Materials and Methods
Study setting and cohort
This retrospective cohort study included parturients with CHD who gave birth via caesarean section from 1994 to 2019 and whose information was in a database maintained by the Department of Obstetrics and Gynaecology at the National Cerebral and Cardiovascular Centre, in Suita, Osaka, Japan. Parturients with simple congenital arrhythmias or cardiomyopathy were excluded. The institutional ethics committee waived the need for patient consent and approved the study before medical record review began (R19094).
Outcomes
Cardiovascular events were defined based on the existing literature.14-17 The primary outcome was a postpartum cardiovascular event consisting of a composite of all-cause death, new onset or worsening heart failure, pulmonary hypertension, sustained or non-sustained arrhythmia, and thromboembolic complications (myocardial infarction, mechanical valve thrombosis, venous thromboembolism, or cerebrovascular event) that required treatment with intravenous or oral medication or electrical defibrillation. Heart failure that required treatment with only oral diuretics and cardiovascular events that occurred during pregnancy and continued into the postpartum period were not considered primary outcomes.
Secondary outcomes consisted of maternal respiratory events, neonatal all-cause death, neonatal intubation, Apgar score, and umbilical arterial pH. Low Apgar score was defined as Apgar score < 7 at 1 minute or 5 minutes after delivery.18 When a patient was multiple pregnancy, only data on the first newborn was analysed. Composite neonatal outcomes included neonatal intubation and low Apgar score. Primary or secondary outcomes could include events that occurred up to 1 week after caesarean section.
Exposure variables
Anaesthetic technique was analysed as general or neuraxial anaesthesia. We categorized spinal, epidural, or the combination of spinal and epidural anaesthesia as neuraxial anaesthesia and general anaesthesia or the combination of general and epidural anaesthesia as general anaesthesia. Parturients who converted from neuraxial to general anaesthesia were excluded because the aim of the study was to clarify the association between these 2 anaesthetic techniques and outcomes.
Potential confounders
Maternal demographic characteristics included age, multiparity, multiple pregnancy, and body mass index (BMI). BMI was divided into 4 categories based on the existing literature.19 Obstetrics characteristics included multiparity, date of caesarean delivery in periods of approximately 5 years to adjust for technical advances in caesarean sections (1994–1999, 2000–2004, 2005–2009, 2010–2014, and 2015–2019), gestational age at delivery categorized into 3 groups based on existing literature: age <31 years, age of 32–36 years, and age ≥37 years to adjust for the risk of older age,20 type of indication for caesarean section (maternal cardiovascular indication or obstetric indication),21 and emergency caesarean delivery status.
Maternal cardiac demographics included New York Heart Association (NYHA) classification before pregnancy and modified World Health Organization (mWHO) classification for maternal cardiovascular risk.22 Two certified experts, an adult cardiologist specializing in caring for obstetric patients and a paediatric cardiologist at our institution, assessed mWHO class for maternal cardiovascular risk, which is the best available assessment model for estimating cardiovascular risk in pregnant women with CHD.23
Systemic ventricular ejection fraction (EF) values from the most recent echocardiographic examination before caesarean section were collected. Cardiovascular events during pregnancy included any of the following: heart failure, pulmonary hypertension, sustained or non-sustained arrhythmia, and thromboembolic complications (myocardial infarction, mechanical valve thrombosis, venous thromboembolism, or cerebrovascular event) that required treatment with intravenous or oral medication, electrical defibrillation, or caesarean delivery.
Statistical Analysis
Stata version 15 (College Station, TX, USA) was used for analysis. For baseline patient characteristics, continuous variables are presented as means with standard deviation for normally distributed variables and were compared using the t-test. Non-normally distributed variables are presented as medians (interquartile range) and were compared using the Mann-Whitney U test. Categorical baseline variables were compared using Fisher’s exact test.
To account for the heterogeneity of CHD, the association between anaesthetic technique and composite postpartum cardiovascular events was analysed using generalized linear mixed models assuming a binomial distribution (i.e., mixed-effects logistic regression) with a random intercept for each mWHO class for maternal cardiovascular risk. Confounding factors included cardiovascular events during pregnancy, gestational age at delivery, and emergency caesarean section status. The association between anaesthetic technique and composite neonatal outcomes was also analysed using mixed-effects logistic regression with a random intercept for each mWHO classification for maternal cardiovascular risk. Confounding factors included cardiovascular events during pregnancy, gestational age at delivery, birth weight, and emergency caesarean section status. Confounding factors considered to be relevant to primary or secondary outcomes were selected based on existing clinical knowledge. Only parturients with complete information available were included in the analysis.
We also performed sensitivity analyses. First, we added date of caesarean delivery as a confounding factor for both composite maternal and neonatal outcomes. Second, we replaced emergency caesarean status with gestational weeks at delivery or type of indication for caesarean section as potential confounding factors for composite maternal or neonatal outcomes, respectively. Third, we applied inverse probability of treatment weighting (IPTW) to analyse whether anaesthetic technique is associated with composite postpartum cardiovascular events. We used multivariate logistic regression to estimate propensity scores using the following factors: age, BMI, multiparity, multiparous, cardiac events during pregnancy, gestational weeks at delivery, emergency caesarean section status, systemic ventricular EF, date of caesarean delivery, NYHA class before pregnancy and mWHO class for maternal cardiovascular disease. To reduce bias and the amount of missing data, 2 authors abstracted data from electronic medical records from 2008 to 2019 and paper charts from 1994 to 2007.