Outcomes of cardiac surgery in patients less than 2.5 kg: impact of patient-dependent and patient-independent variables

Abstract

Objective

A recent STS database study showed that low weight (<2.5 kg) at surgery was associated with a high operative mortality (16%). We thought to assess outcomes after cardiac repair in patients <2.5 kg as compared to 2.5-4.5kg in an institution with a dedicated neonatal cardiac program; and to determine the potential role played by prematurity, STAT risk categories, uni/biventricular pathway, and timing of surgery.

Methods

We analyzed outcomes (hospital mortality, early reintervention, postoperative length of stay (postopLOS), mortality (at last follow-up) in patients <2.5kg at time of surgery (n=146; group1) and 2.5-4.5kg (n=622; group2), who underwent open or closed cardiac repairs from January 2006 to December 2012 at our institution. The statistical analysis was stratified by prematurity, STAT risk categories, uni/biventricular pathway and “usual”/“delayed” timing of surgery. A uni/multivariate risk analysis was performed. Mean follow-up was 21.6±25.6 months.

Results

Hospital mortality in group 1 was 10.9% (n=16) vs. 4.8% (n=30) in group 2 (p=0.007). PostopLOS and early un-planned reintervention rates were similar between the two groups. Late mortality in group 1 was 0.7% (n=1). In Group 1, early outcomes were independent of STAT risk categories, uni/biventricular pathway or timing of surgery, as opposed to group 2. Lower gestational age at birth was an independent risk factor for early mortality in group 1.

Conclusions

A dedicated multidisciplinary neonatal cardiac program yields good outcomes for neonates and infants <2.5kg independently of STAT risk categories and uni/biventricular pathway. Lower gestational age at birth was an independent risk factor for hospital mortality.

Despite improvements in outcomes in neonatal cardiac surgery over the last 20 years, low weight remains a risk factor for increased mortality in neonates and infants undergoing cardiac surgery ¹. A Society of Thoracic Surgeons (STS) Congenital Heart Surgery Database study with 32 participating centers recently demonstrated that the average operative mortality rate in patients with a low weight (≤2.5 kg) at surgery was as high as 16% ². Moreover, the risk factors for mortality and reintervention in this specific population are still controversial ^3-10. To our knowledge, the potential role played by the STAT risk categories, the uni/biventricular pathway, timing of surgery and gestational age were never investigated in a study comparing directly 2 groups of patients (≤2.5 kg and >2.5 kg). Thus, the objectives of our study were to 1) assess the early and mid-term outcomes of cardiac repair in patients ≤2.5 kg in an institution with a dedicated neonatal cardiac program; 2) compare these results to those of patients between 2.5-4.5 kg operated in the same institution; 3) determine the potential role played by gestational age, STAT risk categories, the uni/biventricular pathway and timing of surgery; 4) and perform a uni/multivariate risk analysis in the group of patients ≤2.5 kg.

Materials and Methods

Methods

This retrospective single center study included patients who had open or closed cardiac surgery at the Morgan Stanley Children's Hospital - New-York Presbyterian, Columbia University, from January 2006 to December 2012 with a weight ≤2.5 kg at the time of surgery (group 1) or between 2.5 and 4.5 kg (group 2). Patients who underwent ductus arteriosus closure alone were not included in the study. Perioperative data were retrospectively collected by reviewing the hospital records and the computerized database of our department. Follow-up data were obtained from the institution outpatient records and the same computerized database. The dedicated neonatal cardiac program offered care to neonates or young infants with congenital heart disease from birth to discharge. A dedicated medical and nursing team staffed this program and included members from the Divisions of Neonatal Intensive Care, Pediatric Cardiology and Pediatric Cardiac Surgery. Practitioners in this team had either received advanced training in pediatric cardiac intensive care and/or had advanced understanding and are skilled in the management of newborn babies with congenital heart disease. Dedicated neonatal cardiac intensive care nurses and neonatal nurse practitioners, neonatal respiratory therapists, neonatal nutritionists and feeding specialists staffed the neonatal cardiac intensive care section. Patients who were born at our institution or transferred from outside medical centers with known or suspected congenital heart disease were admitted to the neonatal cardiac intensive care section of the NICU. While there were no clear cut restrictions to admission to the neonatal cardiac intensive care section, infants older than 6-8 weeks of age at the time of transfer or admission are preferably admitted to our pediatric cardiac intensive care unit.

