Tobacco smoking and oral clefts: a meta-analysis.

213Bulletin of the World Health Organization | March 2004, 82 (3) 1 Professor of Epidemiology, Department of Medicine and Therapeutics, University of Aberdeen, Foresterhill, Aberdeen AB25 2ZD, Scotland (email: a.h.cardy@abdn.ac.uk). Correspondence should be sent to this author. 2 Research Fellow, Epidemiology Group, Department of Medicine and Therapeutics, University of Aberdeen, Aberdeen, Soctland. 3 Professor, Department of Nutrition and Food Sciences, Utah State University, Logan, UT, USA. Ref. No. 03-002451 (Submitted: 10 February 03 – Final revised version received: 05 September 03 – Accepted: 09 September 03) Tobacco smoking and oral clefts: a meta-analysis Julian Little,1 Amanda Cardy,2 & Ronald G. Munger3 Objective To examine the association between maternal smoking and non-syndromic orofacial clefts in infants. Methods A meta-analysis of the association between maternal smoking during pregnancy was carried out using data from 24 case–control and cohort studies. Findings Consistent, moderate and statistically significant associations were found between maternal smoking and cleft lip, with or without cleft palate (relative risk 1.34, 95% confidence interval 1.25–1.44) and between maternal smoking and cleft palate (relative risk 1.22, 95% confidence interval 1.10–1.35). There was evidence of a modest dose–response effect for cleft lip with or without cleft palate. Conclusion The evidence of an association between maternal tobacco smoking and orofacial clefts is strong enough to justify its use in anti-smoking campaigns. Keywords Smoking/adverse effects; Cleft lip/ etiology/ epidemiology; Cleft palate/ etiology/epidemiology; Pregnancy; Infant, Newborn; Dose–response relationship, Drug; Meta-analysis; Case–control studies; Cohort studies (source: MeSH, NLM). Mots clés Tabagisme/effets indésirables; Bec de lièvre/ étiologie/ épidémiologie; Fente palatine/ étiologie/ épidémiologie; Grossesse; Nouveau-né; Relation dose-effet médicaments; Méta-analyse; Etude cas-témoins; Etude cohorte (source: MeSH, INSERM). Palabras clave Tabaquismo/efectos adversos; Labio fisurado/ etiología/ epidemiología; Paladar fisurado/ etiología/ epidemiología; Embarazo; Recién nacido; Relación dosis-respuesta a droga; Meta-análisis; Estudios de casos y controles; Estudios de cohortes (fuente: DeCS, BIREME). Bulletin of the World Health Organization 2004;82:213-218. Voir page 217 le résumé en français. En la página 217 figura un resumen en español. Introduction Smoking is practised by about a third of the world’s population aged 15 years or older, including some 12% of women (1). The proportion of women who smoke in developed coun- tries is currently estimated to be 24%, while the proportion in developing countries is about 7% (1). Although tobacco consumption by women may be declining in some countries, there are signs overall that the proportion of women who smoke is increasing, particularly in developing countries (2), where it is estimated that tobacco consumption is rising by about 3.4% per year (3). Globally, each year approximately 12 million women smoke during pregnancy (4). In view of the known adverse effects of smoking on reproductive health, as well as on health more generally, there is a need to consider new methods for tobacco control. In this paper we review the evidence for a relationship between oral clefts and tobacco smoking. We propose that this evidence might be used as a basis for raising awareness of the adverse consequences of tobacco smoking in a subgroup of the population known to have a relatively high motivation to cease smoking (5). This would reinforce a message delivered in a recent report on craniofacial anomalies (6). Orofacial clefts (OFCs) occur at an average frequency of about 1 per 600 live births (7) but there is substantial geographical variation. Some hospital-based surveys in developing countries have revealed markedly higher frequencies (8). Affected infants require multidisciplinary surgical and non-surgical care from birth until adulthood. Thus, in addition to the burden on families, OFCs represent an important cost for health and related services. OFCs are commonly divided into two etiologically distinct groups: cleft lip with or without cleft palate (CL ± P) and cleft palate only (CP). A further division can be made within these groups into isolated clefts (those not associated with other malformations), syndromic clefts (those that are part of a recognized syndrome) and those associated with multiple defects that are not part of a recognized syndrome (henceforth “multiple”). Minor variations occur in the definitions of these groups given by different workers. A previous meta-analysis of the association between ma- ternal cigarette smoking and nonsyndromic oral clefts, based on 11 studies, indicated odds ratios (ORs) for both cleft lip and cleft palate of about 1.3 (9). The present meta-analysis, which also focuses on maternal smoking, is based on 24 studies. .217 214 Bulletin of the World Health Organization | March 2004, 82 (3) Research Tobacco smoking and oral clefts: a meta-analysis Julian Little et al. Methods Publications of potential relevance to our study were identi- fied by using both exploded MeSH headings and text words in a search of MEDLINE (1966−2002) and EMBASE (1980−2002) databases. The main search terms were “cleft lip”, “cleft palate”, “smoking”, “tobacco”, “cigarette”, “mother” and “maternal”. Since OFCs may not be indexed specifically when a study is related to a number of different types of congenital anomaly, we also conducted searches on congenital anomalies and malformations in general. Further articles of potential relevance were identified by manual