Peripartum outcomes and anaesthetic management of parturients with moderate to complex congenital heart disease or pulmonary hypertension*


Correspondence to B. G. Maxwell



We performed a retrospective cohort analysis of pregnancies among women with moderate to complex congenital heart disease or pulmonary hypertension over a 12-year period, resulting in a cohort of 107 cases in 65 women. Neuraxial analgesia or anaesthesia was provided in 84%, 89% and 95% of spontaneous vaginal, operative vaginal and caesarean deliveries, respectively. The caesarean delivery rate was 43% compared to our institution average of 27% over the same period (p = 0.02), and 38% had operative vaginal deliveries compared to a 10.5% institution rate (p < 0.01). Invasive monitoring was used in 28% of all deliveries. There were one maternal and two neonatal deaths. This study provides detailed anaesthetic and peripartum management of women with congenital heart disease, a patient population in whom evidence-based practice and data are largely lacking. We observed a predominance of neuraxial anaesthetic techniques, increased caesarean and operative delivery rates, and favourable maternal and neonatal outcomes. Multicentre studies and registries to compare anaesthetic and obstetric management strategies further and delineate risk factors for adverse outcomes are required.

Cardiac disease is a leading cause of maternal mortality in the USA and UK [1, 2]. Peripartum mortality and morbidity due to cardiac disease have been increasing over the past few decades [1-3]. Contributing to this mortality and morbidity are women with congenital heart disease (CHD). Improvements over the past several decades in the surgical and medical treatment of CHD have allowed increasing numbers of women with CHD to survive well into their reproductive years [4, 5]. An increasing number of women with pulmonary hypertension, either primary or secondary, are also becoming pregnant and carrying a pregnancy to completion [6]. Patients with CHD and pulmonary hypertension present complex obstetric and anaesthetic management challenges.

Several studies including a systematic review [7], a prospective cohort study [8], analyses of national administrative databases [9, 10], and a collection of case series [11-18] have examined obstetric outcomes in obstetric patients with congenital cardiac disease. However, there is limited information concerning the anaesthetic management of these patients and the impact of anaesthesia on peripartum maternal and neonatal outcomes. Information regarding anaesthetic management is generally limited to classification of general vs neuraxial techniques [19-21].

The aim of this study was to examine anaesthetic management and peripartum outcomes in parturients with CHD or pulmonary hypertension presenting at a tertiary academic centre over the past decade. We sought to examine in detail the anaesthetic and obstetric management of these patients compared with our institution's standard management and to evaluate maternal and neonatal outcomes.


After obtaining Stanford University Institutional Review Board approval, we performed a retrospective cohort analysis of all pregnancies occurring in women with a diagnosis of moderate to complex congenital cardiac disease during the period from 1 January 2000 to 31 December 2011 at Lucile Packard Children's Hospital (LPCH), is a tertiary academic referral centre with an average delivery rate of 5030 per year, a caesarean delivery rate of 26.5%, and a labour epidural rate of 83% over the same 12-year study period.

Cases were initially identified through a combination of International Statistical Classifications of Diseases diagnostic codes for congenital cardiac disease (codes 745.x, 746.x, 747.1, 747.3, 747.4, 648.5 and 416.0) and Current Procedural Terminology codes for obstetric and obstetric anaesthesia procedures (01960-9, 59400, 59409, 59410, 59412, 59414, 59425, 59426, 59430, 59610, 59612, 59614, 59618, 59620 and 59622). These cases were cross-referenced with a high-risk obstetric anaesthesia clinic database and an adult congenital cardiology database, then reviewed individually to exclude cases that did not meet inclusion criteria and to avoid double-counting of cases. The clinical database techniques used in this project were supported by the Stanford NIH/NCRR CTSA award number UL1 RR025744. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center for Research Resources or the National Institutes of Health [22].