The STAT risk categories, uni/biventricular pathway and timing of surgery were assigned for each patient included in the study. The STAT risk categories were based on the definition of the categories established by O'Brien et al ¹¹. The surgical pathway was defined as “univentricular” (Norwood procedure, aortopulmonary shunt, pulmonary artery band and/or atrial septectomy performed for single ventricle disease), “biventricular” (primary biventricular complete repair) or “palliation towards biventricular” (systemico-pulmonary shunt, pulmonary artery band, unifocalization and/or transannular patch as a first step before biventricular repair). The timing of surgery was defined as “usual” (based on usual divisional management paradigms in infants of normal weight) or “delayed” (intentional delay in early intervention or unsual choice of a palliative approach to permit growth or maturation), as previously described by Hickey et al ¹². This adjudication of the surgical timing was based on an accurate and precise retrospective chart review made by a senior co-author. The mean age at surgery for patients with usual timing of surgery was 10.9±11.9 days versus 50.6±36 days for those with delayed surgery.

The primary end-point was mortality occurring prior to hospital discharge or within 30 days after the operation. Secondary endpoints were post-operative hospital length of stay (postopLOS), unplanned early reintervention (during the same hospital stay) and late mortality. This study was approved by the Institutional Review Board of Columbia University Medical Center.

Population

A total of 146 and 622 patients were included in group 1 and group 2 respectively. Demographic and surgical characteristics, cardiac diagnoses, STAT categories, surgical pathway and timing of surgery of both groups are presented and compared in Table 1. Patients in Group 1 had a STAT 4 or 5 risk category in 70% (n=102), a biventricular pathway in 63% (n=92) and a delayed timing in 18.5% (n=27). Compared to group 2, group 1 is characterized by a statistically significant lower male/female ratio (p=0.01), older age at surgery (p<0.001), more patients with tetralogy of Fallot (TOF) (p=0.006), and aorto-pulmonary window (p=0.02), but fewer with simple transposition of great arteries (TGA) (p=0.04), more STAT 2 category (p=0.01) and a shorter bypass time (p=0.01). Single ventricles, right heart lesions, aortic arch lesions and TGA were the most frequent group of diseases in both groups. Patients in group 1 had extracardiac malformations and a genetic syndrome in 18.4% and 13% of cases respectively. The mean gestational age at birth in patients in group 1 was 35.5±2.8 weeks. Fifty-six % of patients (n=82) were premature (<37 weeks) and 42% (n=61) had a gestational age <36 weeks. Forty-two % were small for their gestational age (SGA, defined as <10 percentile). Forty-one% (n=33) of STAT 4 patients in group 1 had a gestational age <36 weeks, versus 25% (n=6) in STAT 5 patients (p=0.15). Twenty-two % (n=9) of patients who underwent a univentricular pathway in group 1 had a gestational age <36 weeks, versus 48% (n=44) in those with a biventricular pathway (p=0.007) and 61% (n=8) in those with a palliation towards a biventricular repair (p=0.014).

Table 1.

Patient characteristics and overall outcomes.

