Effects of environmental tobacco smoke on perinatal outcomes: a retrospective cohort study
Dr J Crane, Department of Obstetrics and Gynaecology, Eastern Health, 300 Prince Phillip Drive, St John’s, NL, Canada A1B 3V6. Email email@example.com
Please cite this paper as: Crane J, Keough M, Murphy P, Burrage L, Hutchens D. Effects of environmental tobacco smoke on perinatal outcomes: a retrospective cohort study. BJOG 2011;118:865–871.
Objective To evaluate the effects of environmental tobacco smoke (ETS) on perinatal outcomes.
Design Retrospective cohort study.
Setting Newfoundland and Labrador, Canada.
Population Nonsmoking women with singleton gestations who delivered 1 April 2001–31 March 2009, identified through the Newfoundland and Labrador Provincial Perinatal Database.
Methods Women who self-reported exposure to ETS were compared with those who reported no exposure. Univariate analyses and multivariate linear and logistic regression analyses (adjusting for maternal age, parity, partnered status, work status, level of education, body mass index, alcohol use, illicit drug use and gestational age) were performed and odds ratios(OR; or adjusted differences) with 95% confidence intervals were calculated.
Main outcome measures Birthweight, birth length, head circumference and stillbirth. Secondary outcomes included gestational age at delivery, preterm birth <37 and <34 weeks of gestation, prelabour rupture of membranes, Apgar score, endotracheal intubation for resuscitation, neonatal intensive care unit admission, congenital anomalies, respiratory distress syndrome, intraventricular haemorrhage, neonatal bacterial sepsis, jaundice and neonatal metabolic abnormalities.
Results A total of 11 852 women were included: 1202(11.1%) exposed to ETS and 10 650 (89.9%) not exposed. Exposure to ETS was an independent risk factor for lower mean birthweight (−53.7 g, 95% CI −98.4 to −8.9 g), smaller head circumference (−0.24 cm, 95% CI −0.39 to −0.08 cm), shorter birth length (−0.29 cm, 95% CI −0.51 to −0.07 cm), stillbirth (OR 3.35, 95% CI 1.16–9.72, P = 0.026), and trends towards preterm birth <34 weeks (OR 1.87, 95% CI 1.00–3.53, P = 0.05) and neonatal sepsis (OR 2.96, 95% CI 0.99–8.86).
Conclusions Exposure of nonsmoking pregnant women to ETS is associated with a number of adverse perinatal outcomes including lower birthweight, smaller head circumference and stillbirth, as well as shorter birth length. This information is important for women, their families and healthcare providers, and reinforces the continued need for increased public policy and education on prevention of exposure to ETS.
Maternal cigarette smoking during pregnancy is associated with a variety of obstetric complications including spontaneous abortion, placental abruption, growth restriction, preterm prelabour rupture of membranes, preterm birth, miscarriage and stillbirth, resulting in increased perinatal morbidity and mortality.1–4 Longer-term adverse effects include higher rates of attention deficit hyperactivity disorder,3,5 asthma,4,6 adverse effects on the immune system4,6 and possibly childhood cancers.3 There appears to be a dose–response relationship with increased risks associated with increased number of cigarettes and duration of smoking through pregnancy.1,2
There is less information about exposure to environmental tobacco smoke (ETS; second-hand smoke) and adverse pregnancy outcomes.7,8 Undiluted side-stream smoke contains many harmful chemicals and in greater concentration than cigarette smoke inhaled through a filter.9–11 Previous meta-analyses have found that environmental tobacco smoke reduces mean birthweight by 28–60 g,12–14 may increase the risk of birthweight <2500 g by up to 22%,12,13 but does not seem to affect gestational age or the rate of preterm birth.12–14 One meta-analysis suggested an increase in birth length, but a trend towards smaller head circumference with ETS exposure.12 There is limited information on other adverse perinatal outcomes with ETS exposure, with a previous meta-analysis not finding any studies evaluating stillbirth.12 The objective of this study was to evaluate the effects of ETS, or second-hand smoke exposure, on perinatal outcomes.