searches of reference lists in articles identified during the electronic searches and of current journals covering the fields of epidemiology and teratology. A literature search conducted in the preparation of a mono- graph on neural tube defects (NTDs) was also used to identify earlier work (10). The present review is limited to case–control and cohort studies that include data on smoking by women during preg- nancy. It does not cover animal studies and studies of case series. Where studies overlapped, only the largest data set, or that with the most relevant data, was included. The OR estimates of relative risk (RR) and associated 95% confidence intervals (CIs) for use in the meta-analysis were calculated from the data presented in each paper and were therefore unadjusted, with the exception of three studies where only adjusted RRs were available (11−13). A qualitative assessment was made of the effect of potential confounding. We carried out analyses by etiological type (isolated or multiple) wherever possible. In most studies that considered clefts associated with additional defects a distinction was made between multiple and syndromic cases and the latter were excluded from analysis. We also omitted syndromic cases where possible, although we could not always distinguish between multiple and syndromic groups (11, 14, 15). A meta-analysis was not conducted for syndromic cases alone because of the small numbers of studies presenting relevant data and because the number of cases in these studies was small. We carried out meta-analyses for CL ± P, CP and total clefts. However, the last-mentioned were only analysed in respect of studies that did not distinguish between CL ± P and CP. The analyses of CL ± P and CP were repeated following stratification by the presence or absence of malformations in addition to the cleft, after the exclusion of studies that used malformed controls. All analyses were performed by means of Stata Version 7 software (16). The “meta” command was used in order to obtain fixed and random effects models. Heterogene- ity was assessed using the Q test (17) and, in accordance with convention, a random effect model was used if the P-value was less than 0.1, because this is a weak test. We tested for publica- tion bias by means of Begg’s funnel plot and the formal tests proposed by Begg (18) and Egger (19). Tests for the presence of a linear dose–response effect were carried out where suitable data were available. Results Ten cohort and 22 case–control publications were identified as potentially eligible for inclusion, one of which included data from two separate cohort studies (20). One cohort study and seven case–control studies were excluded, leaving 24 studies for the meta-analysis. A list of the excluded papers is available from the authors. The reasons for exclusion included insuffi- cient data (one publication), data very similar to or superseded by those in other publications (six publications) and the use of a control group unlikely to be representative with regard to smoking exposure (one publication). Most studies derived their cases from various registers of malformations. The methods of exposure assessment varied widely, ranging from the use of specifically designed question- naires or interviews to the examination of standard records collected during the medical histories of mothers or children. The time of data collection ranged from the first trimester to more than three years after birth. The amount of exposure detail varied from the simple presence or absence of smoking during pregnancy to the number of cigarettes smoked and exposure by month of pregnancy (Table 1, web version only, available at: http://www.who.int/bulletin). Our meta-analysis of maternal smoking and CL ± P included thirteen case–control studies (11, 14, 15, 21−30) and two cohort studies (20, 31), one of which (20) included data from two separate cohorts. The initial analysis was based on the maximum number of cases per study after the exclusion of syndromic cases. Data were included for six studies on isolated and multiple defects combined, six studies on isolated defects and three studies in which the pres- ence or absence of additional defects was not stated. Each study was represented only once. A formal test for heterogeneity did not give a significant result (Q = 16.6, P = 0.28) and a fixed effects model was therefore used. We found an overall RR of 1.34 (95% CI = 1.25−1.44) with a range of 1.0 (20) to 2.73 (22) (Fig. 1). The RR was 1.2 or above in 13 of the studies, ranging from 1.3 to 2.3 for case–control studies and from 1.0 (20) to 1.17 (31) for cohort studies. The RR was substantially unchanged on limiting the analysis to studies that considered only isolated CL ± P (1.35; 95% CI = 1.25−1.46; twelve studies) or multiple CL ± P (1.38; 95% CI = 1.14−1.66; seven studies), and on excluding studies that used malformed controls (1.34; 95% CI = 1.22−1.46; nine studies). The Begg funnel plot showed no striking evidence of publication bias (Fig. 2). Neither Begg’s adjusted rank correlation test nor Egger’s regression asymmetry test gave a statistically significant result. The evidence for a relationship between maternal smoking and CP appears somewhat similar to that for CL ± P. The studies used were the same as those used for the analysis CL ± P. A formal test for heterogeneity did not give a significant result (Q = 15.5, P = 0.28) and a fixed effects model was therefore used. The combined estimate of RR was 1.22 (95% CI = 1.1−1.35) and the range was from 0.7 (20) to 2.3 (22) (Fig. 3). Ten of the studies had RR values of 1.2 or more. The two cohort studies had RRs of 0.7 (20) and 1.3 (31). The RR was little changed when analysis was