Environmental tobacco smoke exposure and perinatal outcomes: a systematic review and meta-analyses

Authors

  • GISELLE SALMASI,

    1. Department of Health Sciences, McMaster University, Hamilton, Ontario, Canada L8N 3Z5
    Search for more papers by this author
  • ROSHEEN GRADY,

    1. Department of Health Research Methodology, McMaster University, 1200 Main St. West, Hamilton, Ontario, Canada L8N 3Z5
    Search for more papers by this author
  • JENNIFER JONES,

    1. Department of Health Sciences, McMaster University, Hamilton, Ontario, Canada L8N 3Z5
    Search for more papers by this author
  • SARAH D. MCDONALD,

    Corresponding author
    1. Division of Maternal-Fetal Medicine, Departments of Obstetrics & Gynecology, Diagnostic Imaging, and Clinical Epidemiology & Biostatistics, McMaster University, 1200 Main St. West, Hamilton, Ontario, Canada L8N 3Z5
      Sarah D. McDonald, McMaster University, Division of Maternal-Fetal Medicine, Departments of Obstetrics & Gynecology, Diagnostic Imaging, and Clinical Epidemiology & Biostatistics 1200 Main St. West, HSC 3N52B, Hamilton, Ontario, Canada L8N 3Z5. E-mail: mcdonals@mcmaster.ca
    Search for more papers by this author
  • On behalf of the Knowledge Synthesis Group*


Sarah D. McDonald, McMaster University, Division of Maternal-Fetal Medicine, Departments of Obstetrics & Gynecology, Diagnostic Imaging, and Clinical Epidemiology & Biostatistics 1200 Main St. West, HSC 3N52B, Hamilton, Ontario, Canada L8N 3Z5. E-mail: mcdonals@mcmaster.ca

Abstract

Background. While active maternal tobacco smoking has well established adverse perinatal outcomes, the effects of passive maternal smoking, also called environmental tobacco exposure (ETS), are less well studied and less consistent. Objective: To determine to the effect of ETS on perinatal outcomes. Search strategy. Medline, EMBASE and reference lists were searched. Selection criteria. Studies comparing ETS-exposed pregnant women with those unexposed which adequately addressed active maternal smoking. Data collection and analysis. Two reviewers independently assessed titles, abstracts, full studies, extracted data and assessed quality. Dichotomous data were pooled using odds ratios (OR) and continuous data with weighted mean differences (WMD) using a random effects model. Main results. Seventy-six articles were included with a total of 48,439 ETS-exposed women and 90,918 unexposed women. ETS-exposed infants weighed less [WMD –60 g, 95% confidence interval (CI) –80 to –39 g], with a trend towards increased low birthweight (LBW, < 2,500 g; RR 1.16; 95% CI 0.99–1.36), although the duration of gestation and preterm delivery were similar (WMD 0.02 weeks, 95% CI –0.09 to 0.12 weeks and RR 1.07; 95% CI 0.93–1.22). ETS-exposed infants had longer infant lengths (1.75 cm; 95% CI 1.37–2.12 cm), increased risks of congenital anomalies (OR 1.17; 95% CI 1.03–1.34) and a trend towards smaller head circumferences (–0.11 cm; 95% CI –0.22 to 0.01 cm). Conclusions. ETS-exposed women have increased risks of infants with lower birthweight, congenital anomalies, longer lengths, and trends towards smaller head circumferences and LBW.

Introduction

While active maternal tobacco smoking has well established adverse perinatal outcomes, the effects of passive maternal smoking, also known as environmental tobacco exposure (ETS), have been less well studied and are less consistent in the literature. Active maternal smoking increases the risks of perinatal mortality (1), preterm delivery (PTD) (1), miscarriage (2), ectopic pregnancy (3), antepartum hemorrhage (4), and placenta previa (4). Neonates born to active smokers weigh approximately 200 g less at birth, are at an increased risk of being small-for-gestational age (SGA) (5) and having smaller head circumferences (6). Newborns of women who smoke have increased risks of congenital anomalies such as orofacial clefts (7).

There are a number of proposed biologic mechanisms for the adverse reproductive effects of maternal smoking, including fetal hypoxia secondary to increased levels of carboxyhemoglobin, decreased unloading of blood oxygen, nicotine-induced placental vasoconstriction, and placental vascular disease (8). Tobacco smoke contains over 4,000 compounds including a number of reproductive toxins (9, 10). ETS exposure is a combination of exhaled smoke (mainstream) and smoke emitted directly from tobacco products (sidestream), the latter of which contains greater concentrations of many harmful constituents than active smoke (11).

ETS represents a potentially large public health problem given that 22–30% of non-smoking pregnant women are exposed to ETS (12). A variety of methods has been reported in the literature to determine the presence and extent of ETS exposure, including self-report and assays of nicotine or cotinine. Cotinine is a major metabolite of nicotine with a longer half-life and it is believed to be a more accurate measure of total ETS exposure than questionnaire methods (8, 13).

Although the risks of active maternal smoking are well documented, a complete summary of the effect of ETS exposure on perinatal outcomes is lacking. We undertook a systematic review of the literature to ensure a comprehensive and unbiased summary of the available evidence on the effect of ETS exposure alone without active maternal smoking on all relevant perinatal outcomes.

Methods

The Meta-analysis of Observational Studies in Epidemiology consensus statement was followed (14).

Sources

MEDLINE (1966–April 23, 2009) and EMBASE (1980–April 23, 2009) were searched using separate comprehensive search strategies for MEDLINE and EMBASE (Supplementary Material 1, S1). The reference lists of identified articles were searched for additional references.

