Secondary Logo

Journal Logo

Original Article

Environmental Tobacco Smoke and Risk of Spontaneous Abortion

George, Lena*; Granath, Fredrik; Johansson, Anna L. V.*; Annerén, Göran; Cnattingius, Sven*

Author Information
doi: 10.1097/01.ede.0000229984.53726.33
  • Free


Smoking has been associated with spontaneous abortion in some, but not all, studies.1 Most studies of smoking and risk of spontaneous abortion are based on self-reported exposure, which may underestimate the risk in comparison with biochemical measures of exposure.2

Studies of environmental tobacco smoke (ETS) and risk of spontaneous abortion are limited to a few studies of self-reported exposure to ETS or paternal smoke, and the results have been inconsistent.3–7 Studies based on self-reported ETS exposure probably provide even larger validity concerns than self-reported active smoking and do not properly account for all possible exposures at home, work, and in public places.8,9

Measurement of cotinine, a biomarker of nicotine, has been shown to be a valid summary measure of the dose received from ETS as well as from active smoking.10–12 To our knowledge, there are no studies on the association between cotinine measurements of ETS and risk of spontaneous abortion. In the present Swedish case–control study, we have investigated risk of early spontaneous abortion related to both ETS and active smoking as defined by plasma cotinine levels.


Study Participants

This population-based case–control study was conducted in Uppsala County, Sweden, from January 1996 through December 1998. Cases were identified at the Department of Obstetrics and Gynecology of Uppsala University Hospital, which is the only place in Uppsala County for the care of women with spontaneous abortions. Cases were defined as women with spontaneous abortions diagnosed at a gestational age of 6 to 12 completed weeks whose pregnancies had been confirmed by a positive human chorionic gonadotropin test. We identified 652 women as potential case patients, of whom 562 (86%) agreed to participate.

Controls were primarily selected from pregnant women seeking prenatal care in Uppsala County. Potential controls had contacted the prenatal care clinics by phone to book time for their first prenatal visit. We regularly contacted the prenatal care clinics to receive information about potential controls who were frequency-matched to the cases by gestational week. Of 1037 prenatal care patients asked to participate as controls, 953 (92%) agreed. All controls underwent vaginal ultrasonography before the interview to verify the viability of the fetus. In Uppsala County, there are approximately 3 induced abortions for every 10 completed pregnancies, and some of these pregnancies would have resulted in spontaneous abortion if the pregnancy had continued. Women with induced abortions may differ from women continuing their pregnancies in terms of factors associated with risk of spontaneous abortions such as age, smoking, and possibly other lifestyle-related factors. To avoid this potential bias in the selection of controls, women planning to have induced abortions were therefore added to the control group. In total, 310 women who would later undergo induced abortion were asked to participate, and 274 (88%) agreed. Of these women, 75 women were randomly selected and added to the control group, a number estimated according to the gestational age distribution of induced abortions in Uppsala County during the study period. Thus, in all, we included 1028 controls (953 prenatal care patients and 75 women who planned to have an induced abortion). Gestational age was calculated from the first day of the last menstrual period for all cases and controls. The methods of this matched case–control study have been described in detail elsewhere.13

Collection of Data

All in-person interviews with cases and controls recruited among women receiving prenatal care were conducted by 3 midwives during office hours in the outpatient clinic at the Department of Gynecology and Obstetrics. Two doctors conducted interviews with the control subjects who would undergo induced abortions. Ninety percent of cases were interviewed within 2 weeks after the diagnosis of spontaneous abortion, and the remaining 10% were interviewed within 7 weeks. All controls were interviewed within 6 days after their last completed week of gestation. A structured questionnaire was used to reduce bias; the interviewers could not be blinded as to case–control status. At the time of the interview, results of blood sample (plasma cotinine and folate levels) and karyotype analyses were unknown to cases, controls, and interviewers.