limited to studies that did not use malformed controls (1.34; 95% CI = 1.18−1.52; nine studies). Analysis by subgroups of CP indicated an increased RR for isolated CP (1.31; 95% CI = 1.17−1.47; twelve studies) but not for multiple CP (0.97; 95% CI = 0.77−1.21; seven studies). The Begg funnel plot shows no striking evidence of publication bias (Fig. 4). Neither Begg’s adjusted rank correlation test or Egger’s regression asymmetry test gave a statistically significant result. Studies of total clefts indicated an overall RR that was intermediate between those for CL ± P and CP (1.26; 95% CI = 1.10−1.43). Seven cohort (12, 13, 32−36) and two case–con- trol studies (37, 38) were included. One of the case–control studies (38) included data from two separate control groups. A random effects model was used because of the P-value of 0.003 in the Q test for heterogeneity. Nine studies of CL ± P included sufficient information to allow dose–response analysis (Table 2, web version only, available at: http://www.who.int/bulletin). Five suggested a 215Bulletin of the World Health Organization | March 2004, 82 (3) Research Julian Little et al. Tobacco smoking and oral clefts: a meta-analysis weak dose–response relationship, two showed increasing risk with dose but a reduced risk in the highest exposure group, and one found equal (raised) risks in the low-exposure and medium-exposure groups and an increased risk in the high- exposure group. Our dose–response analysis of CP included eight studies (Table 3, web version only, available at: http://www. who.int/bulletin). No clear evidence was observed: four studies suggested a weak positive dose–response relationship and four did not indicate any such relationship. We did not carry out our own dose–response analysis for total clefts as insufficient raw data were available, although a positive association was reported in four of the eight studies of total clefts which considered this matter and were included in the meta-analysis (13, 32, 37, 38). Some consideration was given to confounding in all but six of the studies in Table 1 (see web version). In most of the studies there was adjustment for maternal sociodemographic characteristics, predominantly maternal age and education, but some studies also considered parity, marital status and race/ ethnicity. Alcohol consumption was also considered as a potential confounder in five studies, and one study omitted women who had consumed alcohol during pregnancy. Adjustment for mul- tivitamin use was made in three studies. In general, adjusted relative risks were similar to crude relative risks. Discussion The present study lends support to the hypothesis that maternal smoking increases the risk of OFCs. Although the studies in- cluded in our analysis varied in terms of case definition, control selection and exposure assessment, the relationship, although of only moderate strength, is largely consistent. The effect was observed for both isolated and multiple clefts and was stronger and more consistent for CL ± P than for CP. We did not conduct a separate analysis for syndromic clefts. Evidence of a dose–response effect for CL ± P may be seen as support for a causal relationship, although a reduced risk was seen in some of the high-exposure groups. This reduced risk may, however, be an artefact attributable to the small sample sizes of these groups or may indicate that high levels of exposure are so toxic to the fetus that they result in fetal death (39). There was little evidence of a dose–response effect for CP. 216 Bulletin of the World Health Organization | March 2004, 82 (3) Research Tobacco smoking and oral clefts: a meta-analysis Julian Little et al. Our analysis was limited to published studies. However, we found no strong evidence of publication bias, although the statistical tests for detecting this have limited power (18), especially for relatively small numbers of studies. Since smoking is correlated with alcohol use, diet and other lifestyle factors in many settings, the control of confounding is an important issue. In most of the studies there was adjustment for various sociodemographic factors that may be related to lifestyle. Of the five studies that controlled for maternal alcohol intake, three treated intake as a dichotomous variable, poten- tially making residual confounding an issue. However, studies of the association between alcohol and OFCs are inconsistent and publication bias may be an issue (6). There is some evi- dence of possible associations with binge drinking but none of the studies included in our meta-analysis considered this. Potentially, vitamin supplementation may also confound the association between OFCs and maternal smoking. In the studies that considered this, however, the adjusted and unadjusted RRs were similar. Moreover, as with alcohol consumption, the relationship is unclear (40). In the study that was most com- prehensive in its assessment of confounding, which included a consideration of alcohol consumption, similar adjusted and unadjusted RRs were found (25). Since the mid-1990s a number of epidemiological studies have sought to investigate interactions between various genes (TGFα, TGFβ3, RARA, MSX1, CYP, GST and EPHX1) and smoking by women during pregnancy in relation to OFCs in their offspring (41). The findings have been inconsistent and suggest that any interaction would probably explain only a small proportion of OFCs. However, this is a promising area of research that can be expected to expand. There has been considerable discussion about the potential importance of recall bias in case–control studies but few studies have attempted to demonstrate or quantify it. The available evidence comparing information collected before