Active smoking has long been considered deleterious to health and a risk factor for a number of diseases.1 Cigarette smoke contains more than 4000 chemicals and at least 60 known carcinogens.2 Non-smoking individuals subjected to secondhand smoke (SHS) are exposed to these same toxins, but in varying amounts.3 Research has found that sidestream smoke, the smoke from the tip of a cigarette, contains greater amounts of tar, nicotine and other chemicals than mainstream smoke.4 5 SHS also contains a number of substances that are believed to be reproductive toxins, including carbon monoxide, nicotine, anabasine, cadmium, benzene, toluene, formaldehyde and others.3 In recent years, the number of people exposed to SHS has declined principally because of the decrease in smoking prevalence and because of restrictions placed on smoking.6 Despite these advances, 65% of non-smokers are still exposed to SHS as detected by serum cotinine.7 However, exposure to SHS as an adult may be lower because of the clean indoor air laws.8

The effects of tobacco use and exposure on pregnancy outcomes remain a public health priority because 15% of mothers continue to smoke throughout pregnancy, and an estimated 43 million women in the United States are exposed to cigarette smoke from others.9 10 Several articles have linked smoking during pregnancy to increased risk of spontaneous abortions.11 12 Other research has shown that maternal smoking during pregnancy is linked to a higher rate of sudden infant death syndrome (SIDS)11 13 in addition to decreased fecundity.14 15

Although many studies link maternal smoking to various pregnancy difficulties, fewer studies have examined SHS exposure and pregnancy difficulties. Two studies published in the early 1990s showed that exposure to SHS was associated with an increase in spontaneous abortions.16 17 More recent research found that SHS exposure during pregnancy was associated with an increased occurrence of spontaneous abortion.3 18 Other studies found that exposure to SHS early in life (in utero or in childhood) was linked with miscarriage19 20 and reduced fecundity in adulthood.21 22

Because evidence linking exposure to SHS and adverse pregnancy outcomes was limited, we conducted a study investigating these associations. The aim of this study was to assess the relation between both childhood and adult SHS exposure and a variety of pregnancy difficulties. This study population was selected because of the detailed information on pregnancy difficulties in addition to data on SHS exposure that existed from women who presented for care at the Roswell Park Cancer Institute.

METHODS

Study population

The population for this study included female patients who were seen at Roswell Park Cancer Institute, Buffalo, NY, between 1982 and 1998. At their first visit, each patient was invited to complete an epidemiological questionnaire (called the Patient Epidemiology Data System or PEDS). Approximately 50% of new patients completed and returned the questionnaire. Eligible participants were females who reported at least one previous pregnancy attempt, and were lifelong never-smokers. Of the 4804 women who were considered to be eligible, 2926 were diagnosed with a malignant disease. The remaining 1878 patients were considered to be cancer-free. The women included in this study were predominately white (99%) and ranged in age from 18 years to 96 years, with a median age of 58 years.

Dependent variables

All the participants in this study completed the PEDS questionnaire. This 16-page instrument gathered information on reproductive and medical histories, family history of cancer, occupational and environmental exposures, tobacco and alcohol use and diet.23 Eligibility for this study was restricted to those who answered “yes” to the question, “Have you ever been pregnant?”

Fetal loss

Fetal losses included affirmative responses to any of the following survey items: “Did you ever have a stillbirth (carried for 5 months or more)” and, if so, how many? “Have you ever had a miscarriage (carried for less than 5 months)” and, if so, how many? Multiple fetal losses were defined as any combination of two of more fetal losses.

Fecundity

Difficulty becoming pregnant was based on affirmative responses to survey items that assessed for difficulty in becoming pregnant persisting for more than one year, which is consistent with the accepted medical definition.24

Fetal loss or difficulty becoming pregnant

Respondents who reported either at least one fetal loss (or example, miscarriage or stillbirth) or who reported difficulty becoming pregnant lasting more than 12 months were classified as having a history of fetal loss or difficulty become pregnant.

Although while previous manuscripts had linked SHS exposure to other pregnancy outcomes, such as SIDS and low infant birth weight, these variables were not assessed in the PEDS questionnaire.