Each woman was asked in detail about previous and current smoking habits. Women were asked about all nicotine exposures on a weekly basis beginning 4 weeks before the last menstrual period until the most recently completed week of gestation. Information was collected on active cigarette smoking (number of cigarettes smoked per day), use of oral snuff, and nicotine replacement therapy (chewing gum and transdermal patches). Active smoking of the partner was also recorded. Women were asked about potential risk factors for spontaneous abortion, including sociodemographic, anthropometric, and lifestyle factors, and obstetric and medical history. Women also reported intake of various caffeine sources during each week of pregnancy. Average daily caffeine intake was calculated from the time of estimated conception through the most recently completed week of gestation. The presence and severity of pregnancy-related symptoms (nausea, vomiting, and fatigue) were reported on a week-by-week basis.

Laboratory Methods

Blood samples were obtained from the cases at the emergency ward at the time of miscarriage diagnosis and from the controls when they were interviewed. Blood samples were kept frozen at −80°C until assayed. Plasma cotinine was measured by gas chromatography with use of N-ethylnorcotinine as an internal standard (detection limit 0.1 ng/mL).14 Folate analyses were performed with an immunoassay analyzer (AxSYM; Abbott Laboratories, Abbott Park, IL) using ion capture reaction technology.

Curettage was performed in all women with incomplete spontaneous abortion at diagnosis. Fetal karyotype analysis was possible if chorionic villi were identified in intrauterine tissue obtained by curettage. Cytogenetic analysis was performed using direct preparation, and the chromosomes were banded with Giemsa stain. Eleven cells in metaphase were routinely analyzed, and karyotyping was considered unsuccessful if fewer than 3 cells in metaphase were obtained.

Definition of Smoking Status

The focus of the present investigation was to estimate the risk of spontaneous abortion related to ETS and active smoking. Active smokers were defined as women with a plasma cotinine >15.0 ng/mL. Women were classified as ETS-exposed if they had plasma cotinine concentrations from 0.1 to 15.0 ng/mL and as nonexposed to tobacco smoke if plasma cotinine levels were <0.1 ng/mL.15 There were 508 cases and 1000 controls with information on plasma cotinine.

In our main analyses, we first excluded self-reported users of oral snuff, nicotine chewing gum, or transdermal nicotine patches (23 cases and 58 controls). Next, we excluded 10 cases and 32 controls with cotinine levels from 0.1 to 15.0 ng/mL who stated that they had smoked during pregnancy and 12 cases and 46 controls with cotinine levels <0.1 ng/mL who stated that they had smoked during pregnancy. In all, we included 463 cases and 864 controls in the main analyses.

Statistical Analysis

Data were analyzed with the use of conditional logistic regression analysis using SAS PROC PHREG.16 The controls were frequency-matched to the cases by week of gestation. We calculated odds ratios (ORs) with 95% confidence intervals (CIs). Variables were included in the multivariable analyses (maternal age, country of birth, education, marital status, shift work, parity, previous spontaneous abortions, average caffeine intake during pregnancy, folate levels, and the following pregnancy symptoms: change of eating habits, nausea, vomiting and tiredness) if they were judged a priori to be potential confounders. Interaction with pregnancy symptoms was tested by introducing an interaction term between nausea and cotinine in the logistic model and was assessed by a likelihood ratio test. The interaction between fetal karyotype and ETS exposure status was assessed in a case-only comparison by a χ2 test.

Oral informed consent was obtained from all the women. The study was approved by the ethics committee of the medical faculty at Uppsala University.


The women with spontaneous abortion were older than the control women (29% of cases vs 10% of controls were ≥35 years) and were more likely to be born outside the Nordic countries (10% vs 5%), multiparous (10% vs 4% had ≥3 births), with previous spontaneous abortions (10% vs 3% had ≥2 spontaneous abortions), and to have a higher caffeine intake (14% vs 9% consumed at least 500 mg caffeine daily). Pregnancy symptoms (change of eating habits, nausea, vomiting, and tiredness) were more prevalent and severe among control women compared with cases.