pregnancy with that collected retrospectively does not demonstrate any severe bias in relation to women who have had adverse preg- nancy outcomes compared with women who have had normal births (10). Furthermore, it has been shown that recall bias can lead to spurious inferences only under extreme conditions (42−44). One study covered by the present review included 217Bulletin of the World Health Organization | March 2004, 82 (3) Research Julian Little et al. Tobacco smoking and oral clefts: a meta-analysis two control groups, one with and one without malformations that were thought to be unrelated to smoking. Essentially the same RRs were obtained for both groups (38). Maternal smoking has been considered in relation to other types of congenital malformation. No consistent associa- tion has been observed with NTDs, cardiovascular anomalies or Down syndrome. Some studies suggest a positive association with limb defects. In four of five studies on gastroschisis a positive association was observed. We are not aware of any evidence for common etiological factors between clefts and gastroschisis, which is also a midline defect (40). It has been suggested that some aspects of the etiology of NTDs and OFCs may be shared, on the basis of involvement of cranial neural crest cells in pathogenesis. However, there are differences in the pattern of geographical variation in preva- lence at birth, in distribution by sex, and in time trends (7). In addition, while folate and folic acid have a clear role in the etiology of NTDs, evidence on the role of folate in the etiology of OFCs is conflicting (6). The contrast in the findings on maternal smoking provides further evidence of differing etiolo- gies for these anomalies. We found consistent, moderate and statistically signifi- cant associations between both CL ± P and CP and maternal smoking. Assuming causality and a 24% prevalence of smoking during pregnancy, the overall RRs from the meta-analyses suggest population excess fractions of 7.5% for CL ± P and 5% for CP. Young women continue to smoke even though the adverse effects of smoking on reproductive health are well known. The demonstration of an association between tobacco smoking during pregnancy and OFCs, and the provision of an easily understandable measure of the association, may create the basis for a new approach to reducing the uptake and continuation of smoking. The way in which the association is presented for the purpose of risk communication is likely to be important. For example, its presentation as a change in absolute risk from 1 per 600 births to 1 per 450 births may have a smaller impact than pointing out that a woman has approximately a 30% increased risk of having a child with CL ± P and a 20% increased risk of having one with CP if she smokes during pregnancy. O Conflicts of interest: none declared. Résumé Méta-analyse du lien entre le tabagisme et les fentes faciales Resumen Tabaquismo y hendiduras bucales: metanálisis Objetivo Estudiar la relación entre el tabaquismo materno y las hendiduras orofaciales en los recién nacidos. Métodos Partiendo de los datos de 24 estudios de casos y testigos y de cohortes, se realizó un metanálisis de la relación entre el hábito de fumar en las mujeres durante el embarazo y la incidencia de hendiduras orofaciales en su descendencia. Resultados Encontramos relaciones coherentes, moderadas y estadísticamente significativas entre el tabaquismo materno y el labio leporino, con o sin hendidura del paladar (riesgo relativo: 1,34, intervalo de confianza del 95%: 1,25-1,44), y entre el tabaquismo materno y la hendidura del paladar (riesgo relativo: 1,22, intervalo de confianza del 95%: 1,10-1,35). La evidencia reunida muestra una relación dosis-respuesta moderada para el labio leporino, con o sin palatosquisis. Conclusión Las pruebas de la existencia de una relación entre el tabaquismo materno y las hendiduras orofaciales son lo bastante sólidas para justificar su uso en las campañas antitabáquicas. Objectif Examen du lien entre le tabagisme pendant la grossesse et les fentes faciales non syndromiques chez le nourrisson. Méthodes Méta-analyse de cette association à partir de 24 études cas-témoins et études de cohortes. Résultats Des liens cohérents, modérés et statistiquement significatifs ont été trouvés entre le tabagisme pendant la grossesse et la fente labiale, avec ou sans fente palatine (risque relatif 1,34, intervalle de confiance à 95 % : 1,25 – 1,44) et entre le tabagisme maternel et la fente palatine (risque relatif 1,22, intervalle de confiance à 95 % : 1,10 – 1,35). Il semble qu’il y ait une relation dose-effet modeste pour la fente labiale avec ou sans fente palatine. Conclusion L’association entre le tabagisme maternel et les fentes faciales est suffisamment bien établie pour pouvoir être utilisée dans les campagnes antitabac. 218 Bulletin of the World Health Organization | March 2004, 82 (3) Research Tobacco smoking and oral clefts: a meta-analysis Julian Little et al. References 1. World Health Organization. Trends in substance use and associated health problems. Geneva: World Health Organization; 1996 (Fact Sheet 127). 2. Ernster V, Kaufman N, Nichter M, Samet J, Yoon SY. Women and tobacco: moving from policy to action. Bulletin of the World Health Organization 2000;78:891-921. 3. World Health Organization. Combating the tobacco epidemic. The world health report 1999 — making a difference. Geneva: World Health Organization; 1999. p. 65-79. 4. Samet JM, Yoon SY. Women and the tobacco epidemic: challenges for the 21st century. Geneva: World Health Organization; 2001. 5. Moore L, Campbell R, Whelan A, Mills N, Lupton P, Misselbrook E, et al. Self-help smoking cessation in pregnancy: cluster randomised controlled trial. BMJ 2002;25:1383-8. 