Criteria for inclusion in the retrospective cohort were documentation of a viable pregnancy in a patient with a prior diagnosis of a congenital cardiac lesion of moderate to severe complexity by existing severity classification guidelines [23], or a diagnosis of moderate (mean pulmonary artery pressure > 40 mmHg) or severe (mean pulmonary artery pressure > 55 mmHg) pulmonary arterial hypertension. Patients with a simple congenital lesion without sequelae, patients with isolated mild pulmonary hypertension, and patients with purely acquired cardiac disease (e.g. ischaemic heart disease, rheumatic valvular disease) were not studied.

We used an institution-specific electronic clinical database, Stanford Translational Research Integrated Database Environment (STRIDE), that aggregates both paper-based and electronic medical records for all patients cared for at LPCH. Data were collected on pregnancy course, labour and delivery outcomes, anaesthetic management during provision of labour analgesia or anaesthesia for caesarean delivery, postpartum maternal outcomes within the same hospitalisation, and neonatal outcomes within 30 days of birth.

Abortion (spontaneous or therapeutic) was defined as pregnancies that lasted less than 20 weeks. Preterm delivery was defined as occurring between 20 weeks' and 36 weeks and 6 days' gestation, in accordance with American College of Obstetricians and Gynecologists practice guidelines [24]. Term deliveries were defined as those occurring at 37 weeks' gestation or later. Operative vaginal delivery was defined as vacuum-assisted or forceps-assisted delivery. Multidisciplinary care conference documents, delivery notes and operative reports were examined to determine whether maternal cardiac status dictated the obstetric management. Maternal cardiac indication was defined as follows: termination of pregnancy recommended by the multidisciplinary care team because of the maternal risks of continuing the pregnancy; induction of labour before 39 weeks because of deteriorating maternal cardiac status; planned operative vaginal delivery; or elective caesarean delivery without a trial of labour in the absence of obstetric indications.

While the Royal College of Obstetricians and Gynaecologists classification of urgency of caesarean delivery [25] is not used in clinical practice in our institution, we assigned a designation (category 1, 2, 3 or 4) to each caesarean delivery for the purposes of this analysis.

Labour analgesic techniques were classified as intravenous opioid only (no neuraxial technique), epidural or combined spinal–epidural (CSE). Anaesthetic techniques for caesarean delivery were classified based on anaesthesia records and operative reports as general anaesthesia, single-shot spinal, epidural in situ, epidural de novo, low-dose CSE, and standard CSE. Single-shot spinal was defined as intrathecal (spinal) dosing without placement of an epidural or intrathecal catheter. Epidural in situ was defined as the successful use of an existing epidural catheter originally placed for labour analgesia. If the in-situ catheter was not successfully used for caesarean delivery and a second block was placed, the technique of the second block was recorded as the technique for the case. Epidural de novo was defined as the initial placement of an epidural catheter in the operating room and epidural medication dosing without an intrathecal dose component. Institutional preference is for lidocaine 2% to provide epidural anaesthesia. Low-dose CSE was defined as a CSE technique placed in the operating room in which the intrathecal local anaesthetic dose was lower than the standard institutional dose of bupivacaine 12 mg with subsequent supplementation via the epidural catheter. Standard CSE was defined as an intrathecal dose consistent with institutional practice for single-shot spinal dosing, plus the concomitant placement of an epidural catheter.

Maternal intensive care unit (ICU) and coronary care unit admissions were recorded based on the immediate or delayed (≤72 h) postpartum disposition of the patient. Admission to a telemetry bed on a standard-acuity surgical floor or the use of continuous pulse oximetry on the standard postpartum floor was not counted as ICU admission. Neonatal mortality was defined as any death within 30 days of birth.

Descriptive statistics were used to summarise anaesthetic management and obstetric, maternal and neonatal outcomes. (IQR [Range]) proportion Reference comparison values were obtained from the overall obstetric anaesthesia database (the population of all deliveries at our institution) for the study period. Operative vaginal delivery comparison data were available only for the period from 2009 to 2011.