	Group 1 (<2.5kg) (n=146)	Group 2 (2.5-4.5 kg) (n=622)	p-value
Sex: Male	49% (n=71)	60.7% (n=377)	0.01*
Birthweight (mean ± SD)	2.1 ± 0.4 kg	3.1 ± 0.5 kg	NA(a)
Weight at surgery (kg) (mean ± SD)	2.2 ± 0.3 kg (range: 1.1-2.50)	3.2 ± 0.4 kg (range: 2.51-4.4)	<0.001*
Age at surgery (d): (mean ± SD)	18.2 ± 24.2 d (range: 1-193)	10.1 ± 15.3 d (range: 0-215)	<0.001*
Proportion of neonates (<31 day-old)	n=112/146 (76%)	n=592/622 (95%)
Cardiac diagnoses
Single ventricle	26.1% (n=38)	30.5% (n=190)	0.27
Hypoplastic left heart syndrome	14.4% (n=21)	20.2% (=126)	0.13
Tricuspid atresia	4.1% (n=6)	2.9% (n=18)	0.42
Double-inlet left ventricle	3.4% (n=5)	3.7% (n=23)	1
Unbalanced AVSD	2.1% (n= 3)	1.8% (n=11)	0.72
Heterotaxy syndromes	1.4% (n=2)	1.8% (n=11)	0.75
Single ventricle, NOS	0.7% (n=1)	0.6% (n=1)	1
Right heart lesions	20.5% (n=30)	11,4% (n=71)	0.006 *
Tetralogy of Fallot	11.6% (n=17)	4.8% (n=30)	0.006 *
Pulmonary atresia, VSD	4.8% (n=7)	3.2% (n=20)	0.61
Pulmonary atresia, no VSD	2.7% (n=4)	2.9% (n=18)	0.57
Pulmonary stenosis	1.4% (n=2)	0.5% (n=3)	0.32
Aortic arch lesions	15% (n=22)	18% (n=112)	0.78
Interrupted aortic arch	4.1% (n=6)	3.7% (n=23)	0.18
Coarctation of aorta + hypoplastic arch	5.5% (n=8)	9.2% (n=57)	0.24
Coarctation of aorta + VSD	5.5% (n=8)	5.1% (n=32)	0.63
TGA	15% (n=22)	21.5% (n=134)	0.13
TGA, intact ventricular septum	8.9% (n=13)	13.7% (n=85)	0.045*
TGA, VSD	5.5% (n=8)	5.3% (n=33)	0.83
TGA VSD aortic arch hypoplasia	0.7% (n=1)	2.1% (n=13)	0.33
ccTGA	n=0	0.5 % (n=3)	1
VSD	6.2% (n=9)	3% (n=19)	0.15
Truncus arteriosus	5.5% (n=8)	2.2% (n=14)	0.06*
TAPVR	6.2% (n=9)	7.6% (n=47)	0.37
AVSD	1.4% (n= 2)	0.6% (n=4)	0.41
Aorto pulmonary window	2.7% (n= 4)	0.16% (n=1)	0.02*
Ebtsein anomaly	0.7% (n=1)	0.16% (n=1)	0.09
Double outlet right ventricle	n=0	2.6% (n=16)	0.07
Cardiac tumor	0.7% (n=1)	0.8% (n=5)	1
Aortic stenosis	n=0	1% (n=6)	NA
Coronary anomalies	n=0	0.3% (n=2)	1
Pathway:
- Univentricular surgery	- 28.1% (n=41)	- 33.9% (n=208)	0.2
- Palliation towards biventricular repair	- 8.9% (n=13)	- 7.7% (n=48)	0.73
- Biventricular repair	- 63% (n=92)	- 58.6% (n=365)	0.37
Timing of surgery:
- Usual	- 81.5% (n=119)	- 100% (n=622)	NA
- Delayed	- 18.5% (n=27)	- 0%
STAT risk categories:
- 1	- 6.8% (n=10)	- 3.7% (n=23)	0.11
- 2	- 11% (n=16)	- 5.1% (n=32)	0.013*
- 3	- 12.3% (n=18)	- 18.5% (n=115)	0.09
- 4	- 53.4% (n=78)	- 49.5% (n=308)	0.4
- 5	- 16.4% (n=24)	- 23.1% (n=144)	0.095
Open-heart cases	94.5% (n=138)	91.5% (n=569)	0.3
Bypass time: mean ± SD	114±46 min	123±44 min	0.011*
Cross clamp time: mean ± SD	52±22 min	57±30 min	0.11
Early mortality: N (%)	10.9% (n=16)	4.8% (n=30)	0.0069
Postoperative length of stay (d): mean ± SD	20.3±24.5	19.6±24.7	0.46
Early reintervention: N (%)	6.2% (n=9)	4.9% (n=31)	0.55
Mortality at last follow-up: N (%)	0.7% (n=1)	3% (n=18)	0.15

AVSD, Atrioventricular septal defect; NOS, not otherwise specified; TGA, transposition of the great arteries; VSD, ventricular septal defect.

^(a)Missing birthweight data in group 2 made the comparison not methodologically acceptable.

Frequencies of the main procedures in groups 1 and 2 are described in the additional Table 1. The four most frequent procedures performed in Group 1 were the Norwood procedure (n=25; 17%), the arterial switch operation (n=13; 8.9%), a primary repair of TOF (n=12; 8.2%) and the shunt palliation or unifocalization for TOF/pulmonary atresia with ventricular septal defect (n=12; 8.2%). The latter procedure were perfomed more frequently in group 1 (8.2%; n=12) than in group 2 (3.2%; n=20) (p=0.01). The frequencies of all other procedures were not significantly different between the 2 groups. Twenty% (n=5) of patients who underwent a Norwood procedure in Group1 had a gestational age <36 weeks, versus 62% (n=5) in those who had a hypoplastic aortic arch repair (p=0.036).

Additional Table 1. Frequencies of the main procedures.