We performed a retrospective cohort study of nonsmoking women with singleton gestations who delivered between 1 April 2001 and 31 May 2009, identifying women using the Newfoundland and Labrador Provincial Perinatal Program (NLPPP) Database. This computerised database collects information on pregnancy outcomes for several regions of Newfoundland and Labrador including Eastern region, Western region and Labrador. Data collected include demographic information, antenatal, intrapartum and postpartum events and perinatal outcomes for deliveries of at least 20 weeks of gestation. Quality assurance and data quality are ensured through the NLPPP Database’s routine edit checking process on all extracted data. Exposure to environmental tobacco smoke was recorded on the prenatal record during the first prenatal visit. Women who self-reported exposure to ETS were compared with those who reported no exposure to ETS. Women were excluded from the study if they did not self-report exposure or no exposure to ETS. Women were also excluded if they reported smoking during pregnancy or did not report this information.
Maternal characteristics including maternal age, parity, partnered status, work status, level of education, body mass index (BMI), weight gain during pregnancy (less than recommended, as recommended, or more than recommended),15 alcohol or illicit drug use in pregnancy, hypertension (pre-existing or gestational) and diabetes (pre-existing or gestational) were described and compared between women who self-reported ETS exposure and those who reported no exposure. Primary outcomes of this study were birthweight, head circumference, birth length and stillbirth. Secondary outcomes included birthweight <2500 g, prelabour rupture of membranes, gestational age at delivery (including preterm birth <37 weeks and <34 weeks of gestation), Apgar score, endotracheal intubation for resuscitation, neonatal intensive care unit (NICU) admission, neonatal length of hospital stay, congenital anomalies, respiratory distress syndrome, intraventricular haemorrhage, necrotising enterocolitis, neonatal bacterial sepsis, jaundice and neonatal metabolic abnormalities (including hypoglycaemia, hypomagnesaemia and hypocalcaemia). Outcomes were based on the International Classification of Diseases version 10 coding, with data entered by trained hospital coders.
Statistical analyses were performed using spss for Windows version 18.0 (SPSS Inc., Chicago, IL, USA) and pepi 3.01, 2000 (Computer Programs for Epidemiologists, Stone Mountain, GA, USA). Descriptive statistics were used for nonsmoking women exposed and not exposed to ETS. Univariate analyses including chi-square and Fisher exact tests were used to compare dichotomous outcomes between the two groups and Student’s t test was used to compare continuous outcomes. Data that were ordinal were compared with Mann–Whitney U test. Multiple logistic and linear regression models were used to evaluate outcomes, adjusting for maternal age, parity, partnered status, work status, level of education, BMI, alcohol use, illicit drug use, and gestational age (for outcomes other than gestational age at delivery). Adjusted odds ratios (OR) (or adjusted differences) and 95% confidence intervals were calculated. For the primary outcomes, a value of P < 0.05 was considered significant. A trend was noted if the P value was between 0.05 and 0.075. For secondary outcomes, a P value of significance was based on Hochberg variation, with P < 0.017,16 with a trend noted if P value was between 0.017 and 0.05. The number of nonsmoking women exposed to ETS associated with one extra or one less case of the outcome of interest was calculated.
Ethics approval for the study was obtained from the Human Investigation Committee of Memorial University and the Research Proposal Approval Committee of Eastern Health.
Of 17 723 women with singleton gestations during the study period, 14 650 (82.7%) did not smoke, 2862 (16.1%) smoked during pregnancy, and smoking status was not known for 211 (1.2%). Of the 14 650 women who met the inclusion criteria (nonsmoking women with singleton gestations), 11 852 (80.9%) reported either exposure or no exposure to ETS. Of these women, 1202 (11.1%) were exposed to ETS and 10 650 (89.9%) self-reported no exposure. Demographic characteristics of exposed and not exposed women are summarised in Table 1. Nonsmoking women exposed to ETS were younger, had a higher BMI, were more likely to be nulliparous, less likely to have a partner, less likely to work, more likely to use alcohol or illicit drugs during pregnancy, and had a lower level of education than women not exposed to ETS.
