the association between maternal smoking during pregnancy and birth defects in children was not investigated due to lack of information in most of the original studies.Future studies should address this issue.The funnel plot and the Egger test suggest the presence of publication bias, due to non-publication of small studies that would demonstrate the protective effect of maternal smoking against defects in children.We believe that the number of this type of study is limited, and the lack of data from these studies does not significantly affect the results of the meta-analysis.
We conclude, from this systematic review with meta-analysis, that maternal smoking during pregnancy is associated with a higher risk of birth defects in children, and that this is a dosedependent association.

Figure 1 Search
Figure 1Search strategy of studies in the Ovid MEDLINE database.

Figure 2 Flowchart
Figure 2Flowchart of the selection of studies included in the review.

Figure 3 Maternal
Figure 3Maternal smoking during pregnancy and birth defects in children according to the body systems involved.

Figure 4 Dose
Figure 4Dose-response between maternal smoking and birth defects in children.

Table 1
General characteristics of the studies included.

Table 2
Association between maternal smoking during pregnancy and birth defects in children: results of the 188 studies with birth defect of any type.

Table 3
Analysis of subgroups about the association between maternal smoking during pregnancy and birth defects in children.I 2 measures the heterogeneity of the results among the sutdies (> 75% indicates significant heterogeneity). *