Population and methods

All pregnant women booking for delivery at the Department of Obstetrics and Gynaecology, Aarhus University Hospital, from September 1989 to August 1996 were invited to participate in the study. The women were asked to fill in three questionnaires: the first two before the routine antenatal visit at 16 weeks of gestation, and the third before the visit at 30 weeks of gestation.

Information from the first questionnaire was used to establish the women's medical record. It provided information on medical and obstetric history, maternal age, and smoking habits before pregnancy and during the first trimester, and alcohol intake during pregnancy. The second questionnaire provided information on marital status, education, occupational status, and caffeine intake during pregnancy. The third questionnaire was used only to register smoking habits during pregnancy. Information about delivery was obtained from a birth registration form filled in by the attending midwife immediately after delivery. Before data entry, all birth registration forms were manually checked and compared with the medical records by a research midwife who also registered all complications during the first days after delivery.

Information about deaths during the first year of life was obtained from the Registry of Causes of Death, administered by the Danish National Board of Health, and from the Danish Civil Registration System, administered by the Danish government. Furthermore, we reviewed the hospital records of all cases of death before one year of life to validate the cause of death. The International Classification of Diseases, 8th Revision (ICD-8) was used in Denmark to classify causes of death through 1993, and from January 1994 the 10th Revision (ICD-10) was used. SIDS was defined as death ICD-8 code 795.80 and ICD-10 code R95.9. The medical records of the children with SIDS were manually checked to ensure that the diagnosis was correct

The study population was restricted to Danish speaking women who completed the first questionnaire and carried their pregnancy to a live birth of a single child (n =  24 986). This number corresponds approximately to 90% of all Danish speaking women booking for delivery during the study period. The second questionnaire was completed by 18 517 (74%) women, and the third by 16 436 (66%). Information about birth weight and gestational age at delivery was missing in 87 and four cases, respectively.

The women were defined as smokers if they smoked one or more cigarettes per day when 16 or 30 weeks pregnant. Only minimal changes in smoking habits appear after 16 weeks of gestation.19 Smoking was analysed in two categories (smokers versus non-smokers), and also in ordered categories: 0, 1–9, and 10 or more cigarettes per day.

STATISTICAL ANALYSIS

The association between smoking during pregnancy and SIDS is presented as odds ratios (OR) with 95% confidence intervals (CI). Potential confounding variables were categorised as in table 1. They were evaluated in logistic regression analyses, and if they changed the measure of association between smoking and SIDS by more than 10% they remained in the final model.20 All covariates were entered as ordinal, creating a number of dummy variables equal to the number of categories minus one. To take into account the time of death after delivery, all measures of association were evaluated in a Cox regression analysis. However, as the results were similar to those from logistic regression analyses, they are not presented. Differences in mean birth weight were evaluated by analysis of variance. The attributable fraction (AF) was calculated from the following formula21:

View this table:

• View inline
• View popup

Table 1

SIDS according to smoking habits during pregnancy and other lifestyle and sociodemographic factors

The study was approved by the regional ethics committee and by the Danish National Board of Health.

Results

The study group consisted of 24 986 live born children. In 7450 (30%) pregnancies the mother was categorised as a smoker and in 17 536 (70%) pregnancies as a non-smoker. Among those categorised as non-smokers, 2642 (15%) had smoked before pregnancy but stopped during the first trimester. Among smokers, 3249 (44%) smoked 1–9 cigarettes per day, and 4201 (56%) 10 or more cigarettes per day. The mean number of cigarettes smoked per day was 9 (5).

The overall rate of SIDS was 0.80 per 1000 live births (n = 20). The median age at death among children who died from SIDS was 69 days (25% percentile 40 days; 75% percentile 109 days). The mean age at death was 27 days (95% CI −15 to 69 days) less among children born to women who had smoked during pregnancy compared with children of non-smokers.

