Advertisement

Caffeine and miscarriage: case closed?

      To the Editors:
      Weng et al report in a highly publicized article that caffeine intake increases the risk of miscarriage.
      • Weng X.
      • Odouli R.
      • Li D.K.
      Maternal caffeine consumption during pregnancy and the risk of miscarriage: a prospective cohort study.
      They note that a caffeine-miscarriage link has been debatable because of the methodologic limitations of prior studies, including retrospective data collection and recall bias, small sample sizes, and confounding by miscarriage risk factors and pregnancy symptoms. It is unclear to us, however, that Weng et al were able to overcome these same problems. Although the investigation is described as a prospective cohort study, the majority of the miscarriage cases were initially contacted and interviewed about caffeine intake after the miscarriage had occurred. In an article also published this month, Savitz et al demonstrated that this can result in recall bias generating a positive result, whereas when caffeine exposure is ascertained before miscarriage, the findings indicate no effect of caffeine.
      • Savitz D.A.
      • Chan R.L.
      • Herring A.H.
      • Howards P.P.
      • Hartmann K.E.
      Caffeine and miscarriage risk.
      Weng et al also state that these cases reported on intake “up to the end of the pregnancy,” which would have included time after fetal demise (but before clinical recognition), during which time an abating of pregnancy symptoms and a rebounding caffeine intake can occur.
      • Signorello L.B.
      • McLaughlin J.K.
      Maternal caffeine consumption and spontaneous abortion: a review of the epidemiologic evidence.
      As would be expected, most of the participants in the Weng et al study (79%) reduced their caffeine consumption during pregnancy (presumably reflected in the single average-daily-intake variable used in their analysis). This reduction would reflect not only physician recommendations but pregnancy symptoms (nausea, vomiting, coffee aversion) that are associated with fetal viability. The authors argue that by “direct control” of intake changes related to fetal viability (ie, by stratifying on change patterns) they have demonstrated a true effect of caffeine, because it persisted across the groups. However, within this largest subset of women who reduced intake, the hazards ratio associated with 200+ mg/day fell from 2.23 (95% CI, 1.34-3.69) to a nonsignificant 1.47 (95% CI, 0.87-2.51), suggesting that confounding likely accounts for the original result.
      Confounding by factors other than pregnancy symptoms is also a concern. Comparing women who consumed 200+ mg/day to 0 mg/day, the former group was more likely to be >35 years of age, to have had previous miscarriages, to smoke cigarettes and drink alcohol during pregnancy, and to lack pregnancy symptoms. Incomplete adjustment for any of these variables could spuriously produce a positive result (eg, the miscarriage rate for women over age 35 is several times higher than for women in their twenties,
      • Heffner L.J.
      Advanced maternal age–how old is too old?.
      and women who consumed 200+ mg/day were almost twice as likely to be in this older age group).
      Finally, the authors state that “…the increased risk of miscarriage appeared to be due to caffeine itself rather than other possible chemicals in coffee because caffeine intake from noncoffee sources showed the similarly increased risk of miscarriage.” However, this claim is based on 8 women who consumed 200+ mg/day from noncoffee sources, 2 of whom had a miscarriage.
      The data used by Weng et al were derived from a study designed to investigate the potential effect of magnetic fields, not caffeine, on miscarriage risk,
      • Li D.K.
      • Odouli R.
      • Wi S.
      • et al.
      A population-based prospective cohort study of personal exposure to magnetic fields during pregnancy and the risk of miscarriage.
      and the analysis does little to improve upon the methods of previous investigations, some of which gave more focused attention to the collection and timing of caffeine intake and pregnancy symptom data. We agree that this issue warrants serious scientific attention, but the present study fails to provide the methodologic rigor needed to answer the question.

      References

        • Weng X.
        • Odouli R.
        • Li D.K.
        Maternal caffeine consumption during pregnancy and the risk of miscarriage: a prospective cohort study.
        Am J Obstet Gynecol. 2008; 198: 279.e1-279.e8
        • Savitz D.A.
        • Chan R.L.
        • Herring A.H.
        • Howards P.P.
        • Hartmann K.E.
        Caffeine and miscarriage risk.
        Epidemiology. 2008; 19: 55-62
        • Signorello L.B.
        • McLaughlin J.K.
        Maternal caffeine consumption and spontaneous abortion: a review of the epidemiologic evidence.
        Epidemiology. 2004; 15: 229-239
        • Heffner L.J.
        Advanced maternal age–how old is too old?.
        N Engl J Med. 2004; 351: 1927-1929
        • Li D.K.
        • Odouli R.
        • Wi S.
        • et al.
        A population-based prospective cohort study of personal exposure to magnetic fields during pregnancy and the risk of miscarriage.
        Epidemiology. 2002; 13: 9-20

      Linked Article

      • Maternal caffeine consumption during pregnancy and the risk of miscarriage: a prospective cohort study
        American Journal of Obstetrics & GynecologyVol. 198Issue 3
        • Preview
          The objective of the study was to examine whether the risk of miscarriage is associated with caffeine consumption during pregnancy after controlling for pregnancy-related symptoms.
        • Full-Text
        • PDF
      • Reply
        American Journal of Obstetrics & GynecologyVol. 199Issue 5
        • Preview
          With regard to potential “recall bias”: had we only included the women who were interviewed before their miscarriage, the association between caffeine intake during pregnancy and the risk of miscarriage would have even been stronger: adjusted hazard ratio (aHR) = 2.78, 95% confidence interval (CI), 1.11-6.96 for daily caffeine intake 200 mg or more compared with aHR = 2.23 (95% CI, 1.34-3.69) when all miscarriages were included. Therefore, our results provided no evidence for the speculated recall bias.
        • Full-Text
        • PDF