Eligibility criteria

Case-control and cohort studies were included if they compared mothers who were exposed to ETS with mothers who were unexposed and reported on any of the outcomes of interest below. ETS exposure was defined either as a positive response on a patient questionnaire or biochemical assays within the accepted passive smoking range (serum, hair, or saliva measurements of 2–10 ng/ml of cotinine; none of the included studies assayed nicotine). To control for potential confounding by maternal smoking, studies were included only if they excluded active maternal smoking or controlled for it. Studies published in any language were eligible for inclusion.

Duplicate publications were excluded. Studies were also excluded if mothers actively smoked or if it could not be determined from the article if active maternal smoking was a confounding factor and if no response could be ascertained from the author.

Study selection

Two independent reviewers (two of G.S., R.G., J.J.) screened the titles and abstracts of all citations identified in the search. The full-text article was retrieved if either reviewer considered the citation potentially relevant. Disagreements were resolved by discussion and consensus. An independent adjudicator (S.M.) was consulted for uncertainty or disagreement.

Outcome measures

Our primary outcome was perinatal mortality. The four main secondary outcomes were birthweight, gestational age at delivery, PTD (< 37 weeks gestation), and low birthweight (LBW, < 2,500 g).

Other secondary outcomes included infant length, infant head circumference, SGA (defined in the original studies as infant birthweight below the 10th percentile for gestational age), intrauterine growth restriction (IUGR) (defined in the original studies as infant length at birth below the 10th percentile for gestational age), congenital malformations, spontaneous abortion, caesarean section, Apgar score at 1 minute and at 5 minutes, ectopic pregnancy, antepartum hemorrhage, stillbirth, and placenta previa.

Data abstraction

A data collection form was used to collect raw data. A pilot instrument was tested, and modifications were made. A perinatologist (S.M.) reviewed the data collection instrument to ensure that all relevant perinatal outcomes were included. Two independent reviewers (two of G.S., R.G., J.J.) performed the extraction of data from full articles. Disagreements were resolved by discussion and consensus and by the use of the independent adjudicator (S.M.) when necessary. Authors were contacted for further information where necessary. Study information, which included the country of origin, years of study, exposure by self-report or biochemically, characteristics of cases and control subjects, potential confounders, outcomes and quality assessment, was collected.

Data analysis and synthesis

Statistical analysis was performed with the Review Manager software (Revman 5.0; the Cochrane Collaboration, Oxford, England). When multiple categories of ETS exposure were provided in the original studies, the results of the individual categories were combined in a weighted fashion according to the number of patients in each category. Most of the studies presented unadjusted data and so unadjusted data were pooled for all studies. Dichotomous data were meta-analyzed with odds ratios (OR) using a random effect model (15) since observational studies have more variability than randomized control trials. Continuous data were analyzed with a weighted mean difference (WMD). Where required and when the incidence of the outcome was rare, in order to be able to pool data, adjusted RR were calculated from adjusted OR (16). As is typical in meta-analyses there was no adjustment for multiple analyses. The I2 value was used to examine heterogeneity in our pooled analyses. An I2 value represents the percentage of total variation across studies due to heterogeneity rather than chance and we considered an I2 value 25% as low (17). Two post-hoc sensitivity analyses were performed analyzing the effect of ETS exposure on birthweight according to (i) maternal self-report versus biochemical analysis and (ii) prospective versus retrospective data.

Quality assessment

Two independent reviewers (two of G.S., R.G., J.J.) performed the quality assessment for each study. Disagreements were resolved by the previously described consensus process. An independent adjudicator (S.M.) was available if uncertainty or disagreement existed. Quality assessments of studies were performed with the Cochrane Handbook guidelines for observational studies (18). Selection bias was defined as systematic differences in the methods of ascertainment of the cases or exposed participants and controls or unexposed participants, including factors such as whether consecutive participants were included. Performance bias was defined as systematic differences in the care or observation provided to participants in the comparison groups other than the exposure under investigation. Attrition bias was defined as systematic differences in the loss to follow-up. Detection bias was defined as systematic differences in the ascertainment of our outcomes. Publication bias was assessed using funnel plots.

Results

Description of studies

One thousand thirty-nine non-duplicate titles and abstracts were identified (Figure 1), 241 full articles reviewed and 76 articles (19–94) met inclusion criteria with a total of 48,439 women exposed to ETS and 90,918 unexposed women (Table 1). The studies originated mainly from developed countries. The most common reasons for exclusion were that ETS exposure or perinatal outcomes were not examined and active maternal active smoking was not accounted for.

Figure 1.

Study process.

Table 1. Characteristics of included studies.
StudyCountry (state, province or city where provided)Years of studyStudy designNumber of ETS exposedNumber of ETS unexposed
  1. Note: ETS, environmental tobacco exposure; NR, information not reported by author and was not provided by authors when an attempt was made to contact.

