Bupropion and specific cardiovascular malformations

Trial description: The study is an extension of earlier work based on a retrospective epidemiologic study of infants born to women who were exposed to bupropion in their estimated first trimester of pregnancy using data from a large US health plan affiliated with i3 Drug Safety (Clinical study ID WWE113694) (Cole JA, Oh KS, Chiang CC, Walker AM, Haight BR, Modell JG. Bupropion in pregnancy and the prevalence of congenital malformations Pharmacoepidemiology and Drug Safety, 2007; 16: 474-484). The cohorts developed for the earlier work consisted of all infants born to women exposed to bupropion during the estimated first trimester and outside the first trimester, and a random sample of infants born to women exposed to other antidepressants during the first trimester between 01 January 1995 and 30 September 2004. The objectives for this study include refining of both the original first trimester bupropion cohort and the original bupropion outside the first trimester cohort into mono-therapy and mono- or poly-therapy. Exposure to other antidepressants during the first trimester will also be refined into mono-therapy and mono- or poly-therapy. With input from pediatric cardiology expert, lists of specific cardiovascular malformations and malformation groupings will be created. The groupings will be created among the refined first trimester bupropion cohort as well as in two comparison cohorts of bupropion outside the first trimester and first trimester antidepressant use (mono-therapy and mono-or poly-therapy). The prevalence in each cohort will be calculated as the number of infants with a specific cardiovascular malformation divided by the number of live born infants. Prevalence will be reported per 1,000 infants. Confidence intervals will be calculated using Wilson’s approximation to exact binomial intervals when the number of cases is five or greater and exact binomial intervals when the number of cases is fewer than five. The appropriateness of further calculations will be evaluated. Where numbers permit, adjusted odds ratios for specific cardiovascular groups/malformations will be calculated and if appropriate, stratified according to maternal dispensing of medications suspected to be teratogenic. The following comparisons, if numbers permit, will be performed: 1) bupropion first trimester mono-therapy cohort versus other antidepressant first trimester mono-therapy cohort; 2) bupropion first trimester mono- or poly-therapy cohort versus other antidepressant first trimester mono- or poly-therapy cohort; 3) bupropion first trimester mono-therapy cohort versus bupropion outside of first trimester mono-therapy cohort, and 4) bupropion first trimester mono- or poly-therapy cohort versus bupropion outside of first trimester mono- or poly-therapy cohort. Adjusted odds ratios will be calculated through a generalized estimated equations form of multivariate logistic regression to account for births associated with multiple infants. The same covariates identified in the original study will be included in this re-analysis. Covariates included: diagnoses of bipolar disorder and eclampsia within one year before delivery; dispensings of lithium, phenytoin, and fluconazole within one year before delivery through the end of the first trimester; and the number of physician visits within 10 to 12 months before delivery, maternal age, geographic region of the health plan, and infant gender. If generalized estimating equation form of the logistic regression model does not converge, adjusted odds ratios will be presented from a conventional multivariate logistic model. If the conventional multivariate logistic model does not converge, only the crude odds ratio will be presented.