Table 1. Demographic data
|Maternal age (years)||27.1 ± 5.9||30.2 ± 4.8||<0.0001a|
|Nulliparous||813/1202 (67.6%)||5262/10 650 (49.4%)||<0.0001b|
|Parity||0 [0, 1]||1 [1, 2]||<0.0001c|
|Gravidity||1 [1, 2]||2 [1, 3]||<0.0001c|
|BMI (kg/m2)||27.5 ± 7.2||26.8 ± 6.3||0.019a|
|Less than recommended||89/671 (13.3%)||701/5935 (11.8%)||0.37b|
|As recommended||158/671 (23.5%)||1524/5935 (25.7%)|| |
|More than recommended||424/671 (63.2%)||3710/5935 (62.5%)|| |
|Hypertension||154/1202 (12.8%)||1217/10 650 (11.4%)||0.16b|
|Diabetes||44/1202 (3.7%)||374/10 650 (3.5%)||0.79b|
|Partnered||599/1098 (54.6%)||7947/10 002 (79.5%)||<0.0001b|
|Not working||338/957 (35.3%)||1591/7668 (20.7%)||<0.0001b|
|Part time||104/957 (10.9%)||638/7668 (8.3%)|| |
|Full time||515/957 (53.8%)||5439/7668 (70.9%)|| |
|Level of education|
|Less than high school||123/993 (12.4%)||233/9074 (2.6%)||<0.0001b|
|High school||230/993 (23.2%)||846/9074 (9.3%)|| |
|Beyond high school||640/993 (64.5%)||7995/9074 (88.1%)|| |
|Alcohol use in pregnancy||33/1199 (2.8%)||143/10 594 (1.3%)||<0.0001b|
|Elicit drug use in pregnancy||6/1199 (0.5%)||8/10 634 (0.08%)||<0.0001d|
The results of univariate analyses for the outcomes of interest are summarised in Table 2. A number of adverse outcomes were seen more frequently in women exposed to ETS including smaller head circumference, lower birthweight, higher incidence of birthweight <2500 g, increased rates of stillbirth, preterm birth <37 and <34 weeks of gestation, Apgar score <7 at 1 minute, need for endotracheal intubation during resuscitation, NICU admission, bacterial sepsis, jaundice and longer neonatal length of stay, as well as shorter birth length.
Table 2. Perinatal outcomes in nonsmoking women by exposure to ETS by univariate analyses
|Head circumference (cm)||34.43 ± 1.94||34.70 ± 1.86||<0.0001a|
|Birth length (cm)||49.77 ± 2.93||50.02 ± 2.81||0.003a|
|Birthweight (g)||3428 ± 642||3508 ± 603||<0.0001a|
|Birthweight <2500 g||89/1202 (7.40%)||493/10 650 (4.63%)||<0.0001b|
|Stillbirth||10/1202 (0.83%)||39/10 650 (0.37%)||0.028c|
|Prelabour rupture of membranes||313/1201 (26.1%)||2702/10 647 (25.4%)||0.63b|
|Preterm birth <37 weeks||122/1201 (10.16%)||871/10 645 (8.18%)||0.019b|
|Preterm birth <34 weeks||39/1201 (3.25%)||233/10 645 (2.19%)||0.020b|
|Apgar <7 at 1 minute||101/1195 (8.45%)||717/10 611 (6.76%)||0.029b|
|Apgar <7 at 5 minutes||22/1196 (1.84%)||149/10 610 (1.40%)||0.23b|
|Endotracheal tube/use||26/1202 (2.16%)||151/10 650 (1.41%)||0.044b|
|NICU admission||158/1202 (13.14%)||1070/10 650 (10.05%)||0.001b|
|Length of stay (days)||4.55 ± 9.66||3.91 ± 9.00||0.028a|
|Congenital anomalies||157/1202 (13.06%)||1224/10 650 (11.49%)||0.11b|
|Respiratory distress syndrome||30/1202 (2.50%)||187/10 650 (1.76%)||0.070b|
|Intraventicular haemorrhage||3/1202 (0.25%)||51/10 650 (0.48%)||0.26b|
|Necrotising enterocolitis||3/1202 (0.25%)||12/10 650 (0.11%)||0.19c|
|Bacterial sepsis||13/1202 (1.08%)||54/10 650 (0.51%)||0.012b|
|Jaundice||86/1202 (7.15%)||566/10 650 (5.31%)||0.008b|
|Metabolic abnormality||38/1202 (3.16%)||310/10 650 (2.91%)||0.63b|
Exposure to ETS was not reported in 19.1% of otherwise eligible women. We therefore compared the demographic characteristics and outcomes (by univariate analyses) between those women in whom exposure was known (either exposed or not exposed) and those in whom exposure was unknown (Tables 3 and 4). Those in whom exposure was unknown were younger, lighter weight, had a lower level of education and were less likely to have hypertension, have a partner or be working (Table 3). Regarding perinatal outcomes, the babies of women with unknown ETS exposure status were lighter birthweight, had smaller head circumferences and birth lengths, were more likely to be born preterm and were more likely to be stillborn (Table 4). They also had higher rates of NICU admission, endotracheal tube use, respiratory distress syndrome, intraventricular haemorrhage, jaundice and longer length of stay.