Adamek 2005PolandNRRetrospective cohort356426
Ahlborg 1991Sweden1980–1983Prospective cohort6782,262
Borlee 1978Belgium1972–1974Case control14692
Campbell 1988UKNRRetrospective cohort143247
Carbone 2007Italy1998–2002Case control182202
Carmichael 2008USA1997–2003Case control4433,527
Chatenoud 1998Italy1990–1998Case control1,2711,044
Chen 1989ChinaJune–December 1981Retrospective cohort304446
Chen 1995USAJanuary–September 1991Case control70165
Chevrier 2008France1998–2001Case control167173
Chung 1997NRNRRetrospective cohort1366
Dejin-Karlson 1998SwedenSeptember 1991–September 1992Prospective cohort323240
Dollberg 2000IsraelJanuary 1998–March 1998Prospective cohort5531
Ekwo 1993USA (Iowa)1985–1987Case control184184
Eliopoulos 1996Canada (Toronto)September–December 1992Prospective cohort2135
Ermis 2004TurkeyNRRetrospective cohort1415
Eskenazi 1995USA (California)1964–1967Prospective cohort771,610
Fantuzzi 2007Italy1999–2000Case control277677
Fantuzzi 2008Italy1999–2000Case control355583
Fayol 2005NRNRProspective cohort1417
Fortier 1994CanadaJanuary–October 1989Retrospective cohort3341
George 2006Sweden1996–1998Case control272859
Goel 2004IndiaNRRetrospective cohort141435
Gomolka 2006Poland2002–2005Prospective cohort118113
Hanke 1999Poland1996–1997Retrospective cohort827924
Haug 2000NorwayNovember–December 1992Retrospective cohort4,93411,496
Honein 2007USAOctober 1997–December 2001Case control7782,955
Hrubá 2000Czech Republic1999Retrospective cohort127600
Jaddoe 2008The Netherlands2002–2006Prospective cohort2,1453,681
Jaakola 2001Finland1996–1997Retrospective cohort261258
Janghorbani 1998IranJune–December 1994Prospective cohort278424
Kalinka 2005PolandNRProspective cohort848
Kharrazi 2004USA1992Prospective cohort1,7441,033
Lackmann 2000GermanyNRProspective cohort167187
Lazzaroni 1990Italy1989Prospective cohort220428
Lie 2008Norway1996–2001Case control196658
Little 2004UKSeptember 1997–January 2000Case control178121
L∅drup-Carlsen1997NorwayJanuary 1992–1993Prospective cohort3333
Luciano 1998ItalyNRProspective cohort3950
Macmahon 1966USAMay and June 1963Retrospective cohort3,5402,395
Maconochie 2006UK1995–2001Case control2,0104,560
Mainous 1994USA1988–1994Retrospective cohort332,410
Martin 1986USA1985Retrospective cohort8531,620
Martinez 1994USANRRetrospective cohort191716
Mathai 1990IndiaJanuary–May 1990Prospective cohort54133
Mathai 1992IndiaNRRetrospective cohort520474
Matsubara 2000Japan1989–1991Prospective cohort1,444560
Miller 2009USA1997–2003Case control1,1813,515
Mitchell 2002New ZealandOctober–December 1997Case control6592,995
Nafstad 1998Norway1995Case control5468
Nakamura 2004BrazilNRProspective cohort267329
Ogawa 1991JapanJune–August 1987Retrospective cohort4,0442,454
Peacock 1998UK1982–1984Prospective cohort164169
Pichini 2000Spain1997–1998Retrospective cohort139146
Pierik 2004Netherlands1999–2001Case control4141
Rashid 2003Saudi Arabia2001Retrospective cohort440428
Rauh 2004New York1998–2002Retrospective cohort91135
Roquer 1995Spain1994Retrospective cohort3341
Sadler 1999USA1988–2002Prospective cohort6061,677
Saito 1991JapanNRRetrospective cohort1,4021,311
Savitz 1991USA1959–1966Retrospective cohort3121
Schwartz-Bickenbach 1987GermanyJanuary–December 1985Retrospective cohort2826
Seidman 1990Israel1974–1976Retrospective cohort5,4459,032
Shaw 1996USA1987–1989Case control198579
Steuerer 1999Germany1992–1994Prospective cohort57160
Steyn 2006South Africa1990Prospective cohort739559
Suarez 2008USA1995–2000Case control153196
Tsui 2008Taiwan2003–2004Prospective cohort184175
Venners 2004China1996–1998Prospective cohort310216
Ward 2007UK2000–2002Retrospective cohort2,2598,100
Windham 1999USA (California)1990–1991Prospective cohort7592,887
Windham 2000USA (California)1990–1991Prospective cohort1122,312
Wong-Gibbons 2008USA (Iowa)1997–2003Case control8063,471
Wu 2007, p. 124TaiwanAugust 2003–October 2004Prospective cohort183176
Wu 2007, p. 313China1996–2000Prospective cohort708680
Zhang 1993China1986–1987Retrospective cohort1,033752
Total   48,43990,918

There were no studies investigating our primary outcome, perinatal mortality. A statistically significant association was found between ETS exposure and lower mean birthweight [WMD –60 g; 95% CI –80 to –39 g, 44 studies, Table 2 and Figure 2], although the duration of gestation was similar (WMD 0.02 weeks, 95% CI –0.09 to 0.12 weeks, 17 studies, Figure 3). There was a trend towards increased PTD in the crude data (OR 1.20; 95% CI 0.99–1.46, 18 studies, Figure 4), but not in the adjusted data (RR 1.07; 95% CI 0.93–1.22, 7 studies, Table 2). There was not a significant increase in LBW in the crude data (OR 1.16; 95% CI 0.93–1.45, 19 studies, Figure 5), although there was a trend towards an increase in the adjusted data (RR 1.16; 95% CI 0.99–1.36, 9 studies, Table 2 and Figure 6).

