Material and Methods

This was a retrospective cohort study of all women with a diagnosis of d-TGA who underwent an atrial switch procedure from 1970-1989 and subsequently were evaluated in our pediatric cardiology or adult congenital cardiology clinics during the 10-year period from 1999-2009. All patients were of childbearing age, which was defined as >18 years old, for the purposes of this study. Patients who died or underwent cardiac transplantation before reaching reproductive age were excluded from further analysis.

This study was deemed exempt by the University of Utah Institutional Review Board. Maternal obstetric, cardiac, and neonatal electronic medical records were reviewed to abstract demographic data. Recorded pregnancy variables included current age, age at each attempted conception, presence of infertility (defined as unsuccessful conception after 1 year of unprotected sexual intercourse or use of fertility treatments to conceive), age at each pregnancy, number of medically documented miscarriages, and indication for delivery. Presence or absence of complication after the pregnancy was noted: pregnancy-induced hypertension (new onset hypertension at ≥20 weeks of gestation, >140 mm Hg systolic, or 90 mm Hg diastolic without proteinuria), preeclampsia (gestational hypertension with >0.3 g of proteinuria/24-hour urine sample), eclampsia, HELLP syndrome (hemolysis, elevated liver enzymes, low platelets), and gestational diabetes mellitus. Other recorded obstetric complications included operative (forceps/vacuum/caesarean) delivery, preterm premature rupture of membranes, and premature labor (spontaneous onset of labor at <37 weeks of gestation). Cardiac data that included the type of cardiac surgery (Mustard or Senning), and the following baseline and peripartum variables were recorded: presence or absence of arrhythmias, pacemaker, need for cardiac medications, and New York Heart Association functional class. The following neonatal complications were documented: premature birth (delivery at <37 weeks of gestation), fetal death (intrauterine death at ≥20 weeks of gestation), neonatal death (within the first month after birth), and/or recurrence of congenital heart disease.

All women in the cohort who achieved pregnancy were observed prospectively throughout the pregnancy and the postpartum period by perinatologists and cardiologists. Most of the women had undergone annual echocardiograms as part of routine follow-up evaluation before conception; those women who had experienced a pregnancy had at least 1 additional echocardiogram performed during the pregnancy, and/or after delivery. All echocardiograms in our system are stored electronically and were reviewed by a single cardiologist for qualitative and quantitative measures of right ventricular function. Because an ejection fraction is not considered to be a reliable measure of systemic right ventricular function, the change in ventricular pressure over time (dP/dt) was used to quantify right ventricular function ; dP/dt is an indirect measure of the force of ventricular contraction that requires the presence of some degree of tricuspid regurgitation to be assessed. Additional recorded echocardiographic parameters included the degree of tricuspid insufficiency that was judged qualitatively as absent, mild, moderate, or severe and evidence of atrial baffle obstruction.

With the exception of the analysis of quantitative cardiac function (dP/dt), the Student t test was used for comparison of continuous variables. Fisher’s exact test was used for comparison of dichotomous variables. To compare dP/dt values over time, a mixed-effects linear regression model was used to account for multiple pregnancies within a single study subject. In this mixed-effects model, the outcome variable was quantitative cardiac function (dP/dt); the predictor variable was an indicator variable for time period (during pregnancy, relative to prepregnancy, in 1 model and after pregnancy, relative to prepregnancy, in a separate model). First, a mixed-effects model with time periods nested within the pregnancy and the pregnancy nested within the study subject was fitted. With the use of a likelihood ratio test, it was discovered that multiple pregnancies that were nested within a subject contributed nothing to the model fit. Therefore, the subject level was dropped, and the models were reduced to time periods that were nested within pregnancy. Statistical analyses were performed in STATA software (version 11.0; StataCorp, College Station, TX); statistical significance was defined as a 2-sided probability value of < .05.