Table 3. Comparison between those in whom exposure was reported (either exposed or not exposed) and those in whom exposure was unknown
|Maternal age (years)||29.4 ± 5.3||29.9 ± 5.0||<0.0001a|
|Nulliparous||1420/2798 (50.8%)||6075/11 852 (51.3%)||0.63b|
|BMI (kg/m2)||26.5 ± 6.8||26.9 ± 6.4||0.04a|
|Less than recommended||133/1179 (11.3%)||790/6606 (12.0%)||0.77b|
|As recommended||307/1179 (26.0%)||1682/6606 (25.4%)|| |
|More than recommended||739/1179 (62.7%)||4134/6606 (62.6%)|| |
|Hypertension||267/2798 (9.5%)||1371/11 852 (11.6%)||0.002b|
|Diabetes||109/2798 (3.9%)||418/11 852 (3.5%)||0.35b|
|Partnered||1810/2467 (73.4%)||8546/11 100 (77.0%)||<0.0001b|
|Not working||348/766 (45.4%)||1929/8625 (22.4%)||<0.0001b|
|Part time||53/766 (6.9%)||742/8625 (8.6%)|| |
|Full time||365/766 (47.7%)||5954/8625 (69.0%)|| |
|Level of education|
|Less than high school||132/2328 (5.7%)||356/10 067 (3.5%)||<0.0001b|
|High school||311/2328 (13.4%)||1076/10 067 (10.7%)|| |
|Beyond high school||1885/2328 (81.0%)||8635/10 067 (85.8%)|| |
|Alcohol use in pregnancy||31/2720 (1.1%)||176/11 793 (1.5%)||0.16b|
|Elicit drug use in pregnancy||4/2687 (0.15%)||14/11 833 (0.12%)||0.76c|
Table 4. Perinatal outcomes between those in whom exposure was reported (either exposed or not exposed) and those in whom exposure was unknown
|Head circumference (cm)||34.44 ± 2.10||34.67 ± 1.87||0.002a|
|Birth length (cm)||49.74 ± 3.18||50.00 ± 2.82||<0.0001a|
|Birth weight (g)||3415 ± 707||3500 ± 608||<0.0001a|
|Birth weight <2500 g||211/2798 (7.54%)||582/11 852 (4.91%)||<0.0001b|
|Stillbirth||20/2798 (0.71%)||49/11 852 (0.41%)||0.036b|
|Prelabour rupture of membranes||688/2796 (24.6%)||3020/11 848 (25.5%)||0.33b|
|Preterm birth <37 weeks||349/2790 (12.5%)||993/11 846 (8.38%)||<0.0001b|
|Preterm birth <34 weeks||147/2790 (5.27%)||272/11 846 (2.30%)||<0.0001b|
|Apgar <7 at 1 minute||209/2781 (7.52%)||818/11 806 (6.93%)||0.28b|
|Apgar <7 at 5 minutes||52/2781 (1.87%)||171/11 806 (1.45%)||0.10b|
|Endotracheal tube/use||60/2797 (2.15%)||177/11 843 (1.49%)||0.014b|
|NICU admission||407/2798 (14.5%)||1228/11 852 (10.4%)||<0.0001b|
|Length of stay (days)||4.94 ± 11.14||3.97 ± 9.07||<0.0001a|
|Congenital anomalies||358/2798 (12.8%)||1381/11 852 (11.7%)||0.093b|
|Respiratory distress syndrome||93/2798 (3.32%)||217/11 852 (1.83%)||<0.0001b|
|Intraventriuclar haemorrhage||25/2798 (0.89%)||54/11 852 (0.46%)||0.004b|
|Necrotising enterocolitis||7/2798(0.25%)||15/11 852 (0.13%)||0.17c|
|Bacterial sepsis||24/2798 (0.86%)||91/11 852 (0.77%)||0.077b|
|Jaundice||205/2798 (7.33%)||652/11 852 (5.50%)||<0.0001b|
|Metabolic abnormality||99/2798 (3.54%)||348/11 852 (2.94%)||0.096b|
Multivariate logistic and linear regressions showing outcomes independently associated with ETS exposure are summarised in Table 5. These models adjusted for maternal age, parity, partnered status, work status, BMI, alcohol or illicit drug use, gestational age (for outcomes other than preterm birth) and level of education (for outcomes other than stillbirth). Univariate analysis revealed that education level was not significantly associated with stillbirth in our population (P = 0.29), and so education level was not