Table 2. Summary table of results of meta-analyses of environmental tobacco smoke exposure and perinatal outcomes.
OutcomeNumber of studiesNumber of ETS exposed with outcome/total number of ETS exposedNumber of ETS unexposed with outcome/total number of ETS unexposedPooled outcome (95% CI)
  1. Statistically significant results are in bold. Note: ETS, environmental tobacco exposure; IUGR, intrauterine growth restriction (infant length < 10th percentile for gestational age); LBW, low birth weight < 2,500 g; OR, odds ratios; PTD, delivery < 37 weeks gestation; SGA, small-for-gestational age (birthweight < 10th percentile for gestational age); WMD, weighted mean difference.

Perinatal mortality0NANANA
Birthweight4424,12547,538WMD –60 g (95% CI –80 to –39)
Birthweight (ETS assessed by self report)3321,82944,590WMD –58 g (95% CI –81 to –36)
Birthweight (ETS assessed biochemically)123,0043,628WMD –62 g (95% CI –101 to –24)
Gestational age at delivery173,3777,076WMD 0.02 weeks (95% CI –0.09 to 0.12)
PTD (crude data)18773/11,2371,432/24,596OR 1.20 (95% CI 0.99–1.46)
PTD (adjusted data)7NANARR 1.07 (95% CI 0.93–1.22)
LBW (crude data)19664/15,6601,033/25,130OR 1.16 (95% CI 0.93–1.45)
LBW (adjusted data)9NANARR 1.16 (95% CI 0.99–1.36)
Infant length91,2771,632WMD 1.75 cm (95% CI 1.37–2.12)
Infant head circumference71,1411,151WMD –0.11 cm (95% CI –0.22 to 0.01)
IUGR191/4,04462/2,454OR 0.89 (95% CI 0.64–1.23)
SGA181,165/7,4062,493/12,891OR 1.06 (95% CI 0.75–1.50)
Congenital malformations121,428/10,4743,258/26,541OR 1.18 (95% CI 1.04–1.34)
Spontaneous abortion91,134/5,9791,834/11,579OR 1.17 (95% CI 0.97–1.41)
Cesarean section4167/869277/1,225OR 1.10 (95% CI 0.88–1.39)
Apgar score at 1 min2581863WMD –0.05 (95% CI –0.35 to 0.25)
Apgar score at 5 min46941,033WMD 0.01 (95% CI –0.10 to 0.13)
Ectopic pregnancy10/131/66Inestimable
Antepartum hemorrhage0NANANA
Stillbirth0NANANA
Placenta praevia0NANANA
Figure 2.

Forest plot of environmental tobacco smoke and birthweight. Environmental tobacco exposure (ETS) was defined as passive maternal exposure to tobacco smoke assessed by self-report questionnaire or biological marker. Sizes of data markers indicate the weights of each study in the analysis. CI, confidence interval; Random, random effects model used for statistical pooling.

Figure 3.

Forest plot of environmental tobacco smoke and gestational age at delivery. Environmental tobacco exposure (ETS) was defined as passive maternal exposure to tobacco smoke assessed by self-report questionnaire or biological marker. Gestational age is given in weeks. Sizes of data markers indicate the weights of each study in the analysis. CI, confidence interval; Random, random effects model used for statistical pooling.

Figure 4.

Forest plot of environmental tobacco smoke and preterm delivery. Environmental tobacco exposure (ETS) was defined as passive maternal exposure to tobacco smoke assessed by self-report questionnaire or biological marker. Preterm delivery is defined as delivery at < 37 weeks gestation. Sizes of data markers indicate the weights of each study in the analysis. CI, confidence interval; Random, random effects model used for statistical pooling.

Figure 5.

Forest plot of environmental tobacco smoke and low birthweight (LBW). Environmental tobacco exposure (ETS) was defined as passive maternal exposure to tobacco smoke assessed by self-report questionnaire or biological marker. LBW is defined as birthweight < 2,500 g at delivery. Sizes of data markers indicate the weights of each study in the analysis. CI, confidence interval; Random, random effects model used for statistical pooling.

Figure 6.

Forest plot of environmental tobacco smoke and low birthweight (LBW, adjusted data). Environmental tobacco exposure (ETS) was defined as passive maternal exposure to tobacco smoke assessed by self-report questionnaire or biological marker. LBW is defined as birthweight < 2,500 g at delivery. Sizes of data markers indicate the weights of each study in the analysis. CI, confidence interval; Random, random effects model used for statistical pooling.

ETS exposed infants were longer at birth by 1.75 cm (95% CI 1.37–2.12 cm, 9 studies, Table 2) with a trend towards a smaller head circumference (–0.11 cm, 95% CI –0.22 to 0.01 cm, 7 studies, Table 2). No significant differences were found between SGA (defined in the original studies as birthweight < 10th percentile for gestational age, OR 1.06 (95% CI 0.75–1.50), 18 studies), spontaneous abortion (OR 1.17; 95% CI 0.97–1.41, 9 studies), cesarean section (OR 1.10; 95% CI 0.88–1.39, 4 studies), or Apgar scores at 1 minute (WMD –0.05; 95% CI –0.35 to 0.25, 2 studies) and 5 minutes (WMD 0.01; 95% CI –0.10 to 0.13, 4 studies). The risk of IUGR (defined in the original studies as birth length < 10th percentile for gestational age, OR 0.89; 95% CI 0.64–1.23) and ectopic pregnancy (0/13 in ETS exposed and 1/66 in unexposed) were each examined in a single study. Women exposed to ETS had an increased risk of having an infant with a congenital anomaly (OR 1.17; 95% CI 1.03–1.34, 12 studies). No studies examined the risk of antepartum hemorrhage, stillbirth, or placenta previa.