Among both cases and controls, ETS-exposed women (plasma cotinine levels from 0.1 to 15.0 ng/mL) were, with respect to maternal characteristics, previous obstetric history, and caffeine intake, generally more similar to nonexposed women (cotinine level <0.1 ng/mL) than to active smokers (cotinine level >15 ng/mL, Table 1). For example, compared with nonexposed and ETS-exposed women, active smokers were more likely to be low educated, unmarried, multiparous, working shift hours, to have a high caffeine intake during pregnancy, and to have low folate levels.

Characteristics of Cases With Spontaneous Abortion and Controls in Relation to Plasma Cotinine Levels

The distribution of plasma cotinine levels among case and control women was skewed toward zero. Among ETS-exposed women, the median plasma cotinine levels were virtually identical among cases (3.3 ng/mL [90% central range 1.2–7.5 ng/mL]) and controls (3.4 ng/mL [0.8–8.9 ng/mL]). Among active smokers, the median plasma cotinine level was higher in cases (137.5 ng/mL [20.2–327.9 ng/mL]) compared with controls (98.5 ng/mL [22.4–291.9 ng/mL]).

Women with spontaneous abortion were more likely to be ETS-exposed (cotinine levels 0.1–≤15 ng/mL) than the control group (111 [24%] vs 161 [19%], Table 1). Compared with nonexposed women (<0.1 ng cotinine/mL), ETS-exposed women faced an increased risk of spontaneous abortion in the univariable analysis (OR = 1.56; 95% CI = 1.18–2.07). When the analysis was restricted to women with available information on all covariates included in the multivariable model (419 cases and 792 controls), the corresponding OR was 1.65 (CI = 1.22–2.22), and this risk remained essentially unchanged after adjustment for potential confounding factors (Table 2). Compared with nonexposed women, the risk of spontaneous abortion was doubled among active smokers (cotinine levels >15 ng/mL).

Association of Cotinine Levels With Spontaneous Abortion

In an attempt to study the source of ETS exposure, we stratified the analyses according to smoking status of the partner. Among women with a nonsmoking partner (379 cases and 723 controls), the adjusted odds ratio of spontaneous abortion among ETS-exposed compared with nonexposed women was 1.51 (1.02–2.24). Among women with a smoking partner (76 cases and 81 controls), the corresponding OR was 1.47 (0.39–5.61).

In the analyses presented in Table 2, we excluded women who stated that they had used snuff or nicotine replacement therapy during pregnancy (23 cases and 58 controls). In addition, we excluded women from the ETS-exposed and nonexposed groups who stated that they had been smoking during pregnancy (22 cases and 78 controls). To investigate whether these exclusions influenced the results, we performed a sensitivity analysis including all women with information on plasma cotinine (508 cases and 1000 controls). The adjusted odds ratio of spontaneous abortion related to ETS exposure was then similar to the risks presented in Table 2 (OR = 1.58; CI = 1.14–2.20). When we restricted the ETS and nonexposed groups to women stating that they were lifelong nonsmokers (361 cases and 704 controls), the corresponding adjusted OR was 1.90 (1.26–2.86).

It has been discussed whether impending spontaneous abortions are associated with lack of nausea, which in turn causes women to be more active and thus more likely to be exposed to various environmental agents.17 To investigate whether nausea modified the ETS-associated risks, we performed interaction analyses between cotinine and nausea with regard to risk of spontaneous abortion. In the first model, nausea was defined as the highest level reported during any week of pregnancy (also used in our main analyses); and the P value for the interaction analysis was 0.58. In the second model, nausea was defined as the nausea level reported at the last completed week of gestation (to time the history of nausea more closely to cotinine measurements). The P value was 0.50. Thus, we found no evidence for effect modification by nausea on the cotinine-associated risks.

