Material

Between 1 September 1992 and 31 August 1995, parents of 294 cases of SIDS in the Scandinavian countries were invited by the local paediatrician (Norway and Sweden) or forensic institute (Denmark) to participate in the study. Of these, 244 families (83%) accepted. For each case, six controls matched for sex, date of birth plus two weeks, and maternity hospital were selected. The delay period of two weeks was allowed to obtain a similar age of the controls and the SIDS victims when the questionnaire was completed. The first four were invited to participate in the study. If a family was reluctant to enter the study, one of the two remaining controls was invited. Of a total of 1207 invited control families, 869 (72%) were willing to participate. The cases and controls received a questionnaire consisting of 272 questions without any written guidance included. The material is described in detail elsewhere.10

Results

CAFFEINE

Mean caffeine intake during pregnancy was higher among mothers of cases than among mothers of controls (345 v218 mg/24 hours; 95% CI of the difference, −167 to −87; p < 0.001), as was caffeine intake after pregnancy (375v 263 mg/24 hours; 95% CI of the difference −153 to −71; p < 0.001).

When analysing caffeine in four strata (table 1), the crude OR for caffeine consumption during pregnancy was increased for doses > 400 mg/24 hours. For doses between 401 and 800 mg/24 hours, the OR was 2.0 (95% CI, 1.2 to 3.4) and for doses > 800 mg/24 hours, the OR was 3.9 (95% CI, 1.9 to 8.1). After pregnancy, the crude OR was significantly increased only in the highest stratum; that is, above 800 mg/24 hours, where it was 3.1 (95% CI, 1.5 to 6.3) (table 1). After adjustment for smoking in 1st trimester, no significant effects remained during or after pregnancy, and after adjustment for maternal age, education, parity, and smoking in 1st trimester we found an effect of caffeine > 400 mg/24 hours during pregnancy, but it was not significant even in the highest stratum. After pregnancy we found an OR of 1.4 (95% CI, 0.6 to 3.1) in the highest stratum after adjusting for smoking, and 2.0 (95% CI, 0.8 to 5.2) after adjusting for smoking, maternal age, education, and parity.

During pregnancy, the difference in χ² between a linear and a categorised model was 9.2 (2 df; p < 0.02) for caffeine; after pregnancy the difference was 14.0 (2 df; p < 0.001).

When comparing caffeine ⩽ 400 mg/24 hours with caffeine > 400 mg/24 hours (table 2), the crude OR for caffeine > 400 mg/24 hours was 2.5 (95% CI, 1.8 to 3.6) during pregnancy, and 2.0 (95% CI, 1.5 to 2.9) after pregnancy. Adjustment for smoking lowered ORs considerably (OR, 1.5; 95% CI, 1.0 to 2.2 and OR, 1.3; 95% CI, 0.9 to 1.9 during and after pregnancy, respectively). When adjusting for smoking, maternal age, education, and parity the OR was 1.6 (95% CI, 1.0 to 2.5) during pregnancy and 1.4 (95% CI, 0.9 to 2.1) after pregnancy.

There was a strong relation between caffeine consumption and smoking (table 3). Among the controls, we found that only 1% of the non-smokers consumed > 800 mg caffeine each day, whereas the corresponding figure was 5% of those who smoked 1–4 cigarettes each day and 14% of those who smoked 5–9 cigarettes each day (p < 0.001).

INTERACTIONS

We found no significant interactions with country, maternal smoking, maternal age, education, or parity. However, there was a significant interaction between breast feeding and caffeine in pregnancy (p = 0.027), but not with caffeine after pregnancy (p = 0.06).

ALCOHOL

If the mother ingested alcohol on the reference day or the day before, the risk was increased with an alcohol dose of 5 AU or more (crude OR, 12.2; 95% CI, 2.5 to 59.3) (table 4). After adjusting for paternal unemployment, maternal smoking, age, and education, the effect was still significant (OR, 5.9; 95% CI, 1.0 to 33.9; p = 0.047). However, there was no influence of either prenatal or postnatal maternal ingestion of alcohol on the risk of SIDS.