6. World Health Organization. Global strategies to reduce the health-care burden of craniofacial anomalies: Report of WHO meetings on international collaborative research on craniofacial anomalies. Geneva: World Health Organization; 2002 (Human Genetics Programme 2002. Management of noncommunicable diseases). Available from: http://www.who.int/ncd/ hgn/publications.htm 7. Mossey PA, Little J. Epidemiology of oral clefts: an international perspective. In: Wyszynski DF, editor. Cleft lip and palate: from origin to treatment. New York: Oxford University Press; 2002. p. 127-58. 8. Murray JC, Daack-Hirsch S, Buetow KH, Munger R, Espina L, Paglinawan N, et al. Clinical and epidemiological studies of cleft lip and palate in the Philippines. Cleft Palate–Craniofacial Journal 1997;34:7-10. 9. Wyszynski DF, Duffy DL, Beaty TH. Maternal cigarette smoking and oral clefts: a meta-analysis. Cleft Palate–Craniofacial Journal 1997;34:206-10. 10. Elwood JM, Little J, Elwood JH. Epidemiology and control of neural tube defects. New York: Oxford University Press; 1992. 11. Van Den Eeden SK, Karagas MR, Daling JR, Vaughan TL. A case–control study of maternal smoking and congenital malformations. Paediatric and Perinatal Epidemiology 1990;4:147-55. 12. Malloy MH, Kleinman JC, Bakewell JM, Schramm WF, Land GH. Maternal smoking during pregnancy: no association with congenital malformations in Missouri 1980−83. American Journal of Public Health 1989;19:1243-6. 13. Honein MA, Paulozzi LJ, Watkins ML. Maternal smoking and birth defects: validity of birth certificate data for effect estimation. Public Health Reports 2001;116:327-35. 14. Lorente C, Cordier S, Goujard J, Aymé S, Bianchi F, Calzolari E, et al. Tobacco and alcohol use during pregnancy and risk of oral clefts. American Journal of Public Health 2000;90:415-9. 15. Khoury MJ, Weinstein A, Panny S, Holtzman NA, Lindsay PK, Farrel K, et al. Maternal cigarette smoking and oral clefts: a population-based study. American Journal of Public Health 1987;77:623-5. 16. Stata Corporation. Stata Statistical Software: Release 7.0. College Station, TX: Stata Corporation; 2001. 17. DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled clinical trials 1986;7:177-88. 18. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994;50:1088-101. 19. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629-34. 20. Shiono PH, Klebanoff MA, Berendes HW. Congenital malformations and maternal smoking during pregnancy. Teratology 1986;34:65-71. 21. Saxén I. Cleft lip and palate in Finland: paternal histories, course of pregnancy and selected environmental factors. International Journal of Epidemiology 1974;3:263-70. 22. Ericson A, Källén B, Westerholm P. Cigarette smoking as an etiologic factor in cleft lip and palate. American Journal of Obstetrics and Gynecology 1979;135:348-51. 23. Czeizel A, Nagy E. A recent aetiological study on facial clefting in Hungary. Acta Paediatrica Hungarica 1986;27:145-66. 24. Christensen K, Olsen J, Nørgaard-Pedersen B, Basso O, Støvring H, Milhollin-Johnson L, et al. Oral clefts, transforming growth factor alpha gene variants, and maternal smoking: a population-based case–control study in Denmark, 1991-1994. American Journal of Epidemiology 1999;149:248-55. 25. Shaw GM, Wasserman CR, Lammer EJ, O’Malley CD, Murray JC, Basart AM, et al. Orofacial clefts, parental cigarette smoking, and transforming growth factor-alpha gene variants. American Journal of Human Genetics 1996;58:551-61. 26. Hwang SJ, Beaty TH, Panny SR, Street NA, Joseph JM, Gordon S, et al. Association study of transforming growth factor alpha (TGFα) TaqI polymorphism and oral clefts: Indication of gene-environment interaction in a population-based sample of infants with birth defects. American Journal of Epidemiology 1995;141:629-36. 27. Beaty TH, Wang H, Hetmanski JB, Fan YT, Zeiger JS, Liang KY, et al. A case–control study of nonsyndromic oral clefts in Maryland. Annals of Epidemiology 2001;11:434-42. 28. Khoury MJ, Gomez-Farias M, Mulinare J. Does maternal cigarette smoking during pregnancy cause cleft lip and palate in offspring? American Journal of Diseases of Children 1989;143:333-7. 29. Lieff S, Olshan AF, Werler M, Strauss RP, Smith J, Mitchell A. Maternal cigarette smoking during pregnancy and risk of oral clefts in newborns. American Journal of Epidemiology 1999;150:683-94. 30. Romitti PA, Lidral AC, Munger RG, Daack-Hirsch S, Burns TL, Murray JC. Candidate genes for nonsyndromic cleft lip and palate and maternal cigarette smoking and alcohol consumption: evaluation of genotype- environment interactions from a population-based case–control study of orofacial clefts. Teratology 1999;59:39-50. 31. Källén K. Maternal smoking and orofacial clefts. Cleft Palate–Craniofacial Journal 1997;34:11-6. 32. Evans DR, Newcombe RG, Campbell H. Maternal smoking habits and congenital malformations: a population study. BMJ 1979;2:171-3. 33. Christianson RE. The relationship between maternal smoking and the incidence of congenital malformations. American Journal of Epidemiology 1980;112:684-95. 34. Woods SE, Raju U. Maternal smoking and the risk of congenital birth defects: a cohort study. Journal of the American Board of Family Practice 2001;14:330-4. 35. McDonald AD, Armstrong BG, Sloan M. Cigarette, alcohol, and coffee consumption and congenital defects. American Journal of Public Health 1992;82:91-93. 36. Seidman DS, Ever-Hadani P, Gale R. Effect of maternal smoking and age on congenital anomalies. Obstetrics and Gynecology 1990;76:1046-50. 37. Kelsey JL, Dwyer T, Holford TR, Bracken MB. Maternal smoking and congenital malformations: an epidemiological study. Journal of Epidemiology and Community Health 1978;32:102-7. 