Among women with spontaneous abortion, chromosomal analysis revealed that 85 fetuses (51 male and 34 female) had normal karyotype, 131 had abnormal karyotype, and 247 had unknown karyotype. Of spontaneous abortions with abnormal fetal karyotype, 89 were autosomal aberrations, 11 were sex chromosomal trisomies (47, XXX; 47, XXY; 47, XYY), 13 were X-monosomies (45, X), and 18 were triploidies or tetraploidies. We found that ETS-exposed women and active smokers were at increased risk of spontaneous abortion of unknown and abnormal fetal karyotype compared with nonexposed women (Table 3). Analysis of spontaneous abortion with normal fetal karyotype included only 75 cases, and ETS exposure and active smoking were associated with only modestly increased risks of spontaneous abortion of normal fetal karyotype. However, when we performed an interaction analysis based on a case-only comparison stratified by fetal karyotype among ETS-exposed and nonexposed women, the overall test of an association between cotinine levels and karyotype-specific abortions had a P value of 0.94.

Odds Ratios for Spontaneous Abortion Associated With Cotinine Levels According to Fetal Karyotype


To our knowledge, this is the first study on the relation between ETS and risk of spontaneous abortion based on cotinine measurements. We found that both nonsmoking women exposed to ETS and active smokers were at increased risk of spontaneous abortion. The power to study spontaneous abortion by karyotype was limited, because our analyses included only 75 cases with normal fetal karyotypes and 120 with abnormal fetal karyotypes. We could not show a differential effect of exposure to ETS or active smoking between normal and abnormal fetal karyotype abortions.

The biologic mechanisms underlying a possible association between ETS and spontaneous abortion may involve pathways similar to those for active smoking, because sidestream tobacco smoke contains many of the same constituents as mainstream tobacco smoke. Several components of tobacco smoke (eg, nicotine, carbon monoxide, and cyanide) are toxic for the developing fetus. Nicotine has vasoconstrictive effects leading to reduced placental blood flow. Carbon monoxide binds to hemoglobin, causing maternal and fetal hypoxia,1 which may interfere with the development of the growing conceptus and induce fetal demise.

We used plasma cotinine measurements to define ETS exposure. Cotinine is a well-accepted biomarker of ETS exposure, and studies using self-reported exposure to ETS are likely to misclassify some exposed women as unexposed.8,9 Because raised cotinine levels could be caused by other nicotine-containing products,18 we decided a priori to exclude women using oral snuff or nicotine replacement therapy. In addition, because daily smokers who stopped smoking in early pregnancy may have elevated cotinine levels, these women were excluded from the ETS-exposed and nonexposed groups. However, we found in supplementary analyses that these restrictions had only minor effects on the ETS-related risk of spontaneous abortion, indicating robustness of our findings.

Spousal smoking is often used as a proxy of ETS exposure.4–7 However, women may also be exposed to other sources of ETS exposure, and women living with a smoking partner may not necessarily be exposed to ETS. The quality of self-reported exposure assessment may account for much of the inconsistent results among previous studies on ETS exposure and spontaneous abortion.3–7 It has also been hypothesized that spousal smoking may increase the risk of spontaneous abortion through direct effects of active smoking on sperm.7 In a stratified analysis, we found similar ETS-associated risk estimates for spontaneous abortion among women with smoking and nonsmoking partners (OR = 1.5). Consequently, we believe that we can rule out the possibility that our findings are related solely to active paternal smoking.

We found that active smoking was associated with a 2-fold increase in risk of spontaneous abortion (OR = 2.1). This is a slightly higher risk than was found in previous studies using cotinine measurements. In the study by Ness et al,2 the OR was 1.8 and in our own investigation,13 based on the current case–control study, the OR was 1.5. Because smoking status was dichotomized in these studies, the reference groups included ETS-exposed individuals. Thus, by making cases and controls more similar according to nicotine exposure status, the association with the outcome is probably biased toward the null value. The same misclassification, resulting in underestimation of effects, is probably also present in studies in which smoking status is assessed with self-reported dichotomized information.19

We were also able to adjust for a wide range of potential confounding factors, which had little effect on the results. It is important to consider the possibility of confounding by active smoking. Women exposed to ETS resembled nonexposed women more than active smokers with respect to other potential risk factors, which reduces the plausibility of confounding by unmeasured risk factors or by active smoking. The cutoff level between active smoking and ETS exposure was set to 15 ng/mL15; active smokers had cotinine levels well above this level, and almost all women classified as ETS-exposed had cotinine levels below 10 ng/mL. Considering that we could see no tendency of overlap also strengthens our belief that it is unlikely that active smokers were misclassified as nonsmokers exposed to ETS.