Independent variables

SHS exposure during childhood and as an adult was captured by the PEDS questionnaire, and the study was restricted to lifelong never-smokers in order to assess the sole effect of SHS. For adult SHS exposure the questionnaire asked the number of hours each day a woman is exposed to the smoke from others’ cigarettes in (a) her home, (b) at work and (c) other locations. For past SHS exposure, participants were asked “When you were a child living at home, did either of your parents smoke on a regular basis?” Based on these questions, several exposure variables were created: parental SHS exposure (no, yes), adult SHS exposure (no, yes), adult and childhood dichotomous exposure (childhood exposure but no adult exposure, adult exposure but no childhood exposure, both adult and childhood exposure, and neither childhood nor adult exposure), and total adult SHS hours (divided into four categories: no exposure and tertiles based on hours exposed).

Control variables

The PEDS questionnaire included items on diet, alcohol consumption and other lifestyle factors, allowing multiple items to be considered as potential confounders.10 20 The inclusions of control variables was based on biological and statistical rationales,25 and included age at most recent pregnancy, marital status, education, income, birth control pill use (dichotomous), alcohol consumption (categorical), caffeine consumption (categorical), year of visit (continuous), previous miscarriage (dichotomous), problems getting pregnant (dichotomous), number of children (categorical), body mass index (categorical), menstrual irregularity (dichotomous), race (categorical) and age at admission (continuous).

Statistical analyses

For descriptive statistics, χ² statistics were calculated to test the difference between those who had any adverse pregnancy outcome and those who had no adverse pregnancy outcomes for categorical variables, and Student t tests were used to calculate the difference for continuous variables. Given the exploratory nature of this research, the results were consider statistically significant at the p<0.05 level. Potential confounders considered for these analyses are listed above. All analyses were performed using SPSS version 15.0.

Unconditional multivariate logistic regression was used to calculate odds ratios (OR) and corresponding 95% confidence intervals (CI). None of the covariates met the classic definition of a confounder, where the covariate was associated with both the outcome and the exposure. Model selection was performed in a stepwise forward manner. Covariates were added to the model and included whether the covariate changed the odds of the main parameter by >10%.20 Variables that were evaluated but not included in the final multivariate model included marital status, education, income, oral contraceptive use, alcohol and caffeine consumption, year of visit, number of children, body mass index, menstrual irregularity, race and age at admission.

RESULTS

Approximately 32% of lifelong never-smoking women experienced at least one episode of a fetal loss, with 12% of women experienced multiple episodes of fetal loss. More than 11% reported trouble becoming pregnant. Overall, more than 40% of women reported having some form of prenatal pregnancy difficulties. The prevalence of prenatal pregnancy difficulties in this study was consistent with results from other research using populations of women without cancer.26 27 Exposure to SHS declined as date of admission became more recent.

Descriptive characteristics of those women who did and did not experience pregnancy difficulties are given in table 1. Those who reported having experienced pregnancy difficulties were significantly (p<0.05) younger at admission and older at the time of their most recent pregnancy. Pregnancy difficulties did not differ by marital status, but those with a higher family income and those who attained a higher educational level were more likely to have experienced pregnancy difficulties. While caffeine consumption was not associated with pregnancy difficulties (p = 0.78), women who consumed a greater amount of alcohol reported a higher proportion of pregnancy difficulties (p<0.01). Women who reported pregnancy difficulties had fewer children and a greater proportion of menstrual irregularity (p<0.01).

Table 2 displays the results examining the association between childhood SHS exposure and the various study outcomes based upon unadjusted and adjusted models. Women exposed as children to SHS from their parents reported trouble becoming pregnant (childhood SHS; OR = 1.27; 95% CI 1.03 to 1.56) more often than those not exposed to SHS during childhood. Childhood SHS exposure was not associated with the other study outcomes.

Table 3 gives the results examining the association between SHS exposure as an adult and the various pregnancy difficulties. Women exposed to SHS as an adult reported trouble becoming pregnant (adult SHS; OR = 1.24; 95% CI 1.03 to 1.51) more often than those not exposed to SHS. SHS exposure as an adult was associated with fetal loss (adult SHS; OR = 1.23; 95% CI 1.08 to 1.40). Lastly, adult SHS exposure increased the odds of either pregnancy difficulties or fetal loss (adult SHS; OR = 1.30; 95% CI 1.15 to 1.47).