Heterogeneity

The heterogeneity in the pooled risk estimates of our outcomes ranged from an I2 test of 0–100%, and generally exceeded 75%, which is considered high. The heterogeneity was likely due to a variety of factors, including varying patient selection and the range of sample sizes. Importantly, both the level of the exposure and its measurement (self report versus biochemical markers such as cotinine samples from serum, hair, or saliva) varied. Hence, a post hoc sensitivity analysis was performed analyzing the effect of ETS exposure on birthweight according to (i) maternal self-report and (ii) biochemical analysis. Infants born to mothers with self-reported ETS exposure had a similar mean birthweight (WMD –58 g; 95% CI –81 to –36 g, I2 = 100%, 33 studies, Table 2) to infants born to mothers whose ETS exposure was assessed biochemically (–62 g; 95% CI –101 to –24 g, I2 = 54%, 9 studies, Table 2). The WMD in birthweight was similar in both the retrospective and prospective studies (–58 g; 95 % CI –85 to –29 g, I2 = 100%, and –61 g; 95% CI –79 to –45, I2 = 100%, respectively, data not shown).

Quality assessment

Quality assessment, performed according to the Cochrane Handbook (18), is summarized in Table 3. Selection bias was not felt to play a major role in any of the studies as often consecutive or ‘all available’ women were included. Most studies did not specifically address performance bias, however, it is unlikely given that most clinicians were not aware of the potential risks of ETS at the time patient care was provided. Attrition bias was often not well documented by authors of cohort studies and it is not applicable in case-control studies. Detection bias was felt to be unlikely given that most of the outcomes in this review are objectively assessed and have standard definitions (for instance, LBW is consistently defined as a birthweight < 2,500 g).

Table 3. Quality assessment of included studies based on Cochrane Handbook Criteria for observational studies.
StudySelection biasPerformance biasAttrition biasDetection bias
  1. Note: Unlikely, bias unlikely; NA, attrition not applicable to case control studies.

Adamek 2005UnlikelyUnlikelyUnlikelyUnlikely
Ahlborg 1991UnlikelyUnlikelyUnlikelyUnlikely
Borlee 1978UnlikelyUnlikelyNARecall bias possible as women asked to recall ETS exposure during pregnancy after delivering an infant with a congenital abnormality
Campbell 1988Possible bias–information on selection limitedUnlikelyUnlikelyUnlikely
Carbone 2007UnlikelyUnlikelyUnlikelyUnlikely
Carmichael 2008UnlikelyUnlikelyUnlikelyUnlikely
Chatenoud 1998UnlikelyUnlikelyNAUnlikely
Chen 1989UnlikelyUnlikelyNAUnlikely
Chen 1995UnlikelyUnlikelyNAUnlikely
Chevrier 2008Eligible cases were not recruited consecutively, only on days recruiter was in the hospital. Excluded non-French speaking mothersUnlikelyNAUnlikely
Chung 1997Exposed and unexposed obtained from the same population of infertile womenUnlikelyUnlikelyUnlikely
Dejin-Karlsson 1998UnlikelyUnlikelyDid not include in analysis those lost to follow-up at deliveryUnlikely
Dollberg 2000Possible bias–exposed and unexposed may not be similar to the general population as study focused on nucleated red blood cell counts in the neonateUnlikelyUnlikelyUnlikely
Ekwo 1993UnlikelyUnlikelyNAUnlikely
Eliopoulos 1996UnlikelyUnlikelyUnlikelyUnlikely
Ermis 2004UnlikelyUnlikelyUnlikelyUnlikely
Eskenazi 1995UnlikelyUnlikelyUnlikelyUnlikely
Fantuzzi 2007Population only included CaucasiansUnlikelyUnlikelyUnlikely
Fantuzzi 2008Population only included CaucasiansUnlikelyUnlikelyUnlikely
Fayol 2005Possible bias–did not control for potential confoundersUnlikelyUnlikelyUnlikely
Fortier 1994UnlikelyUnlikelyUnlikelyUnlikely
George 2006UnlikelyUnlikelyUnlikelyUnlikely
Goel 2004UnlikelyUnlikelyUnlikelyUnlikely
Gomolka 2006UnlikelyUnlikelyUnlikelyUnlikely
Hanke 1999UnlikelyUnlikelyUnlikelyUnlikely
Haug 2000UnlikelyUnlikelyUnlikelyRecall bias a possibility, study concerned SIDS and information was obtained after a long period of time
Honein 2007UnlikelyUnlikelyUnlikelyUnlikely
Hrubá 2000UnlikelyUnlikelyUnlikelyUnlikely
Jaakkola 2001UnlikelyUnlikelyUnlikelyUnlikely
Jaddoe 2008UnlikelyUnlikelyUnlikelyUnlikely
Janghorbani 1998Population restricted to full term infantsUnlikelyUnlikelyUnlikely
Kalinka 2005Limited information– study participants recruited from two different antenatal clinicsUnlikelyUnlikelyUnlikely
Kharrazi 2004UnlikelyUnlikelyUnlikelyUnlikely
Lackmann 2000UnlikelyUnlikelyUnlikelyUnlikely
Lazzaroni 1990UnlikelyUnlikelyUnlikelyUnlikely
Lie 2008UnlikelyUnlikelyUnlikelyUnlikely
Little 2004UnlikelyUnlikelyNARecall bias possible as women asked to recall ETS exposure after delivering an infant with a congenital abnormality
L∅drup-Carlsen 1997UnlikelyUnlikelyUnlikelyUnlikely
Luciano 1998UnlikelyUnlikelyNAUnlikely
Macmahon 1966Population restricted to ‘whites’ onlyUnlikelyUnlikelyUnlikely
Maconochie 2006UnlikelyUnlikelyUnlikelyUnlikely
Mainous 1994Information not providedUnlikelyUnlikelyRecall bias possible as women asked to recall ETS exposure after delivering an infant with a congenital abnormality
Martin 1986UnlikelyUnlikelyUnlikelyUnlikely
Martinez 1994Excluded ‘unhealthy’ infantsUnlikelyUnlikelyUnlikely
Mathai 1990Population restricted to ‘white’ European womenUnlikelyUnlikelyUnlikely
Mathai 1992Information on selection limitedUnlikelyUnlikelyUnlikely
Matsubara 2000UnlikelyUnlikelyUnlikelyUnlikely
Miller 2009UnlikelyUnlikelyUnlikelyUnlikely
Mitchell 2002UnlikelyUnlikelyNAUnlikely
Nafstad 1998Information not providedUnlikelyNAUnlikely
Nakamura 2004UnlikelyUnlikelyUnlikelyUnlikely
Ogawa 1991UnlikelyUnlikelyUnlikelyUnlikely
Peacock 1998Population restricted to ‘white’ womenUnlikelyUnlikelyUnlikely
Pichini 2000UnlikelyUnlikelyUnlikelyUnlikely
Pierik 2004UnlikelyUnlikelyNAUnlikely
Rashid 2003UnlikelyUnlikelyUnlikelyUnlikely
Rauh 2004UnlikelyUnlikelyUnlikelyUnlikely
Roquer 1995Information not providedUnlikelyUnlikelyUnlikely
Sadler 1999UnlikelyUnlikelyUnlikelyUnlikely
Saito 1991UnlikelyUnlikelyUnlikelyUnlikely
Savitz 1991UnlikelyUnlikelyUnlikelyUnlikely
Schwartz-Bickenbach 1987Information not providedUnlikelyUnlikelyUnlikely
Seidman 1990UnlikelyUnlikelyUnlikelyRecall bias possible as parents asked to recall ETS exposure after delivering an infant with a congenital abnormality
Shaw 1996Limited informationUnlikelyNARecall bias possible as parents asked to recall ETS exposure after delivering an infant with a congenital abnormality
Steuerer 1999Information not providedUnlikelyUnlikelyUnlikely
Steyn 2006UnlikelyUnlikelyUnlikelyUnlikely
Suarez 2008UnlikelyUnlikelyUnlikelyUnlikely
Tsui 2008UnlikelyUnlikelyUnlikelyUnlikely
Venners 2004Population restricted to married women onlyUnlikelyUnlikelyUnlikely
Ward 2007UnlikelyUnlikelyUnlikelyUnlikely
Windham 1999UnlikelyUnlikelyUnlikelyUnlikely
Windham 2000UnlikelyUnlikelyUnlikelyUnlikely
Wong-Gibbons 2008Cases included live births, fetal deaths, elective terminations while controls only included live birthsUnlikelyUnlikelyUnlikely
Wu 2007, p. 124UnlikelyUnlikelyUnlikelyUnlikely
Wu 2007, p. 313Population restricted to employed married women who had previously obtained permission to have a childUnlikelyUnlikelyUnlikely
Zhang 1993Population restricted to full term newbornsUnlikelyUnlikelyUnlikely