Paternal use of alcohol in moderate doses was associated with low ORs (tables 5 and 6). Crude ORs were 0.6 (95% CI, 0.4 to 0.9) for use of alcohol less than once a week and 0.5 (95% CI, 0.3 to 0.9) for use of alcohol more than once a week during pregnancy. After pregnancy, the crude ORs were 0.5 (95% CI, 0.3 to 0.7) and 0.4 (95% CI, 0.3 to 0.7), respectively. When adjusting for paternal unemployment, smoking, maternal age, and education, the ORs were 0.8 (95% CI, 0.5 to 1.3) for less than weekly and 0.7 (95% CI, 0.4 to 1.4) for more than weekly use during pregnancy, and 0.7 (95% CI, 0.4 to 1.1) for less than weekly and 0.6 (95% CI, 0.3 to 1.1) for more than weekly use after pregnancy.

Discussion

We could not demonstrate an independent effect of prenatal caffeine on the risk of SIDS. Alcohol consumption by the mother during pregnancy did not increase the risk, but large amounts of alcohol on the reference day or the day before increased the risk. Paternal alcohol consumption did not result in an increased risk; in fact, the ORs were low because of an association with beneficial social circumstances, which is supported by the non-significant results after adjusting for social variables.

CAFFEINE

Caffeine has been found to cause malformations in rodents16-19 and to induce long term effects on sleep, locomotion, learning abilities, emotivity, and anxiety in rodent offspring.18 In humans, teratogenic effects are questionable.20 Caffeine does not change breast milk composition, and stimulates milk production.21 Several studies have suggested an increased risk of spontaneous abortion or intrauterine growth retardation.22-24 However, after adjustment for smoking15 25 this effect seems to be low. Recent studies seem to agree that there is no risk of preterm delivery,26-29 but the effect on birth weight is still disputed.26 29-31

Thus, several studies have been published on the association between caffeine and pregnancy outcome, but to our knowledge, only one on SIDS.3 In our study, however, we could not demonstrate such an effect. The OR estimate in the highest dose range (> 800 mg/24 hours) was 2.5 during and 2.0 after pregnancy, which was not significant (table 1), and the same applies to the dichotomised variable in table 2. Unfortunately, we do not have sufficient data to verify an effect, regardless of the model chosen. Also, with the results from the difference in χ² between a categoric and a linear model, we could not verify a dose response relation. The results might be caused by misclassification—mothers might be more honest when answering questions concerning coffee and tea consumption than questions regarding smoking, especially as it is well known that smoking is harmful to the baby. Alternatively, it could be that we did not have enough cases to verify an effect. Because smoking and coffee consumption are strongly correlated (table 3), unreported smoking would tend to increase the ORs for caffeine consumption.

The interaction between breast feeding and caffeine is presumably the result of confounding by smoking. It has been shown that mothers who smoke breast feed less than those who do not.32 Among our controls, 30% of smokers were breast feeding at the interview, in contrast to 70% of the non-smokers. We found that in a model containing smoking in early pregnancy, the interaction term loses its significance.

Because it has been suggested that caffeine could stimulate milk production,21 one might expect an association between caffeine intake and breast feeding, and thus a protective effect of caffeine, because breast feeding seems to lower the OR. Our data showed no such association.

ALCOHOL

In view of the harmful effects of alcohol on the fetus, we found it remarkable that there was no increase in SIDS with prenatal alcohol use. In a case control study, it is possible that mothers with alcohol problems would not admit them, and the consequence would be an underestimation of its effects. Thus, the increased risk attached to high intake of alcohol by the mother within the last days before the event was surprising and might suggest that the lack of effects of prenatal use is true. However, this result is based on only seven cases and two controls. The small numbers render analyses of interactions impossible.

The seemingly protective effect of paternal alcohol ingestion reflects an association with beneficial social circumstances, and thus is a confounder. This is supported by the disappearance of significance after adjusting for maternal age and education, maternal smoking during pregnancy, and paternal unemployment. It has been shown recently that the pattern of alcohol use differs between social classes, and that men in Sweden with a higher socioeconomic index consume more alcohol than men with a lower socioeconomic index (Rosengren A, Dotevall A, Wilhelmsen L, the MONICA study of Göteburg. Unpublished data, 1999). Thus, a moderately high alcohol consumption could be a marker of high social status.

Conclusions

We found no association between caffeine consumption, during or after pregnancy, and increased risk of SIDS. Prenatal alcohol consumption by the mother was not associated with an increased risk, but heavy postnatal intake by the mother might be a risk factor in a very small subgroup. Paternal alcohol consumption prenatally or postnatally is associated with a lower OR, but this is a result of confounding.