	Group 1 (<2.5kg) (n=146) n (%)	Group 2 (2.5-4.5 kg) (n=622) n (%)	p-value
Procedures for Single ventricle
Single ventricle / Norwood	n=25 (17%)	n=141 (22.7%)	0.14
Single Ventricle / Conduit or shunt	n=9 (6.2%)	n=36 (5.8%)	1
Single ventricle / isolated atrial septectomy	n=1 (0.7%)	n=1 (0.2%)	1
Single ventricle / PA band	n=3 (2%)	n=12 (1.9%)	1
Procedures for Right Heart lesions
Tetralogy of Fallot – PA-VSD/ Primary repair	n=12 (8.2%)	n=30 (4.8%)	0.11
Tetralogy of Fallot -PA-VSD/ Shunt palliation or unifocalization	n=12 (8.2%)	n=20 (3.2%)	0.01*
PA-IVS / RVOT repair	n=2 (1.4%)	n=8 (1.3%)	1
PA-IVS / Shunt palliation	n=2 (1.4%)	n=10 (1.6%)	1
Pulmonary stenosis repair	n=2 (1.4%)	n=3 (0.5%)	0.24
Procedures for aortic arch lesions
Coarctation-hypoplastic arch / aortic arch repair	n=8 (5.5%)	n=57 (9.2%)	0.19
Coarctation-VSD / aortic arch+VSD repair	n=8 (5.5%)	n=32 (5.1%)	1
Interrupted aortic arch repair	n=6 (4.1%)	n=23 (3.7%)	0.81
Procedures for simple/complex TGA
TGA-IVS / ASO	n=13 (8.9%)	n=85 (13.7%)	0.13
TGA-VSD / ASO + VSD closure	n=8 (5.5%)	n=33 (5.3%)	1
TGA-VSD-PS / Complete repair	n=1 (0.7%)	n=7 (1.1%)	1
TGA-VSD-PS / Palliation	n=0	n=6 (1%)	0.62
Other procedures	n=9 (6.2%)	n=47 (7.6%)	0.6
TAPVR / Repair
VSD/Primary repair	n=8 (5.5%)	n=19 (3%)	0.2
VSD/PA band	n=1 (0.7%)	n=0	0.2
Truncus arteriosus/Repair	n=7 (4.8%)	n=14 (2.2%)	0.1
Truncus arteriosus / PA band	n=1 (0.7%)	n=0	0.19
AVSD/Primary repair	n=2 (1.4%)	n=3 (0.5%)	0.24
AVSD/PA band	n=0	n=1 (0.2%)	1

Statistical analysis

Descriptive statistics were performed and stratified by the categories ≤2.5kg and >2.5kg. Bivariate testing by weight category for demographic and surgical characteristics and for the outcomes of hospital mortality, postopLOS, early unplanned reintervention and late mortality was performed for categorical variables using Fischer's Exact or Chi-squared tests, and for continuous variables using Student's t-test, Wilcoxon-Rank Sum test, or ANOVA or Kruskal-Wallis. Multiple comparisons were explored using Tukey's test. Trends for ordinal independent variables were performed using P-for trend (ANOVA) or Cochran-Armitage test for trend. Comparison between the endpoints and demographic and surgical characteristic between weight categories were performed by Cochran Mantel-Haenszel or 2-way ANOVA, first testing for interaction by Breslow-Day test or by creating interaction terms in 2-way ANOVA. The statistical analysis was stratified by the STAT risk categories, surgical pathway and timing of surgery.

A risk analysis was performed for the endpoints early mortality and early unplanned reintervention in group 1. Univariate analysis was performed for categorical variables using Fischer's Exact or Chi-squared tests, and for continuous variables using Student's t-test or Mann-Whitney test. Multivariate analysis was performed using a logistic regression model to estimate risk factors for early mortality and early unplanned reintervention. Variables were included into the model by backward elimination if p<0.05. Adjustment factors for multivariate analysis were the surgeon, STAT score, type of procedure (uni/biventricular) and gestational age at birth. Postoperative complications were defined as at least one of the following postoperative major events: cardiac arrest, ECMO, arrhythmia, atrioventricular block requiring a pacemaker, diaphragm paralysis, atelectasis, pleural effusion requiring thoracocentesis, respiratory failure, seizures, cerebrovascular event, renal failure requiring dialysis, necrotizing enterocilitis, mediastinis or any septic syndrome. For all tests, a p-value of less than 0.05 was considered significant. SAS 9.3 (Cary, NC) was used for data analysis.

Results

Overall outcomes are described in Table 1. Hospital mortality in group 1 was significantly higher in group 1 than in group 2 (10.9% vs. 4.8%; p=0.0069). Causes of hospital deaths in group 1 were related to heart failure (n=12), respiratory failure (n=2) and sepsis (n=2). PostopLOS, early un-planned reintervention rate and mortality at last follow-up were similar between the two groups. Late mortality in patients <2.5kg was 0.7% (n=1).

Outcomes stratified by gestational age, STAT scores, surgical pathway and surgical timing

Outcomes in patients<2.5kg, stratified by gestational age, STAT scores, surgical pathway and surgical timing are presented in Table 2. In this group, each outcome was independent of STAT risk categories, uni/biventricular pathway or timing of surgery. Hospital mortality was 8.3% (n=2) in STAT 5 category and 12.2% (n=5) for the patients in the univentricular pathway. Hospital mortality was 9.2% (n=11) when the surgical timing was “usual”, vs. 18.5% (n=5) when surgery was delayed to allow growth and maturation of the patient (p=0.18). Hospital mortality was significantly associated with a gestational age <36 weeks (18% vs. 5.8%; p=0.03) (Table 2). Outcomes of patients>2.5kg, stratified by STAT scores and surgical pathway showed that early mortality rate and postopLOS were significantly increased in higher STAT categories (p<0.001 and p=0.007, respectively). Early mortality, postopLOS and early unplanned reintervention were also significantly more frequent in the univentricular pathway compared the biventricular pathway (p<0.001 for each outcome). In this group 2, hospital mortality was 10.4% (n=15) in STAT 5 category and 9.9% (n=21) for the patients in the univentricular pathway.