Assessment of publication bias

Publication bias was explored using funnel plots for our main secondary outcomes. The plot for birthweight (Figure 7) showed a relative dearth of points to the left of the effect measure, suggesting that there may be small unpublished studies with the potential for worse outcomes in ETS exposed infants. The funnel plots for PTD, gestational age at delivery, LBW, SGA, infant length, and head circumference (data not shown) showed relative symmetry suggesting publication bias is unlikely in these outcomes.

Figure 7.

Funnel plot of environmental tobacco smoke and birthweight. Environmental tobacco exposure (ETS) was defined as passive maternal exposure to tobacco smoke assessed by self-report questionnaire or biological marker. WMD, weighted mean difference; SE, standard error.

Discussion

This meta-analysis, the first to evaluate the effect of maternal ETS exposure on a thorough set of perinatal outcomes, found significantly increased risks of having a lighter, longer neonate with an increased risk of a congenital anomaly and with trends toward a smaller head and an increased risk of LBW. Unfortunately, no comment can be made on the association between ETS and our primary outcome, perinatal mortality, as no studies examined this outcome. We hypothesize that if ETS does have an effect on perinatal mortality, it is likely small given that a meta-analysis of active maternal smoking found an OR 1.23 (95% CI 1.12–1.41) (5), although it might still potentially be important given the significance of the outcome and its preventability. Active maternal smoking is associated with shorter infants and hence we suspect that the finding of longer infants with ETS exposure was likely due to chance. However, active maternal smoking is associated with decreased infant head circumference (6), and hence the trend towards a decreased head circumference that we found is likely valid. This finding is concerning given that the redistribution of blood flow to the brain in the setting of in utero growth restriction is typically thought to result in ‘head sparing’ even when the abdomen is small.