38. Chung KC, Kowalski CP, Kim HM, Buchman SR . Maternal cigarette smoking during pregnancy and the risk of having a child with cleft lip/palate. Plastic and Reconstructive Surgery 2000;105:485-91. 39. Selevan SG, Lemasters GK. The dose–response fallacy in human reproductive studies of toxic exposures. Journal of Occupational Medicine 1987;29:451-4. 40. Little J. Congenital anomalies and smoking. United Kingdom Department of Health; 2003. Available from: http://www.eurocat.ulster.ac.uk/ 41. Murray JC. Gene/environment causes of cleft lip and/or palate. Clinical Genetics 2002;61:248-56. 42. Drews CD, Greenland S. The impact of differential recall on the results of case–control studies. International Journal of Epidemiology 1990;19:1107-12. 43. Swan SH, Shaw GM, Schulman J. Reporting and selection bias in case–control studies of congenital malformations. Epidemiology 1992;3:356-63. 44. Khoury MJ, James LM, Erickson JD. On the use of affected controls to address recall bias in case–control studies of birth defects. Teratology 1994;49:273-81. 219Bulletin of the World Health Organization | March 2004, 82 (3) Research Julian Little et al. Tobacco smoking and oral clefts: a meta-analysis Table 1. Studies of orofacial clefts and maternal smoking included in the meta-analysis Author Study area Period Sample CLP No. of Case Control No. of Control Expo- Tim- Infor- Con- Sub- details/ type a,b cases partici- type controls partici- sure ing c mation founder groups source pation pation assess- re- assess- ment corded ment Europe Evans et al., UK, Cardiff 1965–76 Cardiff TC|NS 111 98% d Cohort >65 000 98% Interview in P Dose No No 1979 (1) Birth puerperium Survey Saxén Finland 1967–71 Finnish CL ± P|I 599 100% Matched 590 98% Interview P 5+/day No TC by 1974 (2) Register of CP|I on resi- after versus pres- Congenital TC|M+S dence delivery none ence of Malforma- area and other tions delivery malfor- time mations Ericson et al., Sweden 1975 Swedish CL ± P|NS 66 88% No malfor- 130 87% Health EP Dose Yes No 1979 (3) delivery CP|NS mations, clinic records matched on records delivery of first unit and visit time, during maternal pregnancy age and parity Kallén Sweden 1983–92 Swedish CL+P|I,M 1834 NS Cohort 1 002 742 99% Interview EP Dose Yes Each 1997 (4) health CL|I,M at first cleft registries: CP|I,M antenatal type by cases PRS clinic visit pres- without ence of chromo- other somal malfor- anomalies mations Czeizel & Hungary 1970–76 Hungarian CL ± P|I 630 58% Matched 824 42% Question- PreC Dose No No Nagy, 1986 (5) Congenital CP|I 179 49% on birth naire T1 Malforma- TC|S 392 78% place and P tion week, sex, Registry birth outcome Christensen Denmark 1991–94 Hospital CL±P|I 302 96% Live births, 567 94% Interview, T1 Dose Yes Each et al., 1999 (6) live births CP|I no malfor- usually cleft TC|I,M mations, within two type by matched weeks of TGF on time birth geno- and place type of birth Lorente et al., France, 1989–92 Live CL ± 161 63% e Live births, 1134 90% Interview T1 Dose Yes Each 2000 (7) Italy, births or P|I,M+S no malfor- France, within one cleft Netherlands stillbirths CP|I,M+S mations, Italy; month of type by and United from matched on 60% birth pres- Kingdom congenital residence Nether- ence of malforma- area and lands; other tion delivery 86% malfor- registries time United mations Kingdom USA Kelsey et al., Connecticut 1974–76 Hospital TC|NS 40 71% f Random 2968 90% Interview Month 3 Dose Yes No 1978 (8) and clinic sample of within births hospital- one year born of birth normal infants g Christianson San Francisco 1959–66 Members TC|NS 24 NS Cohort 14 735 NS Interview EP Yes/No No No 1980 (9) of Kaiser in early Health pregnancy, Plan usually first trimester Shiono et al., Multicentre 1959–66 Collabo- CL ± P|NS NS NS Cohort 53 572 98% h Interview EP Yes/No i No No 1986 (10) a rative at regis- Perinatal tration for Project: prenatal liveborn care 20+ weeks gestation A 220 Bulletin of the World Health Organization | March 2004, 82 (3) Research Tobacco smoking and oral clefts: a meta-analysis Julian Little et al. Author Study area Period Sample CLP No. of Case Control No. of Control Expo- Tim- Infor- Con- Sub- details/ type a,b cases partici- type controls partici- sure ingc mation founder groups source pation pation assess- re- assess- ment corded ment Shiono et al., California 1974–77 Kaiser CL ± P|NS 56 NS Cohort 34 434 94% Self- T1 Yes/No (i) Yes No 1986 (10) b Clinics: CP|NS adminis- live births tered of 20+ question- weeks naire as gestation part of routine prenatal care Shaw et al., California 1987–89 California CL ± P|I,M,S 731 85% Live 734 78% Interview, PeriC j Dose Yes Each 1996 (11) Birth CP|I,M,S births, no mean of cleft Defects malfor- 3.5 years type by Monitoring mations, after pres- Program matched delivery ence of (includes on time malfor- fetuses) period mations; and area TGF of birth geno- type Khoury et al., Maryland 1984 Maryland CL ± 53 98% Births 198 95% Interview P Dose Yes Each 1987 (12). Birth P|I,M+S with after birth cleft Overlap with Defects CP|I,M+S other type by Hwang 1995. Reporting defects, pres- and Infor- excluding ence of mation Down other System: syndrome malfor- births of mations 20+ weeks gestation or >500g birthweight Hwang et al., Maryland 1984–92 Maryland CL ± P|I 183 NS Births 284 NS Interview P Yes/No Yes Each 1995 (13) Birth CP|I with Dose for cleft Defects other CP type by Reporting isolated TGF and Infor- defects, geno- mation matched type System: on year births of and 20+ weeks county gestation of birth, or >500g maternal birthweight age Beaty et al., Maryland 1992–98 Cases of CL ± P|I 171 NS Healthy 182 NS Brief T1 Yes/No Yes Also 2001 (14) various CP|I infants interview passive ethnicities from at