included in the logistic regression model for stillbirth. Exposure to ETS was independently associated with lighter birthweight, shorter birth length, smaller head circumference, stillbirth, and trends towards preterm birth <34 weeks of gestation and bacterial sepsis. Babies born to women exposed to ETS had a mean birthweight 53.7 g (95% CI 8.9–98.4 g) less than women reporting no exposure to ETS. The head circumference of babies of mothers exposed to ETS was 0.24 cm (95% CI 0.08–0.39 cm) smaller than those not exposed to ETS. The number of women exposed to ETS associated with one additional stillbirth was 117, with one additional preterm birth <34 weeks of gestation was 55, and with one additional incident of neonatal bacterial sepsis was 102.
Table 5. Multivariate linear and logistic regression analyses showing outcomes independently associated with environmental tobacco smoke exposure
|Birthweight (g)||−53.7||−98.4 to −8.9||0.019|
|Head circumference (cm)||−0.24||−0.39 to −0.08||0.003|
|Birth length (g)||−0.29||−0.51 to −0.07||0.010|
|Stillbirth||3.35*||1.16 to 9.72||0.026|
|Preterm birth <34 weeks||1.87*||1.00 to 3.53||0.05|
|Bacterial sepsis||2.96*||0.99 to 8.86||0.05|
Our study confirms the findings of previous researchers, noting an association between ETS and reduced birthweight.12–14 Postulated mechanisms for lower birthweight include reduced oxygenation of the fetus secondary to carbon monoxide as well as nicotine-related vasoconstriction resulting in decreased uterine and placental blood flow.2 We also found that pregnancies exposed to ETS resulted in babies with smaller head circumferences and shorter birth lengths. A previous meta-analysis found a trend towards smaller head circumference, but interestingly a longer birth length, a finding that was felt to be probably due to chance.12
Unlike many earlier studies and meta-analyses,12–14 we did find an association between ETS and preterm birth <34 weeks. We are not able to further assess if these preterm births were the result of spontaneous onset of labour with intact membranes or of premature rupture of membranes or were indicated for maternal or fetal reasons. Active maternal smoking has been associated with preterm prelabour rupture of membranes and preterm birth, although the underlying mechanism is not well understood.1,2 Possible explanations for a link with preterm prelabour rupture of membranes include vasoconstriction disrupting the integrity of the amniotic membranes, possible adverse effects on nutritional status by interfering with protein metabolism, and impaired maternal immunity leading to increased risk of infection.2 We did not, however, find a difference in the overall rate of prelabour rupture of membranes between the two groups. Possible biological explanations for the association of direct maternal smoking and spontaneous onset of labour with intact membranes include placental vasoconstriction and increased levels of catecholamines, both of which may initiate labour.1,2 Direct maternal smoking has been linked to preterm birth so it is not surprising that we noted this association with ETS, despite many previous studies not finding this association.