Our study has several strengths. Our aim was to review the association of ETS and a wide breadth of plausible perinatal outcomes, guided by those which are affected by active maternal smoking. Our study adds new information by examining important outcomes including spontaneous abortion, ectopic pregnancy, cesarean section, Apgar scores, antepartum hemorrhage, placenta previa, and congenital malformations. Second, the large number of participants in the included studies yielded adequate power for many important outcomes such as birthweight, duration of gestation, and infant length. The risk of PTD, for instance, was not significantly different in more than 45,000 ETS exposed and unexposed women. Another strength of our meta-analysis was that we performed a sensitivity analysis comparing the results of birthweight according to maternal self-report of ETS exposure compared to biochemical analyses because of concerns in the literature that biochemical methods are superior to self-report. We found similar decreases in birthweight with maternal self-report of ETS exposure and biochemical analyses. These findings could potentially save significant amounts of money for research budgets and for the health care system by justifying the omission of formal biochemical verification of ETS exposure.

The results of our meta-analysis mainly corroborate those of the only other two meta-analyses examining ETS exposure during pregnancy and which focused solely on in utero growth [Windham et al. (95) and Leonardi-Bee et al. (96)] and duration of gestation [Leonardi-Bee et al. (96)]. We found a slightly larger decrease in birthweight in ETS exposed infants (–60 g; 95% CI –80 to –39 g, 44 studies) than either Windham et al. (95) (–31 g; 95% CI –42 to –20 g, 22 studies) or Leonardi-Bee et al. (96) (–34 g; 95% CI –51 to –16 g in 17 prospective studies and –40 g, –54 to –26 g in 27 retrospective studies). Neither we, Windham et al. (95), nor Leonardi-Bee et al. (96) found significant increases in SGA (OR 1.06; 95% CI 0.75–1.50, OR 1.07; 95% CI 1.0–1.15, and OR 1.21; 95% CI 1.06–1.37 in prospective studies, respectively). We found a shorter duration of gestation of ETS exposed infants (–0.4 weeks; 95% CI –0.7 to –0.1 weeks, 17 studies) and Leonardi-Bee et al. (96) did not (–0.04 weeks; 95% CI –0.22 to 0.13, 5 prospective studies, –0.03 weeks, 95% CI –0.28 to 0.22 weeks; 6 retrospective studies). Leonardi-Bee et al. (96) found a significant increase in LBW (OR 1.32; 95% CI 1.07–1.63 in 9 prospective studies and OR 1.22; 95% CI 1.08–1.37 in 17 retrospective studies), while we did not (OR 1.16; 95% CI 0.93–1.45, 19 studies). The differences between our findings and those in the other meta-analyses include variation in whether or not women with active smoking were clearly excluded [not in Windham et al. (95)] and variation in the number of studies included per outcome based on differences in inclusion criteria.

The main limitation of any meta-analysis of observational studies including ours is residual confounding. Although most studies attempted to match or control for some important confounding variables such as maternal age, parity, socioeconomic status, alcohol, and drug use, the extent of the adjustment varied between studies. Moreover, pre-eclampsia has not been well-addressed. It is well documented that active maternal smoking decreases the risk of pre-eclampsia by approximately 30%, which could be associated with fewer inductions and hence longer gestations and higher birthweights (97). ETS exposure could conceivably also decrease the risk of pre-eclampsia but only one study in our systematic review of the literature looked at the effect of ETS on pre-eclampsia (98) (The authors report 13 instances of pre-eclampsia in 606 women exposed to ETS compared to 42 episodes of pre-eclampsia out of 1,677 unexposed to ETS, p = 0.06.). Also, several authors were unable to provide requested data and thus we could not include their studies.

Further research into the potentially additive effects of passive and active maternal smoking should be explored. Additionally, it would be important for future studies to account for important confounders such as pre-eclampsia.

In conclusion, in this systematic review and meta-analyses, we found that mothers exposed to ETS have small but significantly increased risks of having lighter babies with an increased risk of congenital anomalies and trends toward smaller heads and LBW. The results of this meta-analysis suggest that clinicians can report to mothers who are exposed to ETS that the effects on their neonates are likely relatively small. However, for women at a lower threshold for poor perinatal outcomes, such as active smokers or those with other health conditions that put them at higher perinatal risk, the additional effect of ETS may be more significant and deserves further study.

Acknowledgements

We gratefully acknowledge additional data and clarification provided by Drs. Mainous, Eskenazi, Lødrup Carlsen, Borlee, Chen, Hu, Krapels, Steegers, Meeker, Jaddoe, Bakker, Maconochie, Carmichael, Perera, Ekwo and Baker. We are indebted to our translators including Marc Colangelo, Ingrid McTiernan, Sophie Kuziora, Michal Bodhanovic, Mingyu Huang, Irena Szymanska, Katherine Morrison, Issei Nakamura, and Penka Stoyanova.

Declaration of interest: None of the authors have any financial or other conflict of interest. This study was supported by funding from the Canadian Institute of Health Research (CIHR) Knowledge Synthesis/Translation grant # KRS 86242. Dr. McDonald is supported by a CIHR New Investigator Award. CIHR did not have any role in the design, data collection, management, analysis, interpretation of the data, or preparation, review, or approval of the manuscript.

