Question

Pregnant women were recruited during their first trimester to study the relationship between chocolate consumption during pregnancy and risk of preeclampsia (pregnancy-induced hypertension). They were interviewed about chocolate consumption, age, race, education, smoking, body mass index, and previous pregnancy history. Participants were re-interviewed after delivery to obtain information on chocolate consumption during the third trimester of pregnancy. In addition, umbilical cord blood was collected at the time of delivery and measured for concentration of theobromine (the major metabolite of chocolate). Obstetrics records were obtained to determine if preeclampsia developed during the pregnancy. Of 1995 eligible women, 348 were excluded due to preexisting hypertension or preeclampsia at the first study visit, leaving a sample size of 1647 women (Epidemiology 2008;19:459-464).

(a) What study design is this?

(b) Why were women with preexisting hypertension excluded from the study?

(c) What is the primary independent variable (i.e., exposure or predictor variable) in the study?

(d) What is the dependent (outcome) variable in the study?

(e) Chocolate consumption was higher among women who smoked during pregnancy, so smoking is a potential confounder. What else needs to be true for smoking to be a confounder in the relationship between chocolate consumption and preeclampsia?

(f) Logistic regression was used to adjust for possible confounding.

(f-i) What is one advantage of using logistic regression to adjust for confounding?

(f-ii) What is one disadvantage of using logistic regression to adjust for confounding?

(g) Results for the association between umbilical cord theobromine level and preeclampsia were as follows:

Theobromine (ng/mL)	Unadjusted OR (95% CI)	Adjusted OR (95% CI)*
0-155	1.0	1.0
>155-400	0.46 (0.22-097)	0.49 (0.21-1.15)
>400-900	0.32 (0.14-0.73)	0.35 (0.13-0.90)
>900	0.28 (0.12-0.65)	0.31 (0.11-0.87)
* adjusted for age, race, education, smoking, and BMI

(g-i) Why are confidence intervals not reported for the 0-155 ng/mL category?

(g-ii) Interpret the unadjusted odds ratio (OR) for >900 ng/mL.

(g-iii) Interpret the 95% confidence interval (CI) for the unadjusted OR for >900 ng/mL.

(g-iv) Does it appear that age, race, education, smoking, and/or BMI were confounders in the relationship between umbilical cord theobromine level and preeclampsia? Why or why not?

(h) Only women who delivered a live birth were included in the analysis. A concern is that exclusion of women who miscarried could cause bias in the study findings.

(h-i) Would this be a type of information bias or selection bias? Explain.

(h-ii) 8% of women enrolled in the study miscarried. Is it likely that the associations observed above are due to bias from excluding these women? Explain.

(i) The authors were concerned that study findings could be due to residual confounding by smoking, so they conducted a sensitivity analysis by repeating the analyses using only nonsmoking women.

(i-i) What is residual confounding?

(i-ii) What is a sensitivity analysis?

(i-iii) They found no change in the results when they restricted the analysis to only nonsmoking women. Does it appear that residual confounding due to smoking was a problem? Why or why not?

(j) Based on this study, what do you conclude about the association between chocolate consumption during pregnancy and preeclampsia?

Answer 1