Study cohort statistics were compared with the institutional comparison statistics using chi-squared χ2 and Fisher's exact tests, as appropriate. Trends over the time in the study cohort caesarean delivery rate, operative vaginal delivery rate, and epidural rate were examined using a Mann–Kendall test for trend (a non-parametric test to determine the presence and direction of a trend over time [26]). Trends in the reference comparison rates were examined using a random effects linear regression (the equivalent parametric test). Analyses were performed with Microsoft Excel (version 2007; Microsoft Corporation, Redmond, WA, USA) and SAS (version 9.3; SAS Institute, Cary, NC, USA) with a predetermined value of p > 0.05 considered statistically significant.


A cohort of 107 cases (104 pregnancies, with three sets of twins) in 65 women was identified. There were 27 deliveries (25%) in women with complex congenital lesions. The cohort also included 18 (17%) left-sided obstructive lesions, 16 (15%) lesions primarily involving a defect in septation, 12 (11%) right-sided regurgitant lesions, 10 (9%) right-sided obstructive lesions, nine (8%) left-sided regurgitant lesions, and eight (7%) in which the primary defect was one of myocardial function (e.g. ventricular non-compaction, arrhythmogenic right ventricular dysplasia). Pulmonary hypertension of at least moderate severity was present in 21 cases (20%); it represented the primary lesion in seven cases (7%) and was a sequel of another primary structural lesion in 14 cases (13%).

Cyanosis was present in 8% of all cases. The woman had undergone at least one prior cardiac surgical operation in 58% of cases. Maternal characteristics and obstetric data are summarised in Table 1.

Table 1. Patients' characteristics and obstetric data of 104 women with congenital heart disease lesions and/or pulmonary hypertension. Values are mean (SD) or number (proportion)
Age; years30 (7)
Caucasian34 (33%)
Hispanic39 (38%)
Asian29 (28%)
Black2 (3%)
Nulliparous55 (53%)
ASA physical status
1-27 (7%)
3 76 (73%)
421 (20%)

There were 18 abortions (16 first-trimester and two second-trimester) and 89 births. Maternal cardiac status was the indication for 10 (56%) terminations. Of vaginal deliveries, 19 (38%) were operative, compared with a 10.5% operative vaginal delivery rate in our institution (p < 0.01). Operative vaginal deliveries included 10 vacuum-assisted deliveries (53%) and nine forceps-assisted deliveries (47%). Maternal cardiac status dictated the mode of delivery in 12 operative vaginal deliveries (63%). Figure 1 shows the caesarean delivery rate over the 12-year study period. The overall caesarean delivery rate was 43% in the study cohort, compared with our institutional average of 27% over the same period (p = 0.02). Maternal cardiac status was the indication for nine caesarean deliveries (24%), of which two were category-1, two were category-2, three were category-3, and two were category-4. Both category-1 deliveries occurred in patients with pulmonary hypertension. Elective repeat, breech and prior uterine surgery were the obstetric indications in the remaining cases of caesarean delivery. Eight women (22%) had a trial of labour before their caesarean delivery, none of which were category-1. No statistically significant trend was present in caesarean delivery rate (p = 0.20), or operative vaginal delivery rate (p = 0.42) in our study cohort over the study period. There was a significant increase in caesarean delivery (p < 0.01), but no significant trend in operative vaginal delivery (p = 0.25) at our institution over the study period.

Figure 1.

The rate of caesarean delivery in 3-year periods over the 12-year study period (2000–2011). The dark bars indicate the study cohort of patients with congenital heart disease and/or pulmonary hypertension; light bars indicate all cases in the institution over the same time period.