The crude analysis showed that children of smokers had more than three times the risk of SIDS than children of non-smokers (table 2). The risk of SIDS increased with the number of cigarettes smoked per day (test for trend p < 0.05; table 2). Adjustment for parity, alcohol, and caffeine intake during pregnancy, maternal height and weight before pregnancy, years of school, occupational status, marital status, and number of antenatal care visits did not change the results. Adjustment for mother's age reduced the risk of SIDS associated with smoking (table 2). Mean birth weight was 323 g (95% CI 81 to 566 g) lower among children who died from SIDS than among surviving children. Adjustment for birth weight and gestational age at birth also reduced the risk of SIDS associated with smoking (OR = 2.9; 95% CI 1.2 to 7.2). Conclusions were not changed after adjustment for maternal age, birth weight, and gestational age at birth.

Discussion

In this prospective study of smoking during pregnancy and SIDS, children of women who smoked during pregnancy had an increased risk of SIDS compared with children born to women who had not smoked during pregnancy. Despite a small numbrer of cases and controlling for possible confounders the association was statistically significant. The risk of SIDS increased with the number of cigarettes smoked per day during pregnancy.

Information about smoking habits during pregnancy was obtained from a self administered questionnaire completed by the pregnant women at 16 and 30 weeks of gestation. Because the data were collected prospectively, information about smoking could not be biased by the parents' knowledge about the death of their child. With only 20 cases, it was not meaningful to assess any differential effect between smoking at weeks 16 and 30. Furthermore, among 8607 women (34%) information was missing about smoking habits at 30 weeks of gestation. In our population 30–40% of the smokers stopped smoking in early pregnancy, but after the first trimester very few stopped smoking.22 ,23 Thus, smoking habits at 16 weeks of gestation are a valid measure of smoking habits throughout pregnancy.

Women who stopped smoking before 16 weeks of gestation were categorised as non-smokers in the present study. However, if exposure in early pregnancy is associated with an increased risk of SIDS, compared with no exposure during the entire pregnancy, our result is an underestimation of the true association between smoking during pregnancy and SIDS.

Owing to careful prospective collection of information about maternal lifestyle, and sociodemographic and obstetric factors, we could adjust for a variety of potential confounders. Adjustment for maternal age marginally decreased the risk of SIDS associated with smoking. However, maternal age may represent a proxy for other factors associated with smoking and the occurrence of SIDS, for example, attitudes towards childcare. It is possible that controlling for those factors would further decrease the risk of SIDS associated with smoking. Furthermore, information about socioeconomic markers was missing in one third of all women which may constrain the possibility to fully adjust for these factors.

We did not have information on smoking by the mothers after delivery. However, this is likely to be closely associated with smoking habits during pregnancy. For outcomes assessed shortly after birth, any attempt to discriminate between effects of maternal smoking during pregnancy or after birth remains speculative. It is therefore possible that the association identified in this study may be caused by smoking exposure after delivery.

Unlike Schellscheidt and coworkers24 we found that SIDS children had a lower mean birth weight than with surviving infants. Some children might be more vulnerable to tobacco exposure in utero, but it is also possible that women whose children died from SIDS smoked more than they actually reported, or smoked differently from mothers with surviving infants. Biochemical measures of tobacco exposure could throw light on this problem. The risk of SIDS was marginally reduced after adjustment for birth weight. Thus our result may indicate that smoking increases the risk of SIDS in addition to its effect on birth weight and preterm delivery.

Our study adds to earlier evidence for an association between smoking during pregnancy and SIDS. The study's strengths are the possibility to adjust for a number of potential confounders and the fact that information about smoking habits during pregnancy was prospectively collected. Given the prospective nature of the study, the number of deaths is small: however, if the association between maternal smoking and SIDS found in this and previous studies reflects a causal relation, approximately 30–40% of all cases of SIDS could be avoided if all pregnant women stopped smoking in a population with 30% pregnant smokers.