Table 2.

Outcomes of surgery in patients<2.5kg, by STAT scores, uni/biventricular pathway and surgical timing.

Patients<2.5kg at surgery
RISK CATEGORIES: STAT scores	1	2	3	4	5	p-value
Early mortality: N (%)	0/10	3/16 (18.7%)	1/18 (5.6%)	10/78 (12.8%)	2/24 (8.3%)	0.53
Postoperative length of stay (d): mean ± SD	17.7 (±13.8)	24.8 (±27.1)	13.5 (±9)	20.6 (±27.8)	22.8 (±22.6)	0.68
Early reintervention: N (%)	0/10	1/16 (6.2%)	1/18 (5.6%)	6/78 (7.7%)	1/24 (4.2%)	0.93
SURGICAL PATHWAY	Biventricular		Palliation towards biventricular		Univentricular	P-value
Early mortality: N (%)	9/92 (9.8%)		2/13 (15.4%)		5/41 (12.2%)	0.8
Postoperative length of stay (d): mean ± SD	20.5(±26.1)		16.2(±9.8)		21(±24.2)	0.97
Early reintervention: N (%)	4/92 (4.3%)		2/13 (15.4%)		3/41 (7.3%)	0.28
SURGICAL TIMING	Usual timing		delayed timing			p-value
Early mortality: N (%)	11/119 (9.2%)		5/27 (18.5%)			0.18
Postoperative length of stay (d): mean ± SD	20.8 (±26)		18.2 (±16.5)			0.85
Early reintervention: N (%)	7/119 (5.9%)		2/27 (7.4%)			0.67
GESTATIONAL AGE	<36 weeks		≥36 weeks			p-value
Early mortality: N (%)	11/61 (18%)		5/85 (5.8%)			0.03*
Postoperative length of stay (d): mean ± SD	24.2 (±31.6)		18.7 (±19.8)			0.21
Early reintervention: N (%)	1/61 (1.6%)		8/85(10.3%)			0.08

Comparative analysis of outcomes between groups 1 and 2, by diagnostic-procedure group, STAT scores and surgical pathway

Comparison of the mortality rates by diagnostic-procedure group between group 1 and group 2 are displayed in Table 3. The mortality rate was significantly higher in group 1 after aortic arch repair (for hypoplastic arch) compared to group 2 (25% vs. 0%; p=0.02). There was also a trend for a higher mortality in group 1 compared to group 2 after total anomalous pulmonary veinous return repair (22.2% vs. 4.3%; p=0.15), aortopulmonary shunt for TOF (16.7% vs.0%; p=0.16) and arterial switch associated to VSD closure (12.5% vs. 0%; p=0.21). Mortality rates were similar after the Norwood procedure (8% in group 1 vs 10.6% in group 2; p=1.00), or any palliative procedure for single ventricle. No death occured in patients <2.5 kg after arterial switch operation (0/13), interrupted aortic arch repair (0/6), truncus arteriosus repair (0/7), and pulmonary stenosis (0/2).

Table 3. Comparative analysis of mortality rates, by diagnostic-procedure, for both groups.

Diagnostic-Procedure group (n)	Mortality rate, overall % (n)	Mortality rate, patients <2.5kg % (n)	Mortality rate, 2.5-4.5kg % (n)	p-value
Single ventricle / Norwood	10.2% (n=17/166)	8% (n=2/25)	10.6% (n=15/141)	1.00
Single Ventricle / Conduit/shunt	11% (n=5/45)	22.2% (n=2/9)	8.3% (n=3/36)	0.30
Single ventricle / PA band	6.7% (n=1/15)	0% (n=0/3)	8.3% (n=1/12)	1.00
TGA/IVS/ASO	1% (n=1/98)	0% (n=0/13)	1.2% (n=1/85)	1.00
TGA/VSD/ASO+VSD closure	2.4% (1/41)	12.5% (n=1/8)	0% (n=0/33)	0.21
TAPVR/Repair	7.1% (n=4/56)	22.2% (n=2/9)	4.3% (n=2/47)	0.15
Coarctation hypoplastic arch/aortic arch repair via sternotomy	3% (n=2/65)	25% (n=2/8)	0% (n=0/57)	0.02*
Coarctation VSD/aortic arch + VSD repair	2.5% (n=1/40)	12.5% (n=1/8)	0% (n=0/32)	0.22
Interrupted aortic arch repair	3.6% (n=1/28)	0% (n=0/6)	4.3% (n=1/23)	1.00
Truncus arteriosus/Repair	4.8% (n=1/21)	0% (n=0/7)	7.1% (n=1/14)	1.00
Tetralogy of Fallot/Primary repair	4.8% (n=2/42)	8.3% (n=1/12)	3.3% (n=1/30)	0.51
Tetralogy of Fallot/Shunt palliation	6.2% (n=2/32)	16.7% (n=2/12)	0% (n=0/20)	0.16
Aorto pulmonary window repair	20% (n=1/5)	25% (n=1/4)	0% (n=0/1)	1.00
Cardiac tumor resection	16.7% (n=1/6)	100% (n=1/1)	0% (n=0/5)	0.28
Pulmonary stenosis repair	20% (n=1/5)	0% (n=0/2)	33% (n=1/3)	1.00