Table 4 gives the association between combined childhood and adult SHS exposure with pregnancy difficulties. Combined adult and childhood SHS exposure was associated with difficulty becoming pregnant (OR = 1.68; 95% CI 1.28 to 2.20). Women exposed to SHS both as an adult and during childhood had significantly increased odds of fetal loss (OR = 1.39; 95% CI 1.17 to 1.66). Multiple fetal losses were examined in relation to adult and childhood SHS, where the odds ratios found were similar to that of fetal loss (OR = 1.62; 95% CI 1.25 to 2.11).

Table 5 displays the results for pregnancy difficulty odds by tertiles of adult SHS exposure. Those reporting exposure for 6 hours or more a day to SHS had a significantly higher likelihood of fetal loss (OR = 1.30; 95% CI 1.11 to 1.53). There was a positive trend (p<0.05) for increasing SHS hours for odds of multiple fetal loss. Women exposed to 6 hours or more of SHS were also at increased odds of difficulty becoming pregnant (OR = 1.36; 95% CI 1.08 to 1.72), along with a positive trend (p = 0.03). Women in the highest exposure tertile had a significant 40% increase in odds ratio of any pregnancy difficulties (OR = 1.39; 95% CI 1.19 to 1.62; p for trend = 0.01).

DISCUSSION

This study found evidence that self-reported exposure to SHS (both as an adult and during childhood), was associated with an increase in the odds of various prenatal pregnancy difficulties. Various SHS exposure measures were consistently associated with fetal loss, difficulty becoming pregnant (fecundity) and either of these outcomes. Statistically significant increases in the odds for these outcomes ranged between 25% and 70%, and the strongest associations were observed for those with the greatest SHS exposures. In addition, statistically significant positive trends were noted for adult SHS exposure hours and pregnancy difficulties, indicating a dose-response relation. While significant associations were noted for adult SHS exposures and both fetal losses and difficulty becoming pregnant, childhood SHS exposures were associated only with difficulty in becoming pregnancy. No indication of interaction was noted between childhood and adult SHS when included in the same logistic model. No signs of interaction were apparent when the results were stratified by those diagnosed with cancer and non-malignant conditions. The addition of multiple potential confounding variables to the regression models did not dramatically alter crude associations between SHS exposure and pregnancy difficulties. Furthermore, women with a considerable active smoking history were found to have significantly increased odds of experiencing pregnancy loss and difficulty becoming pregnant. However, owing to the prevalent nature of these data, the results must be interpreted with caution, and causality cannot be concluded.

The results presented in this study are consistent with a number of previous publications examining the association between SHS exposure and various pregnancy difficulties.3 16^–20 The odds ratio estimates obtained in this study are similar in magnitude to the risk estimates in the studies that reported an association between SHS and pregnancy difficulties.17 28 Although a small number of studies have found an association between SHS and pregnancy difficulties, not all studies have reported a positive association.29

Overall, the exact aetiologies of prenatal pregnancy difficulties remain uncertain. Some mechanism, which may explain these occurrences, include chromosomal and uterine abnormalities, endocrine abnormalities, stress, exposure to environmental factors (for example, tobacco smoke) and unidentified factors.30 Compared to mainstream smoke (inhaled by the smoker) diluted sidestream smoke (the major component of SHS), especially after it has aged, can be more toxic.31 DNA comet assays demonstrate more cells with DNA damage in never-smokers exposed to SHS than in active smokers.32 Likewise, toxins that create more reactive oxygen species (ROS) with resulting oxidative stress are also found in higher concentrations in sidestream smoke.33

It is possible that SHS exposure involves the inhibition of reproductive hormone synthesis (oestrogens and progesterones) at various sites including the ovary, the fat stores and the placenta if the woman is pregnant. Aromatase converts androgens to estradiol in the granulosa cells of ovarian follicles and in placental trophoblasts. Nicotine, cotinine and anabasine from tobacco smoke are endocrine disruptors that inhibit aromatase activity in granulosa cells and in trophoblasts, which reduce fertility.34 In addition, SHS constituents produce changes in cervical mucus, which can impair fertility.15 35