There were 23 (26%) postpartum maternal ICU admissions, compared to an institutional rate of maternal ICU admission for all maternity patients of 0.3% (p < 0.01). Of ICU admissions, 16 were for monitoring only, without any noted complications. The seven maternal ICU admissions for serious morbidity included three primarily for postpartum haemorrhage and four primarily for cardiac complications (heart failure and/or arrhythmia). There was one maternal death following a category-1 caesarean delivery performed under general anaesthesia at 35 weeks, due to decompensated cor pulmonale in a woman transferred from an outside hospital with primary arterial pulmonary hypertension, cyanosis (oxygen saturation of 85% on presentation), and impending respiratory failure. The woman died in the ICU on postpartum day 1 from right ventricular failure despite mechanical ventilation and maximum inotropic and vasopressor support. The infant survived without sequelae.

Neuraxial analgesia was provided in 83% of spontaneous deliveries and 89% of operative vaginal deliveries (Table 2). Neuraxial analgesia was provided in 88% of all labouring patients. These rates compare with our institution's neuraxial analgesia rate of 83% over the same time period. Thirty-five (95%) caesarean deliveries were performed under a neuraxial technique (Table 3). The dose range of 0.75% hyperbaric bupivacaine for the low-dose CSE technique was 3–6 mg (0.4–0.8 ml); supplementary epidural doses of 2% lidocaine ranged from 100 to 440 mg (5–22 ml). Adjunctive intravenous analgesics (fentanyl and ketamine) were required in one caesarean delivery under an in-situ labour epidural. Two caesarean deliveries were performed under general anaesthesia, both of which were category-1 deliveries in unstable patients with pulmonary hypertension. Eight (22%) caesarean deliveries occurred in the main operating room rather than the obstetric unit operating rooms. Invasive monitoring was used in 14 (38%) caesarean deliveries and 25 (29%) deliveries in total. Central venous catheters were used in 10 (11%) deliveries; one pulmonary artery catheter was placed for a caesarean delivery (in the woman that subsequently died).

Table 2. Analgesic and anaesthetic technique utilised for pain relief during labour. Values are number (proportion)
 Spontaneous vaginal delivery (n = 30)Operative vaginal delivery (n = 19)
  1. CSE, combined spinal–epidural.

Epidural22 (73%)14 (74%)
CSE3 (10%)3 (15%)
Intravenous opioid5 (16%)2 (10%)
Table 3. Anaesthetic technique utilised for caesarean delivery according to urgency category. Values are number (proportion)
 Category 1 (n = 2)Category 2 (n = 2)Category 3 (n = 9)Category 4 (n = 24)All (n = 37)
  1. CSE, combined spinal–epidural.

General anaesthesia2 (100%)0002 (5%)
Single-shot spinal0006 (25%)6 (16%)
Epidural de novo00012 (50%)13 (35%)
Epidural in situ008 (89%)07 (19%)
Low-dose CSE01 (50%)03 (13%)4 (11%)
Standard CSE01 (50%)1 (11%)3 (13%)5 (14%)
Table 4. Neonatal Apgar scores in the study cohort. Values are median (IQR [range]) or number (proportion)
 Spontaneous vaginal delivery (n = 31)Operative vaginal delivery (n = 20)Caesarean delivery (n = 38)
1-min Apgar8.1 (7–9 [2–9])8.3 (8–9 [7–9])7.6 (6–9 [3–9])
5-min Apgar8.6 (7–9 [8–9])8.9 (8–9 [8–9])8.8 (8–9 [6–9])
1-min Apgar < 72 (6%)05 (14%)
5-min Apgar < 71 (3%)01 (3%)

Other surgical procedures performed during pregnancy included the following: one patient with a transcatheter pulmonary valvuloplasty due to worsening right ventricular dysfunction in the setting of pulmonary valve stenosis; one patient with Ebstein's anomaly who had synchronised cardioversions and attempted ablations for Wolf–Parkinson–White syndrome in three different pregnancies; and two patients, both with complex congenital lesions, who underwent open appendicectomy under general anaesthesia during the second trimester. All procedures were performed without any complications. In the immediate postpartum period, bilateral tubal ligation was performed in 10 patients, six at the time of caesarean delivery, and four as a separate postpartum procedure after vaginal delivery (one under spinal, one under low-dose CSE (0.5 ml bupivacaine supplemented with 8 ml epidural 2% lidocaine), two under epidural in situ, and none under general anaesthesia).