Caption: We could not compare procedures that were not performed in one of the 2, groups and procedures that were not complicated by any case of death.

Comparison of hospital mortality between groups 1 and 2, stratified by STAT scores and surgical pathway is showed in Table 4. After such a stratification, the inverse association between infant weight and hospital mortality persisted only in the STAT 4 category subgroup (12.8% in group 1 vs. 3.9% in group 2; p=0.005) and in the biventricular strategy subgroup (9.8% vs. 2.2%; p=0.02).

Table 4. Comparative analysis of mortality rates, by STAT scores and surgical pathway, both groups.

	Mortality rate, <2.5kg (n=146) N(%)	Mortality rate, 2.5-4.5kg (n=622) N(%)	Risk ratio (95%CI)	p-value
RISK CATEGORIES: STAT scores
1	0/10	1/23 (4.3)	NA	NA
2	3/16 (18.8)	1/32 (3.1)	6 (0.67,53.2)	0.1
3	1/18(5.6)	1/115 (0.87)	6.4 (0.42, 98)	0.25
4	10/78(12.8)	12/308 (3.9)	3.3 (1.48, 7.33)	0.005*
5	2/24 (8.3)	15/144(10.4)	0.8 (0.19, 3.3)	1
SURGICAL PATHWAY
Biventricular	9/92(9.8)	8/365 (2.2)	4.4 (1.8, 11.2)	0.02*
Palliation towards biventricular	2/13 (15.4)	1/48 (2)	7.4 (0.73, 75.2)	0.11
Univentricular	5/41 (12.2)	21/208(10)	1.2 (0.5, 3.1)	0.77

The postoperative length of stay stratified on the surgical strategy was significantly longer in group 1 (20.5±26.1 days) compared to group 2 (18.4±16.3 days) after biventricular repair only (p=0.02). The absence of association between weight and early unplanned reintervention persisted after stratification on STAT risk categories and surgical strategy.

Risk analysis

Risk analysis for early mortality in patients <2.5kg is presented in Table 5. The main factors associated with an increased risk of early mortality in the univariate analysis include the lower gestational age at birth (p=0.013), the occurence of postoperative complications (p<0.0001), cardiac arrest (p<0.0001), renal failure requiring dialysis (p=0.004) and an atrioventricular block requiring a pacemaker (p=0.03). The birthweight, the presence of major extracardiac malformations, a genetic syndrome, the age and weight at surgery, the STAT score, the surgical strategy, the timing of surgery, the surgeon and the need for an unplanned early reintervention were not associated with an increased risk of hospital mortality. The multivariate analysis demonstrated that lower gestational age at birth (OR=0.83;(0.69-0.98)) was an independent risk factor for early mortality.

Table 5. Risk analysis for early mortality in patients <2.5kg.