Although oestrogens are associated with fertility and fecundity, progesterone is strongly linked to ovarian cyclicity and maintenance of pregnancy.36 Cadmium, at levels found in SHS, inhibits the synthesis of progesterone in trophoblasts and interferes with multiple intrafollicular processes including meiotic maturation and oocyte cumulus expansion.36 Nicotine, cotinine and anabasine also inhibit progesterone synthesis in granulosa cells.37 Oxidative stress, generated by hypoxia and detoxification of xenobiotics via phase I enzymes and more so, phase II enzymes (the glutathione S transferase family), is implicated in pregnancy difficulties, especially spontaneous abortion, mainly as a result of endothelial cell damage to placental vessels.26 All of the mechanisms are induced via exposure to SHS and may account for the findings of this study.

The present study had a number of advantages and disadvantages. The data used in this study are prevalence data. Because the participants were reporting on past behaviours, it is unknown whether their reported exposure levels were those that prevailed at the time of the prenatal pregnancy difficulty. Research shows that SHS exposure steadily decreases over a woman’s lifetime.38 If misclassification was present, it was most probably in the conservative direction, underestimating SHS exposure, which would bias the odds estimates towards the null.39 While recall bias may be a limitation of this study, it is unclear if this would be differential between cases and controls, especially for an exposure that was generally considered commonplace and socially acceptable. Because of the prevalent nature of the data, it is impossible to establish temporality. Another limitation of this study is the potential of misclassification of the exposure and the various outcomes, since the data used in this study were self-reported. For example, the questionnaire only asked about parental smoking in the home during childhood, but there could have been other relatives living in the house who smoked. Much of this misclassification was an underestimation of their true SHS exposure and there is no evidence to suggest the bias was differential in any direction.40 Data presented in table 5 show a U-shaped relation between hours of daily exposure and study outcomes. While the point estimate for those in 2–5 hours/day category is slightly less than those in the 0.5–1.5 hours/day category, their confidence intervals overlap considerably. Because of this, we have relied on the overall test for trend across the categories as noted on the table. Lastly, only 50% of eligible women who came to Roswell Park Cancer Institute completed the PEDS questionnaire. There is no way to determine whether those individuals who did not complete the questionnaire differed from participants in this study with respect to SHS exposure and the occurrence of pregnancy difficulties, which could affect the generalisability of the results. Other authors have used the PEDS database for case-control studies examining cancer odds and successfully replicated those associations.23 41 Furthermore, both the descriptive characteristics and odds found in this study are consistent with previous literature examining the association between SHS and pregnancy difficulties.10 17 28

Strengths of this study include the diversity of information collected on a number of variables. The exposure variables used in this study included both past exposure to SHS from parents and adult exposure at home, at work and in other locations. This study is also one of the few with the ability to examine the association between SHS exposure and a number of prenatal pregnancy difficulties. A large number of potential confounders that were used in previous publications were used in this study, which reduced the likelihood that the results seen in this study were due to confounding.10 20 Other strengths of this study include the number of women included in the statistical analyses. This study used approximately 4800 lifelong never-smoking women to examine the association between SHS exposure and pregnancy difficulties. The large sample permitted this study to stratify exposures by multiple levels, which allowed for the observation of dose-response relations while maintaining adequate statistical power. Lastly, another strength of this study is the consistency of the association observed across the various SHS exposure measures for each of the various pregnancy difficulties.

The results from this study are consistent with results from a number of other studies that linked SHS exposure to various adverse pregnancy outcomes.3 16^–22 Although one cannot infer from this study that exposure to SHS causes various adverse pregnancy outcomes, this study adds valuable evidence to the current scientific knowledge and demonstrates the imperative need for further research into this area. The significance of the associations between SHS exposure and adverse pregnancy outcomes underscores the public health doctrine that all people, especially women in their reproductive years, should be fully protected from tobacco smoke. Based on the current state of knowledge, clinicians are encouraged to strongly recommend smoking cessation and the reduction SHS exposure to women of child bearing age, and to their household contacts, in an attempt to minimise prenatal pregnancy difficulties.