Our study evaluated a number of perinatal outcomes not evaluated in many previous studies,12 including stillbirth. We found that ETS exposure was associated with stillbirth and bacterial sepsis. As direct maternal smoking has been linked with impaired maternal immunity1,2 it may also interfere with immunity in the newborn infant, and this may be the biological explanation for the association we found between ETS and neonatal bacterial sepsis. Few studies have evaluated the association of ETS and stillbirth, and those studies that did, did not find an association.10,17 Stillbirth has been associated with direct maternal smoking.2 The biological plausibility of this association is not fully understood, but possible causes include its association with growth restriction and placental complications including abruption, as well as a nicotine-induced change in central respiratory control mechanisms.2
To make these associations easy to understand for women, their families and healthcare providers we calculated the number of nonsmoking women exposed to ETS associated with one extra adverse outcome of interest, with the numbers ranging from 55 to 117. For example for every 55 women exposed to ETS one additional preterm birth <34 weeks was noted and for every 117 women exposed to ETS one additional stillbirth was noted.
It is important that the limitations of our study be addressed. Unfortunately, 19.1% of otherwise eligible women did not have ETS exposure recorded in the database, and so were not included in the study. Despite this, we had over 11 000 women in our study who reported either exposure or no exposure to ETS. We did, however, evaluate the demographic characteristics and outcomes of those in whom ETS exposure was unknown compared with those in whom it was known (either exposed or not exposed), noting differences between these two groups. Exposure of ETS was based on women’s self-report as recorded in the prenatal record so we could not verify the exposure. Previous research, however, has suggested that self-reports of second-hand smoke exposure are accurate in most studies18 and are appropriate when studying large populations.19 We were not able to further quantify the exposure, such as the number of hours a day exposed, the degree (such as an enclosed room not ventilated or outdoor environment) or type of exposure (household or work environment), and so are not able to evaluate a possible dose–response relationship. Pregnant women may be reluctant to admit to ETS exposure because of social stigma and so some of these women may be misclassified into the no exposure group. If this occurred, then the true association between ETS exposure and the outcomes of interest may actually be stronger than found in our study. We did not compare nonsmoking women with ETS exposure to women who smoked during pregnancy, as the objective of our current study was to evaluate the effects of ETS on nonsmoking women. In our population of women with singleton gestations delivering during the study period, there were 2862 smokers, of whom 1138 reported ETS exposure, 600 reported no ETS exposure and 1124 with unknown ETS exposure status. We felt that these numbers were too small to include in our study as comparison groups because we would not have adequate power to evaluate the outcomes of interest for these groups. Also we were not able to evaluate women who had smoked in the past but stopped before pregnancy. We only evaluated pregnancy outcomes at 20 weeks of gestation and beyond, and so cannot comment on the effects of ETS on spontaneous abortion rates. Although we controlled for a variety of potential confounders, it is possible that the associations noted with adverse outcomes are the result of other confounders. Importantly, there may be residual confounding by unmeasured risk factors such as socioeconomic status. Socioeconomic status was not consistently recorded in the database so this information could not be used in this study. However, we were able to adjust for partnered status, work status, education level, and alcohol or illicit drug use, some of which may be linked to socioeconomic status. Finally, our population is over 99% Caucasian, and so our findings may not be generalisable to other ethnic groups.
The strengths of the study include the fact that ETS was recorded for the majority of eligible women in the database (81.1%). We were able to adjust for multiple potential confounders including maternal age, parity, partnered status, work status, level of education, BMI, alcohol or illicit drug use, and gestational age at delivery. We evaluated a variety of perinatal outcomes not previously studied, including stillbirth.
Exposure of nonsmoking pregnant women to ETS is associated with a number of adverse perinatal outcomes including lower birthweight, smaller head circumference, shorter birth length, stillbirth, and trends towards preterm birth <34 weeks of gestation and bacterial sepsis. This information is important for women, their families and healthcare providers, and reinforces the continued need for increased public policy and education on prevention of exposure to ETS.
Disclosure of interests
There are no relevant financial, personal, political, intellectual or religious conflicts of interest.
Contribution to authorship
JMGC, contributed to study design, data analysis, and writing and revision of the manuscript; MK, PM and LB contributed to study design, data analysis, and review of the manuscript and DH contributed to study design and review of the manuscript.
Details of ethics approval
Ethics approval for the study was obtained from the Human Investigation Committee of Memorial University and the Research Proposal Approval Committee of Eastern Health (#09.128 July 22, 2009).