Univariate risk analysis for early mortality on patients <2.5kg
	survivors (n=130)	early deaths (n=16)	p-value
gestational age (continuous variable) (mean±DS)(weeks)	35.7±2.8	34±2.7	0.013*
gestational age <36 weeks	50 (38.5%)	11 (68.8%)	0.03*
prematurity (gestational age <37 weeks)	70 (53.8%)	12 (75%)	0.12
Birthweight	2.13±0.43 kg	2.02±0.44 kg	0.22
	AGA: 71 (54.6%)	AGA: 10 (62.5%)	0.26
	SGA: 57 (43.8%)	SGA: 5 (31.3%)
	LGA: 2 (1.5%)	LGA: 1 (6.3%)
Antenatal diagnosis	84 (64.6%)	11 (68.8%)	1.00
Major extracardiac malformation	25 (19.2%)	2 (12.5%)	0.73
genetic syndrome	17 (13.1%)	3 (18.8%)	0.46
Surgeon	surgeon 1: 19(14.6%)	surgeon 1: 2 (12.5%)	0.072
	surgeon 2: 53 (40.8%)	surgeon 2: 2 (12.5%)
	surgeon 3: 25 (19.2%)	surgeon 3: 4 (25%)
	surgeon 4: 33 (25.4%)	surgeon 4: 8 (50%)
Age at surgery (days) (median, range)	7 (1; 91)	12.5 (1;64)	0.25
Weight at surgery	2.22±0.25	2.12±0.35	0.26
STAT score (mean)	3.61±1.1	3.69±0.94	0.89
Pathway:	- Univentricular: 36 (27.7%)	- Univentricular: 5 (31.3%)	0.77
	- Palliation towards biventricular: 11 (8.5%)	- Palliation towards biventricular: 2 (12.5%)
	- Biventricular: 83 (63.8%)	- Biventricular: 9 (56.3%)
Timing of surgery	- Usual: 108 (83.1%)	- Usual: 11 (68.8%)	0.17
	- Delayed: 22 (16.9%)	- Delayed: 5 (31.3%)	0.30
Bypass	122 (93.8%)	16 (100%)	0.30
Bypass time	110±50	102±48	0.67
Postoperative complications	41 (31%)	16 (100%)	<0.0001*
delayed chest closure	15 (11.5%)	5 (31.3%)	0.047*
cardiac arrest	4(3.1%)	14 (87.5%)	<0.0001*
length of mechanical ventilation (median) (days)	4 (1; 32)	12 (1;41)	0.09
Arrhythmia	11 (8.5%)	7 (43.8%)	0.001*
AV block requiring pacemaker	1 (0.8%)	2 (12.5%)	0.03*
Pulmonary complications	17 (13.1%)	6 (37.5%)	0.02*
Renal failure requiring dialysis	0	1 (6.3%)	0.004*
Necrotizing entrocolitis	6 (4.6%)	2 (12.5%)	0.21
Early unplanned reintervention	7 (5.4%)	2 (12.5%)	0.25
Residual lesions at discharge	32 (24.6%)	6 (37.5%)	0.36
Multivariate analysis
	p-value	OR	95% confidence interval
gestational age (continuous variable, in weeks)	0.037	0.83	0.69-0.98

Factors of adjustment for the multivariate analysis: surgeon, STAT score, type of procedure (uni/biventricular), gestational age at birth.

The risk analysis for unplanned early reintervention in patients <2.5kg shows that the occurence of postoperative complications was an independent risk factor for early reinterventon (OR 18.9;(2.2-160). However, early reintervention in patients <2.5kg was independent of the surgeon, the STAT score, and the surgical strategy.

Discussion

Although a weight of <2.5kg at surgery remains a risk factor for early mortality, a dedicated neonatal cardiac program yields good early and mid-term outcomes for low weight neonates or infants independently of the STAT risk categories and surgical strategy. A lower gestational age at birth was an independent risk factor for early mortality in neonates or infants <2.5kg at surgery.

Despite major progress in neonatal cardiac surgery and postoperative care over the last 20 years, low weight at surgery remains associated with high mortality rates (10% to 24%) as well as with high morbidity rates ^3,7,10,13-16. A recent study from Philadelphia, showing patient demographics very similar to ours, reported a 24% hospital mortality rate for patients < 2.5kg ³. Another recent and large report of the STS Congenital Heart Surgery Database analyzed outcomes of cardiac surgery in over 500 infants weighing 1–2.5 kg from 32 participating centers ². This study demonstrated that this low weight cohort had significantly higher mortality for specific operations and that lower weight consistently increased risk of mortality after stratification by risk categories.

Our series confirms that a weight of <2.5kg at surgery remains a risk factor for early mortality, with a hospital mortality rate of 10.9%. The rate of early unplanned reintervention was not significantly different between patients <2.5 kg and >2.5kg, suggesting that technical surgical factors that might occur in tiny neonates were not the primary cause of the higher mortality seen in the < 2.5kg patients. This is further buttressed by the fact that factors like the STAT risk categories, the surgeon or the bypass time were also not related to mortality or early reintervention. Several technically complex procedures such as the arterial switch operation, interrupted aortic arch repair or truncus arteriosus repair were performed with no hospital mortality in the < 2.5kg patients. The expertise and the major role played by the surgical and perfusion team in the management of cardiopulmonary bypass in low weight neonates, with significant emphasis on obsessive fluid restriction such as priming volume reduction ¹⁷ and lack of postoperative bleeding to avoid blood product transfusions and lack of need for permanent pacemaker, cannot be overemphasized. The higher risk of mortality seems therefore to be more related to patient-dependent factors, especially prematurity, as demonstrated in the risk analysis. Finally, the fact that major postoperative complications were highly correlated with early interventions (OR 18.9) on multivariate analysis in patients < 2.5kg emphasizes the need to “get it right the first time” and avoid being forced to take a patient back for revision or interventional catheterization.

The mortality rate after Norwood procedure in patients <2.5kg was low at 8% (n=2/25) in our institution. The mortality rate for low-weight patients undergoing a Norwood procedure in the series based on the STS Congenital Heart Surgery Database was 30% ², with rates of 38% to 51% reported in other series ^{8, 18, 19}. Moreover, overall outcomes other than hospital mortality, such as postoperative length of stay, early unplanned reintervention rate and mid-term mortality rate in patients <2.5kg were not higher than those of patients>2.5kg. Such excellent early and mid-term outcomes reported in our series and others ^{4, 20} may rely on the expertise acquired by a multidisciplinary team dedicated to neonatal cardiac care in the setting of a high-volume program.