To take confounding by nausea into account, we adjusted for pregnancy symptoms in the analyses. Nausea did not explain the association between ETS exposure and spontaneous abortion. Nevertheless, the relation between environmental exposures (such as tobacco smoke), symptoms of pregnancy (such as nausea), and fetal viability is complex. Women with a viable pregnancy experience more nausea and, it has been proposed that this in turn may cause women to avoid various environmental exposures.17 However, in interaction analyses between cotinine and nausea with regard to risk of spontaneous abortion, we found no evidence of an effect modification by nausea.

This study also has limitations. Although plasma cotinine is a precise indicator of exposure over the recent past, a single cotinine measurement is an imperfect marker of the exposure over the etiologic period of interest. The half-life of cotinine is approximately 17 hours for nonpregnant women12 and close to 9 hours among women between 16 and 40 weeks of gestation.20 We did not have information about the time when each blood sample was collected, and systematic differences between cases and controls may have influenced levels of plasma cotinine. In controls, blood was collected during office hours (ie, from 8 am to 4 pm). Although most women with pregnancy bleedings contacted the Department of Obstetrics and Gynecology by phone and were scheduled for a visit during office hours, blood samples were, among cases, also collected in evenings and sometimes at night. We also lacked specific information regarding where the cases and controls spent the hours and days before cotinine measurements. If recency or amount of ETS exposure differed between cases and controls, we would have expected differences in prevalence of ETS exposure as well as differences in the distribution of cotinine values among these classified as being exposed to ETS. However, median plasma cotinine levels were almost the same for cases and controls exposed to ETS, and the distribution of cotinine values among cases and controls exposed to ETS was virtually identical (data available at request) indicating that timing and dose of exposure for cases and controls were similar. Thus, possible misclassification of ETS exposure based on a single cotinine measurement should be nondifferential between cases and controls.

We restricted the definition of ETS to women who stated that they were nonsmokers during pregnancy; still, we cannot rule out that a fraction of women classified as ETS-exposed were smoking cigarettes on an intermittent basis. However, occasional smokers are probably former active smokers and would thus have had maternal characteristics similar to active smokers. In addition, when we restricted the analyses to women who stated that they never in life had been smokers, risk of spontaneous abortion related to ETS became, if anything, even more pronounced.

In conclusion, the high prevalence of ETS exposure of nonsmoking women in our study (24% among cases and 19% among controls) is consistent with the prevalence in other populations of pregnant women for whom cotinine measurements were obtained to define ETS exposure.15,21–23 Given the high prevalence of exposure to ETS and the fact that spontaneous abortion is the most common adverse outcome of pregnancy,24,25 the public health consequences of passive smoking regarding early fetal loss may be substantial.


