Maternal intake of coffee during the first 3 months of the pregnancy was associated in a dose-response manner with risk of delivering an infant with CLP. In contrast, we found no evidence of an association between coffee intake and risk of CPO. The association between coffee and CLP was only slightly decreased by adjustments for confounders, and it persisted for nonsmokers and the subset of infants with isolated CLP.
There is no known mechanism by which coffee intake might increase the risk of CLP. Effects on homocysteine is one potential pathway. Evidence suggests that maternal hyperhomocysteinemia may be linked to increased risk of CLP (13
). Coffee intake increases the plasma concentration of homocysteine (14
), as does smoking (15
), a well-established risk factor for CLP (17
). Folic acid supplements are consistently associated with reduced risk of CLP (18
), and such supplements reduce plasma homocysteine (15
). In the present study, the association between high coffee intake and CLP was slightly stronger for women who did not take a folic acid supplement (crude odds ratio = 1.99, 95% CI: 1.30, 3.04), although this interaction did not approach statistical significance (P
for interaction = 0.89).
Consumption of caffeine-containing tea was strongly associated with a reduced risk of both CLP and CPO. Potential health benefits of tea have been linked to its high content of antioxidants (19
). This is not a likely explanation for the reduced cleft risk, however, because coffee has an even higher total content of antioxidants than tea does (20
). The association of tea with lower risk of both types of clefts may reflect the presence of unmeasured confounding factors. Tea drinkers may have more healthy habits; consumers of high quantities of tea in our study were older, more educated, and drank less alcohol per sitting compared with nonconsumers. Furthermore, the diet of consumers of high quantities of tea contained more of every macronutrient and almost every micronutrient compared with the diet of nonconsumers (data not shown).
The association of coffee with CLP appears to be independent of the role of caffeine. We found no association between total caffeine intake and risk of CLP (reflecting the combination of the positive association with coffee and the negative association with tea). Only 5 women drank decaffeinated coffee exclusively, too few for a separate analysis. When we combined data on decaffeinated coffee with those on caffeinated coffee, the odds ratios for CLP were slightly increased, consistent with the association of coffee and CLP being unrelated to caffeine intake (data not shown).
By the same token, high-quantity coffee consumers (>3 cups/day) were older, less educated, smoked more, and drank more alcohol per sitting compared with nonconsumers. Even though we adjusted for possible confounding factors, we cannot rule out the possibility of residual confounding by known factors such as smoking or confounding by unmeasured factors. Although we did not control for diet, there was no evidence of a poorer diet among consumers of high quantities of coffee, who actually reported eating more vegetables, less sugar, and more of many essential nutrients (data not shown).
Another potential source of bias in this study comes from possible selective participation of controls. While participation was relatively high (88% for cases and 76% for controls), the lower participation of controls leaves room for differential participation by coffee drinking. However, it is unlikely that such selection would occur for one type of facial cleft and not the other. Coffee was strongly associated with CLP but not with CPO in our data. This specificity has been found for other factors associated with facial clefts (6
). By the same argument, the lack of specificity for the tea association (which was protective for both types of facial clefts) raises the question of whether this association might be due to some unmeasured bias.
Women may reduce their coffee consumption as they develop pregnancy symptoms of nausea and vomiting. Given that closure of the lip and palate occurs relatively late in the first trimester (8–12 weeks after the last menstrual period) (1
), it is possible that reported coffee intake is greater than actual consumption at the crucial stage of embryonic development. This bias would tend to weaken the observed association.
Our study had the advantage of enrolling mothers soon (an average of 15 weeks) after delivery. In addition, questions on coffee and other caffeinated beverages constituted only a small part of the questionnaire, with no specific emphasis on these beverages that would be expected to bias reporting. Another strength of the study was collection of extensive data on potential confounders such as folic acid supplement use, alcohol consumption, and smoking. Confounders were adjusted for as continuous variables in our analyses, but adjustment using categorized variables did not alter the results.
A meta-analysis of 3 studies on maternal coffee consumption and orofacial clefts found a slight increase in the risk of clefts (11
). All 3 studies used 0 cups as the reference category. When high coffee intake was compared with low intake, the pooled odds ratio was 1.2 (95% CI: 0.9, 1.6). One of the studies provided results for CLP separately, with an adjusted odds ratio for CLP of 1.3 (95% CI: 0.9, 1.9) associated with more than 3 cups of coffee a day (11
). The 2 other studies combined both types of clefts in their analyses. In the report by Kurppa et al. (21
), drinking more than 4 cups of coffee a day had no association with risk of clefts (unadjusted odds ratio = 1.0, 95% CI: 0.6, 1.6), whereas, according to McDonald et al. (22
), more than 3 cups of coffee daily resulted in an adjusted odds ratio of 1.4 (95% CI: 0.7, 2.7).
A recent cohort study from Denmark found that coffee intake was associated with a reduced risk of CLP (23
). The authors observed an odds ratio of 0.66 (95% CI: 0.27, 1.62) when comparing daily intake of more than 5 cups of coffee with no intake. The Danish study had the advantage of a prospective design, although one cost of this design was a relatively small number of cases (134 with CLP).
In summary, results from our study showed a dose-dependent association between coffee consumption during the first trimester and increased risk of CLP. This association was specific to CLP, with no association found between coffee consumption and risk of CPO. There was little or no evidence for an association between caffeine from other types of beverages and CLP. Even with extensive adjustments for confounders, we cannot eliminate the possibility that the associations between coffee consumption and risk of CLP are due to uncontrolled confounding by factors associated with the habit of drinking coffee. Still, considering our results and the prior mixed evidence for a coffee effect on clefts, women cannot be assured that maternal coffee consumption is entirely benign for the developing fetus. Other clefts studies now in progress should give close attention to a possible association of coffee consumption with facial clefts.