The preterm delivery rate was 25%, compared with the institutional rate over the same time period of 11% (p = 0.03). Two (2%) neonates died, compared with 0.5% in our institution overall (p = 0.48). Both neonatal deaths were very preterm (22 weeks, no resuscitation, and 25 weeks, died within 24 h). There were eight other infants with NICU admission for prematurity, without further complications. Apgar scores are shown in Table 4. Only one neonate had significant CHD evident in the neonatal period (truncus arteriosus with no apparent genetic relationship to the maternal lesion of repaired Tetralogy of Fallot).


There are limited data describing the anaesthetic management of pregnant women with CHD. This retrospective analysis provides detailed information on anaesthetic and obstetric management of a large cohort of these patients. Our results outline past and recent management strategies and demonstrate that the majority of women undergo successful vaginal delivery. However, there is an increased rate of caesarean delivery compared with our overall institutional rate, which appears to be driven by caesarean deliveries performed for maternal cardiac indications.

During a 12-year period, our institutional caesarean delivery rate increased significantly; however, the caesarean delivery rate in the study cohort did not and actually appears to trend down, although this trend did not reach statistical significance. This is consistent with prior reports which suggest a shift in management strategies towards vaginal delivery in patients with significant cardiac disease [6]. The observation in our study of a 43% rate of caesarean delivery with 24% of those performed for maternal cardiac indications is consistent with a large case series from Japan (n = 151), in which the caesarean delivery rate was 44%, with 34% performed for maternal cardiac indications [21].

The comparatively high rate of operative vaginal deliveries in our study (38% compared to the 11% institutional rate) reflects the use of what is often referred to as a ‘cardiac’ delivery (i.e. minimal maternal pushing with operative assistance under dense neuraxial block). It should be noted that comparison institutional rates for this outcome were available only for 2009 through 2011. Although there are no clear randomised trials supporting this strategy compared with spontaneous vaginal or elective caesarean delivery, it appears to be a common management strategy in our population in the absence of emergency circumstances. We appear to utilise operative vaginal delivery less frequently than in a similar retrospective cohort study of obstetric outcomes from Boston (study period 1998–2004), in which 35 of 55 (64%) vaginal deliveries were operative. While it is possible that this simply reflects differing institutional preferences, it may also suggest a temporal shift in obstetric management, with a greater willingness to allow unassisted vaginal deliveries compared with prior time periods.

A key finding from our study was the high usage of neuraxial analgesia or anaesthesia, as the vast majority of deliveries, vaginal or caesarean, were successfully managed with a catheter-based neuraxial anaesthetic technique. Although a single-shot spinal is the preferred anaesthetic technique for healthy patients undergoing caesarean delivery at our institution, this technique accounted for only 16% of caesarean anaesthetics. The frequent utilisation of epidural de novo or in situ and low-dose CSE strategies is likely to be explained by the desire for a more gradual onset of blockade and the ability to allow subsequent dosing if clinically indicated. The varied neuraxial techniques utilised demonstrate that there is no ‘gold-standard’ technique, and that the anaesthetic management was individualised. General anaesthesia was limited to only category-1 caesarean deliveries in unstable pulmonary hypertension patients, so comparisons between anaesthetic techniques cannot be made in this study cohort.

Invasive monitoring was used infrequently for both vaginal and caesarean deliveries. The only use of a pulmonary artery catheter was in a patient with pulmonary hypertension who was decompensating on admission to our institution. Our findings are consistent with a more selective, as opposed to routine, use of invasive monitoring (especially pulmonary artery catheters) in this population.