Outcomes in patients <2.5 kg were independent of the STAT risk categories and the uni/biventricular pathway, contrary to those >2.5kg. Curzon et al ² demonstrated in the STS database series that lower weight consistently increased risk of mortality after risk stratification. Palliation for a univentricular heart, diagnosis of HLHS and performance of a Norwood procedure have been reported as a risk factor for adverse outcome following cardiac surgery in low-weight infants ^{1-4, 8, 19, 21}. The reason for neutralization of risk categories and univentricular palliation as a risk factor in our series does not seem to be explained by a lack of statistical power since the STAT 5 category group and univentricular pathway group are one the most important groups in our series. The neutralization for these factors in patients<2.5kg seems to be related to the excellent outcomes in patients with HLHS (main procedure in the STAT 5 category group and univentricular pathway group) in our series. We speculate that the use of the right ventricle to pulmonary artery conduit has conferred a survival advantage to stage I palliation in low weight patients. Contrary to the Norwood procedure, some specific procedures performed at a weight <2.5kg still seem to be at a higher risk for early mortality such as total anomalous pulmonary venous return repair or aortopulmonary shunt for Tetralogy of Fallot. These findings are also consistent with the STS data ², particularly with respect to mortality after aortopulmonary shunts. Interestingly, the distribution of prematurity among the STAT categories, surgical strategy groups and procedures was not equal: patients with a STAT4 category, palliation towards a biventricular repair or an aortic arch repair had a lower mean gestational age compared to those with a STAT 5 category, a univentricular pathway or a Norwood procedure, which may explain the better outcomes in the STAT 5 category, given that, besides postoperative cardiac arrest, lower gestational age at birth was the strongest and only risk factor that remained after multivariate analysis.

The “usual” or “delayed” timing of surgery was not associated with a higher risk of any of the outcomes in this low weight population, thus implying at first view that delaying repair of congenital heart defects in low birth weight infants does not seem to confer a benefit. However, it could also mean that the “delayed” patients benefitted from the delay since their outcomes could have been worse had they been operated at a “usual” time. These findings are consistent with other series in literature ^{14, 21-27}. In particular, a study from Toronto ¹² showed that for neonates<2.0 kg, imposed delays in intervention neither compromise nor improve survival.

Our multivariate analysis demonstrated that low gestational age at birth was the only independent risk factor for early mortality in the low weight patients. This major variable could not be studied by Curzon et al ² since this variable was not captured by the STS congenital Database at the time this study was performed. However, other studies suggested a probable negative role played by prematurity in this specific population ^{3, 9, 15, 28}. Hickey et al showed that gestational age strongly influenced subsequent risk of death in a predictable way but was not a significant independent risk factor in relation to low birth weight ¹². This risk factor has been confirmed by other studies ²⁹ and should thus be taken into account in the management of such patients, such as delaying birth to full maturity whenever possible.

As for major extracardiac malformations, there have been contradictory reports of their influence on the risk for death in the studied population ^{3, 4, 7, 21, 30, 31}. Such an association is maybe strongest in the subgroup of patients undergoing a Norwood procedure ^{8, 18}. Similarly, the presence of a genetic syndrome was not associated with worse outcomes, contrary to the report by Azakie et al ⁴. This can be explained by the fact that two thirds of patients with a genetic syndrome had either Down syndrome or DiGeorge syndrome.

This study is limited by its retrospective nature. We were unable to answer the question of whether a strategy of deferring surgical intervention for maturation is beneficial. The patient population is very heterogeneous so that we could not provide much insight into therapeutic strategies that may be advantageous for a given clinical situation. The short follow-up of the study did not allow a good analysis of the long-term mortality. Finally, this study did not include less than 2.5 Kg patients admitted for surgical congenital heart disease who did not undergo surgery (n=4). These 4 patients were not operated on because of a contra-indication for cardiopulmonary bypass (severe prematrurity-related morbidity) (n=2) or because they died while the surgery was already scheduled (n=2).

In summary, although a weight of <2.5kg at surgery remains a risk factor for early mortality, a dedicated neonatal cardiac program yields excellent outcomes for low weight neonates or infants independently of STAT risk categories, uni/biventricular pathway and surgical timing. Such good outcomes also extended into the STAT5 risk category and the univentricular pathway. The birthweight, the presence of extracardiac malformations, a genetic syndrome, the STAT category, the surgical strategy, the timing of surgery and the surgeon were not associated with an increased risk of hospital mortality. Lower gestational age at birth was a major independent risk factor for early mortality.