1. Health Consequences of Tobacco Use Among Women. Women and Smoking: A Report of the Surgeon General. Rockville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office of Smoking and Health; 2001:177–450.
2. Ness RB, Grisso JA, Hirschinger N, et al. Cocaine and tobacco use and the risk of spontaneous abortion. N Engl J Med. 1999;340:333–339.
3. Ahlborg G Jr, Bodin L. Tobacco smoke exposure and pregnancy outcome among working women. A prospective study at prenatal care centers in Orebro County, Sweden. Am J Epidemiol. 1991;133:338–347.
4. Windham GC, Swan SH, Fenster L. Parental cigarette smoking and the risk of spontaneous abortion. Am J Epidemiol. 1992;135:1394–1403.
5. Chatenoud L, Parazzini F, di Cintio E, et al. Paternal and maternal smoking habits before conception and during the first trimester: relation to spontaneous abortion. Ann Epidemiol. 1998;8:520–526.
6. Windham GC, Von Behren J, Waller K, et al. Exposure to environmental and mainstream tobacco smoke and risk of spontaneous abortion. Am J Epidemiol. 1999;149:243–247.
7. Venners SA, Wang X, Chen C, et al. Paternal smoking and pregnancy loss: a prospective study using a biomarker of pregnancy. Am J Epidemiol. 2004;159:993–1001.
8. Benowitz NL. Biomarkers of environmental tobacco smoke exposure. Environ Health Perspect. 1999;107(suppl 2):349–355.
9. DeLorenze GN, Kharrazi M, Kaufman FL, et al. Exposure to environmental tobacco smoke in pregnant women: the association between self-report and serum cotinine. Environ Res. 2002;90:21–32.
10. Jarvis M, Tunstall-Pedoe H, Feyerabend C, et al. Biochemical markers of smoke absorption and self reported exposure to passive smoking. J Epidemiol Community Health. 1984;38:335–339.
11. Haddow JE, Knight GJ, Palomaki GE, et al. Cigarette consumption and serum cotinine in relation to birthweight. BJOG. 1987;94:678–681.
12. Benowitz NL. Cotinine as a biomarker of environmental tobacco smoke exposure. Epidemiol Rev. 1996;18:188–204.
13. Cnattingius S, Signorello LB, Anneren G, et al. Caffeine intake and the risk of first-trimester spontaneous abortion. N Engl J Med. 2000;343:1839–1845.
14. Olsson P, Kuylenstierna F, Johansson C-J Pharmacokinetics of nicotine after intranasal administration. In: Adlkofer F, Thurau K, eds. Effects of Nicotine on Biological Systems: Advances in Pharmacological Sciences. Basel, Switzerland: Birkhäuser; 1991:57–61.
15. Peacock JL, Cook DG, Carey IM, et al. Maternal cotinine level during pregnancy and birthweight for gestational age. Int J Epidemiol. 1998;27:647–656.
16. Changes and Enhancements Through Release 6.11 [computer program]. Cary, NC: SAS Institute Inc; 1996.
17. Savitz DA. Magnetic fields and miscarriage. Epidemiology. 2002;13:1–4.
18. Dempsey DA, Benowitz NL. Risks and benefits of nicotine to aid smoking cessation in pregnancy. Drug Saf. 2001;24:277–322.
19. Cnattingius S. The epidemiology of smoking during pregnancy: smoking prevalence, maternal characteristics, and pregnancy outcomes. Nicotine Tob Res. 2004;6(suppl 2):S125–140.
20. Dempsey D, Jacob P 3rd, Benowitz NL. Accelerated metabolism of nicotine and cotinine in pregnant smokers. J Pharmacol Exp Ther. 2002;301:594–598.
21. Eskenazi B, Prehn AW, Christianson RE. Passive and active maternal smoking as measured by serum cotinine: the effect on birthweight. Am J Public Health. 1995;85:395–398.
22. Rebagliato M, Florey Cdu V, Bolumar F. Exposure to environmental tobacco smoke in nonsmoking pregnant women in relation to birth weight. Am J Epidemiol. 1995;142:531–537.
23. Haddow JE, Knight GJ, Palomaki GE, et al. Second-trimester serum cotinine levels in nonsmokers in relation to birth weight. Am J Obstet Gynecol. 1988;159:481–484.
24. Wilcox AJ, Weinberg CR, O'Connor JF, et al. Incidence of early loss of pregnancy. N Engl J Med. 1988;319:189–194.
25. Nybo Andersen AM, Wohlfahrt J, Christens P, et al. Maternal age and fetal loss: population based register linkage study. BMJ. 2000;320:1708–1712.
© 2006 Lippincott Williams & Wilkins, Inc.