*Members of Knowledge Synthesis Group on determinants of low birth weight/preterm birth: Prakesh Shah, Associate Professor, Department of Paediatrics, Mount Sinai Hospital and Department of Health Policy, Management and Evaluation, University of Toronto, Toronto, Canada; Arne Ohlsson, Professor Emeritus, Department of Paediatrics, Mount Sinai Hospital, New York, USA, and Departments of Paediatrics, Obstetrics and Gynaecology, and Health Policy, Management and Evaluation, University of Toronto, Canada; Vibhuti Shah, Associate Professor, Department of Paediatrics, Mount Sinai Hospital, New York, USA, and Department of Health Policy, Management and Evaluation, University of Toronto, Toronto, Canada; Kellie E Murphy, Assistant Professor, Department of Obstetrics and Gynecology, Mount Sinai Hospital, New York, USA, and University of Toronto, Canada; Sarah D McDonald, Assistant Professor, Division of Maternal-Fetal Medicine, Departments of Obstetrics & Gynecology and Diagnostic Imaging, and Clinical Epidemiology & Biostatistics, McMaster University, Hamilton, Canada; Eileen Hutton, Associate Professor, Department of Obstetrics and Gynecology, McMaster University, Hamilton, Canada; Christine Newburn-Cook, Associate Professor & Associate Dean Research, Faculty of Nursing, University of Alberta, Edmonton, Canada; Corine Frick, Director, Alberta Perinatal Health Program and Adjunct Professor, Faculty of Nursing, University of Calgary, Calgary, Canada; Fran Scott, Associate Professor, Dalla Lana School of Public Health, University of Toronto and Toronto Public Health, Toronto, Canada; Victoria Allen, Associate Professor, Department of Obstetrics and Gynaecology, Dalhousie University, Halifax, Canada; Joseph Beyene, Associate Professor and Senior Scientist, Research Institute of The Hospital for Sick Children and Dalla Lana School of Public Health, University of Toronto, Toronto, Canada.

Supplementary Material 1 (S1)

EMBASE Search Strategy

1 secondary smok:.mp. (10)

2 passive smok:.mp. (2839)

3 second hand smok:.mp. (165)

4 environmental tobacco smok:.mp. (1664)

5 involuntary smok:.mp. (30)

6 exp Tobacco Smoke Pollution/ (2513)

7 air pollution, tobacco smoke.mp. (3)

8 environmental pollution, tobacco smok:.mp. (0)

9 smok:, passive.mp. (75)

10 environmental smoke pollution, tobacco.mp. (0)

11 environmental tobacco smoke pollution.mp. (0)

12 paternal smok:.mp. (140)

13 maternal passive smok:.mp. (11)

14 exp SMOKING/ (72511)

15 fathers.mp. or exp FATHERS/ (6259)

16 14 and 15 (361)

17 or/1-13 (4173)

18 16 or 17 (4443)

19 pregnancy/ (62036)

20 pregnancy outcome/ (4297)

21 pregnancy complications/ or fetal death/ or fetal diseases/ or fetal anoxia/ or fetal distress/ or meconium aspiration/ or labor complications/ or abruptio placentae/ or fetal membranes, premature rupture/ or placenta accreta/ or placenta praevia/ or postpartum hemorrhage/ or labor, premature/ or oligohydramnios/ or exp placenta diseases/ or pregnancy complications, cardiovascular/ or pregnancy complications, hematologic/ or pregnancy complications, infectious/ or pregnancy toxemias/ (35005)

22 stillbirth/ or Growth/ or Growth Disorders/ or PRENATAL INFLUENCES/ or Asphyxia/ or Aspiration/ or Meconium/ or aspiration Pneumonia/ or Respiratory Distress Syndrome/ or PLACENTA DISEASES/ or PLACENTA DISORDERS/ or Pregnancy-induced hypertension/ or pre-eclampsia/ or eclampsia/ or PUERPERAL DISORDERs/ or infant, premature/ (57564)

23 fetal growth retardation/ or infant, low birth weight/ or infant, small for gestational age/ or infant, very low birth weight/ or Growth/ or Growth Disorders/ or Birth Weight/ or fetal macrosomia/ (35557)

24 (pregnancy complication: or fet: death or stillbirth: or ASPHYXIA NEONATORUM or fetal anoxia or FET: DISTRESS or fet: disease: or SMALL FOR GESTATIONAL AGE or growth retardation or fetal growth retardation or iugr or intrauterine growth retardation or growth restriction or fetal growth restriction or intrauterine growth restriction or Growth or Growth Disorders or Birth Weight or low birth weight).mp. (525265)

25 (Fetal Diseases or Infant Disease: or Newborn Disease: or stillbirth or FETAL DEATH or PRENATAL INFLUENCES or Asphyxia or Aspiration or Meconium or aspiration Pneumonia or Respiratory Distress Syndrome or labor complication: or abuptio: or Fetal Membranes or prom or placenta: accreta or Placenta Praevia or Placental Insufficiency or PLACENTA: DISEASE: or PLACENTA: DISORDERS or cardiovascular pregnancy complications or hematologic pregnancy complication: or infectious complication: or PREGNANCY TOXEMIA: or Pregnancy-induced hypertens: or pre-eclampsia or eclampsia or PUERPERIAL DISORDER: or infant, premature or premature: or preterm deliver:).mp. (117328)

26 (premature deliver: or preterm labo?r or premature labo?r).mp. (9955)

27 or/21-26 (633108)

28 18 and 27 (513)

29 meta-analysis.mp. (38272)

30 (metaanalysis or systematic review).mp. (32908)

31 29 or 30 (54685)

32 (case-control or cohort or control).mp. (857273)

33 exp Case-Control Studies/ or case-control.mp. (36980)

34 exp COHORT STUDIES/ (51394)

35 exp Infant, Newborn, Diseases/ (257546)

36 or/19-26,35 (863484)

37 18 and 36 (788)

38 study.mp. (3356134)

39 or/32-34,38 (3582322)

40 37 and 39 (533)

Ancillary