Maternal and neonatal outcomes overall were favourable. We observed a high rate of preterm delivery compared with institutional rates. This study did not have the power necessary to demonstrate risk factors for adverse maternal outcomes nor to evaluate the necessity or cost-effectiveness of postpartum ICU admission for monitoring of stable patients. The mortality rate of pulmonary hypertension in this study (1 of 7 patients with primary pulmonary hypertension, 1 of 18 all patients with moderate to severe pulmonary hypertension) is consistent with recent small studies [27, 28]. Our findings suggest that although pulmonary hypertension remains a high-risk condition, the prognosis in the modern era may have improved from the historically quoted 30–60% risk of mortality [6, 29]. Larger multicentre studies are needed to evaluate further whether outcomes of pulmonary hypertension in pregnancy have improved over time.

This study has several potential inherent limitations. Assessment of case severity was retrospective and based on type of lesion and ASA status, not on more robust individual haemodynamic measurements at the time of delivery. These variables are imperfect proxies for true functional status, and adult CHD patients have a significant rate of subclinical abnormalities in exercise tolerance [30, 31], perhaps because their lifelong perception of normal activity level may differ from that of the general population. Despite collecting data over a 12-year period, the cohort size lacked sufficient numbers to allow for robust analysis of anaesthetic and obstetric management and patient outcomes. Our decision to limit the study to cases of moderate to complex CHD or moderate to severe pulmonary hypertension was done in keeping with the established framework for outpatient care of adult CHD patients. It is recommended that patients with simple lesions may not need specialist care, but those with moderate to complex CHD do [23]. In addition, well-compensated patients with simple lesions will be less reliably identified by retrospective or prospective methods, and these lesions, even if recognised, may be of questionable clinical significance. We acknowledge that selection of patients with moderate to severe disease certainly reduced the overall sample size and reduced the power of the study for trend and subgroup analysis. However, we believe that the trade-off between the smaller study size and the more uniform inclusion of cases with a clinically significant lesion favours our strategy. Our study provides both temporal and detailed anaesthetic and peripartum data in a study population where data are largely lacking.

During the initial phase of case identification, we observed a substantial degree of error in the billing codes associated with CHD diagnoses. In addition to the inherent limitations that are present in any study that relies on billing codes for case identification, it appears that obstetrics is an arena where additional potential for error exists. Increased errors may occur because of inclusions of fetal diagnosis of CHD in the maternal medical record. Initial search queries for CHD diagnosis codes yielded a majority of cases that represented fetal, not maternal, CHD. We estimate that the ratio of fetal to maternal cases identified in this manner was approximately 5:1. Although we are confident that subsequent case-by-case review accurately excluded cases in which no maternal CHD diagnosis existed, this finding should lead us to question prior or subsequent studies that rely solely on administrative data without individual case review for validation. This technique is likely to underestimate the rate of adverse maternal outcomes (women who are in fact healthy) and overestimate the rate of adverse neonatal outcomes (neonates with CHD).

Our single-institution study also has limited external validity to the degree that care patterns remain somewhat idiosyncratic. Patient populations may differ between institutions because of geography, referral patterns and associations with the availability and quality of maternal–fetal medicine, adult congenital cardiology, and congenital cardiac surgical services.

In conclusion, this 12-year retrospective cohort analysis provides detailed anaesthetic and peripartum management of women with moderate to complex CHD and pulmonary hypertension. This single-institution cohort study revealed almost universal use of neuraxial anaesthetic techniques, increased caesarean and operative delivery rates, and largely favourable maternal and neonatal outcomes. In the absence of large multicentre studies or a national registry for pregnant women with CHD, descriptive information of large cohorts from single centres provides the best available data, that can be used as a framework for the management of this patient population. Multicentre studies and registries are needed to compare anaesthetic and obstetric management strategies and to delineate risk factors associated with adverse outcomes.


The authors gratefully acknowledge the clinical informatics assistance and support of Susan Weber and Gomathi Krishnan.

Competing interests

No external funding and no competing interests declared.