first smok- from various newborn visit, then ing sources: nursery or detailed figures for attending telephone Whites well baby interview presented clinic Khoury et al., Metropolitan 1968–80 Metropolitan CL ± P|I,M 345 70% k Live 2809 70% l Interview PeriC m Dose Yes Each 1989 (15) Atlanta Atlanta CP|I,M births, cleft Congenital matched type by Defects on birth pres- Program period, ence of hospital other and malfor- ethnicity mations Malloy et al., Missouri 1980–83 Single TC|NS 451 NS Cohort 288 067 95% Birth P Yes/No Yes No 1989 (16) births certificate from Missouri Birth Defects Registry Lieff et al., USA and 1976–92 Slone CL|I,M 1479 84% n Live 2295 77% (n) Interview P Dose Yes Each 1999 (17) Canada, Epidemio- CL + P|I,M births or within six 1976–92; cleft multicentre logy Unit CP|I,M stillbirths months of T1 type by Birth CL±P|I,M with other delivery 1988–92 pres- Defects malfor- ence of Study: mations other Whites, thought malfor- live births unrelated mations or stillbirths to mater- nal smok- ing p (Table 1, cont.) B 221Bulletin of the World Health Organization | March 2004, 82 (3) Research Julian Little et al. Tobacco smoking and oral clefts: a meta-analysis Author Study area Period Sample CLP No. of Case Control No. of Control Expo- Tim- Infor- Con- Sub- details/ type a,b cases partici- type controls partici- sure ingc mation founder groups source pation pation assess- re- assess- ment corded ment Van den Washington 1984–86 Birth CL ± 173 97% p Singleton 4500 96% Birth P Yes/No Yes Each Eeden et al., State records: P|I,M+S liveborn, certificate cleft 1990 (18) single CP|I,M+S no malfor- type by births from mations pres- Washington ence of State other malfor- mations Romitti et al., Iowa 1987–94 Iowa Birth CL ± P|I 225 71% q Live 393 58% Interviews/ Two Dose Yes Each 1999 (19) Defects births, no question- months cleft Registry: malfor- naires, after type by Whites, mations mean 29 con- TGF live births, months ception and stillbirths, after birth MSX1 aborted (cases) or geno- fetuses 35 months type (controls) Chung et al., 46 states 1996 National TC|I 2207 ~99% Live 4414 NS Recorded P Dose Yes No 2000 (20) Centre for births, no by physician/ Health malfor- nurse at Statistics mations birth 1996 Natality Live 9183 NS Recorded P Dose Yes No Database: births with by physician/ live births genitouri- nurse at nary, gastro- birth intestinal or chromosomal anomalies Honein et al., 45 states 1997–8 National TC 5238 NS Cohort 6 161 506 NS Birth P Dose Yes No 2001 (21) plus Centre certificate New York for Health City and Statistics District of 1997-98 Columbia Natality Database: live births Woods and Cincinnati 1998–99 Liveborn TC|NS 7 NS Cohort 18 009 NS Recorded P Yes/No Yes No Raju 2001 hospital at admis- (22) births sion Canada McDonald Montreal 1982–84 Pregnancies TC|NS 96 NS Cohort 89 221 NS NS T1 Dose Yes No et al., 1992 >20 week (23) gestation Eastern Mediterranean Seidman Israel, 1974–76 Live births TC|NS 17 or 18 r NS Cohort 17 152 98% Interview P Dose No No et al., 1990 West or stillbirths in (24) Jerusalem from three puerperium obstetric units a CL ± P = cleft lip with/without cleft palate; CP = cleft palate only; CL = cleft lip only; CL + P = cleft lip and cleft palate; TC = total clefts; PRS = Pierre Robin Syndrome. b I = isolated clefts; M = clefts associated with multiple defects (unknown etiology); S = syndromic clefts; M+S = syndromic and multiple cases probably included; NS = not stated. c PreC = preconceptional; PeriC = periconceptional; P = pregnancy; EP = early pregnancy; T1 = first trimester. d Information on smoking habits of mother obtained for 98% of singleton deliveries recorded in Cardiff Birth Survey. e All types of congenital anomalies combined. f Participation rate for malformations of all types. g Same five hospitals in which 90% of cases of malformations were born; excludes births in private clinics, which accounted for 10% of cases of malformation. h This is overestimated, as an unspecified number of women refused follow-up. It is stated that in 4689 of 59 391 pregnancies the mothers refused follow-up or the pregnancies ended in miscarriage or stillbirth. i Dose-response relationship was analysed but data not presented for oral clefts. j One month before conception to three months after pregnancy began. k Includes subjects with syndromic, multiple and isolated clefts. l Participation rate for mothers of infants with all types of congenital anomalies and mothers of controls combined. m Three months before conception to three months after pregnancy began. n In 1983. p Excluded infants with CNS, CVS and musculoskeletal defects, inguinal hernia, pyloric stenosis and syndromes associated with oral clefts (despite absence of cleft). q Calculated with inclusion of cases having associated malformations; these cases were excluded from analysis. r Case number calculated from other figures presented in paper. (Table 1, cont.) C 222 Bulletin of the World Health Organization | March 2004, 82 (3) Research Tobacco smoking and oral clefts: a meta-analysis Julian Little et al. Table 3. Relative risk of CP by level of maternal cigarette smoking pregnancy (data from first trimester used when available) Relative risk Author Cleft typea Low Medium High Trend (χ2 test) P-value Ericson 1979 (3) NS 2.5 2.1 - 0.68 0.41 Czeizel 1986 (5) I 0.8 0.9 0.3 3.26 0.071 Hwang 1995 (13) I 1.4 1.5 - 1.69 0.194 Shaw 1996 (11) I 1.4 2.2 - 6.84 0.009 Christensen 1999 (6) I 0.9 1.0 0.6 0.24 0.621 Lieff 1999 (17) I 1.3 1.1 1.0 0.43 0.511 Romitti 1999 (19) I 1.5 2.3 - 5.23 0.022 Lorente 2000 (7) I 1.3 1.4 - 0.64 0.423 a I = isolated, NS = not stated. Categories as Table 2, plus Hwang: Low 1–10; Medium 11. Table 2. Relative risk of CL ± P by level of maternal smoking during pregnancy (data from first trimester used when available) Relative risk Author Cleft typea Low Medium High Trend (χ2 test) P-value Ericson 1979 (3) NS 3.4 2.3 - 4.10 0.043 Czeizel 1986 (5) I 1.5 0.9 1.1 1.10 0.295 Khoury 1987b (12) I 1.2 4.2 4.8 8.58c 0.003c Khoury 1989c (15) I 1.6 1.6 1.9 8.03 0.005 Shaw 1996 (11) I 1.6 2.1 - 15.47 0.000 Christensen 1999 (6) I 1.1 1.6 1.5 5.01 0.025 Lieff 1999 (17) I 1.5 1.4 1.1 5.76 0.016 Romitti 1999 (19) I 1.2 1.3 - 0.97 0.324 Lorente 2000 (7) I 1.4 2.0 - 6.36 0.012 a I = isolated, NS = not stated. b Subset of Hwang et al. (13). c From Wyszynski et al. (25). Categories Ericson: Low 1−9: Medium 10+. Czeizel: Low 1−10; Medium 11−20; High 21+. Khoury 1987: Low 1−10; Medium 11−20; High 21+. Khoury 1989, Leiff: Low 1−14; Medium 15−24; High 25+. Shaw: Low 1−19; Medium 20+. Christensen: Low 1−9; Medium 10−19; High 20+. Romitti, Lorente: Low 1−9; Medium 10+. References in Tables 1 and 2 1. Evans DR, Newcombe RG, Campbell H. Maternal smoking habits and congenital malformations: a population study. BMJ 1979;2:171-3. 2. Saxén I. Cleft lip and palate in Finland: paternal histories, course of pregnancy and selected environmental factors. International Journal of Epidemiology 1974;3:263-70. 3. Ericson A, Källén B, Westerholm P. Cigarette smoking as an etiologic factor in cleft lip and palate. American Journal of Obstetrics and Gynecology 1979;135:348-51. 4. Källén K. Maternal smoking and orofacial clefts. Cleft Palate–Craniofacial Journal 1997;34:11-6. 5. Czeizel A, Nagy E. A recent aetiological study on facial clefting in Hungary. Acta Paediatrica Hungarica 1986;27:145-66. 6. Christensen K, Olsen J, Nørgaard-Pedersen B, Basso O, Støvring H, Milhollin-Johnson L, et al. Oral clefts, transforming growth factor alpha gene variants, and maternal smoking: a population-based case- control study in Denmark, 1991-1994. American Journal of Epidemiology 1999;149:248-55. 7. Lorente C, Cordier S, Goujard J, Aymé S, Bianchi F, Calzolari E, et al. Tobacco and alcohol use during pregnancy and risk of oral clefts. American Journal of Public Health 2000;90:415-9. 8. Kelsey JL, Dwyer T, Holford TR, Bracken MB. Maternal smoking and congenital malformations: an epidemiological study. Journal of Epidemiology and Community Health 1978;32:102-7. 9. Christianson RE. The relationship between maternal smoking and the incidence of congenital malformations. American Journal of Epidemiology 1980;112:684-95. 10. Shiono PH, Klebanoff MA, Berendes HW. Congenital malformations and maternal smoking during pregnancy. Teratology 1986;34:65-71. 11. Shaw GM, Wasserman CR, Lammer EJ, O’Malley CD, Murray JC, Basart AM, et al. Orofacial clefts, parental cigarette smoking, and transforming growth factor-alpha gene variants. American Journal of Human Genetics 1996;58:551-61. D 223Bulletin of the World Health Organization | March 2004, 82 (3) Research Julian Little et al. Tobacco smoking and oral clefts: a meta-analysis 12. Khoury MJ, Weinstein A, Panny S, Holtzman NA, Lindsay PK, Farrel K, et al. Maternal cigarette smoking and oral clefts: a population-based study. American Journal of Public Health 1987;77:623-5. 13. Hwang SJ, Beaty TH, Panny SR, Street NA, Joseph JM, Gordon S, et al. Association study of transforming growth factor alpha (TGFα) TaqI polymorphism and oral clefts: Indication of gene-environment interaction in a population-based sample of infants with birth defects. American Journal of Epidemiology 1995;141:629-36. 14. Beaty TH, Wang H, Hetmanski JB, Fan YT, Zeiger JS, Liang KY, et al. A case- control study of nonsyndromic oral clefts in Maryland. Annals of Epidemiology 2001;11:434-42. 15. Khoury MJ, Gomez-Farias M, Mulinare J. Does maternal cigarette smoking during pregnancy cause cleft lip and palate in offspring? American Journal of Diseases of Children 1989;143:333-7. 16. Malloy MH, Kleinman JC, Bakewell JM, Schramm WF, Land GH. Maternal smoking during pregnancy: no association with congenital malformations in Missouri 1980−83. American Journal of Public Health 1989;19:1243-6. 17. Lieff S, Olshan AF, Werler M, Strauss RP, Smith J, Mitchell A. Maternal cigarette smoking during pregnancy and risk of oral clefts in newborns. American Journal of Epidemiology 1999;150:683-94. 18. Van Den Eeden SK, Karagas MR, Daling JR, Vaughan TL. A case-control study of maternal smoking and congenital malformations. Paediatric and Perinatal Epidemiology 1990;4:147-55. 19. Romitti PA, Lidral AC, Munger RG, Daack-Hirsch S, Burns TL, Murray JC. Candidate genes for nonsyndromic cleft lip and palate and maternal cigarette smoking and alcohol consumption: evaluation of genotype- environment interactions from a population-based case-control study of orofacial clefts. Teratology 1999;59:39-50. 20. Chung KC, Kowalski CP, Kim HM, Buchman SR . Maternal cigarette smoking during pregnancy and the risk of having a child with cleft lip/palate. Plastic and Reconstructive Surgery 2000;105:485-91. 21. Honein MA, Paulozzi LJ, Watkins ML. Maternal smoking and birth defects: validity of birth certificate data for effect estimation. Public Health Reports 2001;116:327-35. 22. Woods SE, Raju U. Maternal smoking and the risk of congenital birth defects: a cohort study. Journal of the American Board of Family Practice 2001;14:330-4. 23. McDonald AD, Armstrong BG, Sloan M. Cigarette, alcohol, and coffee consumption and congenital defects. American Journal of Public Health 1992;82:91-3. 24. Seidman DS, Ever-Hadani P, Gale R. Effect of maternal smoking and age on congenital anomalies. Obstetrics and Gynecology 1990;76:1046-50. 25. Wyszynski DF, Duffy DL, Beaty TH. Maternal cigarette smoking and oral clefts: a meta-analysis. Cleft Palate Craniofacial Journal 1997;34:206-10. E