The Risks of Being Tiny: The Added Risk of Low Weight for Neonates Undergoing Congenital Heart Surgery

Abstract

The aims of this study were 1) to describe the additive risk of performing cardiac surgery in neonates born ≤ 2.0 kg, after accounting for the baseline risks of low birth weight, and 2) to describe the additive risk of being born ≤ 2.0 kg in neonates undergoing cardiac surgery. Risk difference analysis in a retrospective cohort, 2006–2016. Neonates born ≤2.0 kg undergoing congenital heart surgery during initial postnatal admission were included. Data were standardized alternatingly for birth weight and cardiac surgical risk using national population data to estimate the number of deaths expected had they not required cardiac surgery or were they of normal weight. Of 105 neonates ≤2 kg, median birth weight was 1.6 kg (IQR 1.3–1.8 kg). Median gestational age was 33 weeks (IQR 31–35 weeks). Observed operative mortality was 14.3%; 0% for neonates ≤1.0 kg (CI 0–33.6%), 20.6% for neonates >1.0–1.5 kg (CI 8.7–37.9%), and 12.9% for neonates >1.5–2.0 kg (CI 5.7–23.9%). Among neonates ≤2.0 kg not undergoing cardiac surgery, expected mortality was 4.8% (CI 1.6–10.8); cardiac surgery increased the risk of mortality 9.5% (CI 1.7–17.4%). Conversely, the expected risk for normal birth weight neonates undergoing cardiac surgery was 5.7% (CI 2.1–12.0%); low birth weight increased the risk of mortality 8.6% (CI 0.5–16.6%). To continue making advancements in cardiac surgery, we must understand that the rate of mortality observed in normal weight infants is not a realistic target and that, despite advances, the risk attributable to the surgery remains higher among low birth weight patients.

Introduction

Over the last two decades, multiple centers have published their outcomes for congenital heart surgery in low birth-weight infants.[1–5] Initially, investigators focused on all infants born at ≤ 2.5 kg, with center-level mortality reported between 7% and 25%—with variation between studies in operative case-mix complexity, patient comorbidities, and providers. As a general theme, these papers highlighted their impressive results and argued that newer, lower weight thresholds for operability should be considered, as surgeons gather proficiency, technical mastery and experience in performing complex procedures on extremely small infants. Recent data have suggested that the relationship between birth weight and survival following cardiac surgery is non-linear. Hickey and colleagues suggested there might be an inflection point at 2.0 kg, below which they observed marked decline in survival.[6] Increasingly, surgeons have moved towards 2.0 kg as the new definition of “low birth weight” for cardiac surgery. Despite this shift, mortality rates have remained higher for low birth weight infants than for normal weight infants, even after considering the effects of a range of confounding factors.[1–6] This has sparked ongoing debates about “how small is too small” and the merits (or risks) of waiting for low birth-weight infants to grow before operating on their hearts.

Low birth weight is, however, a known risk factor for death even in the absence of congenital heart disease.[7] To our knowledge, no study has considered this as a confounder when assessing operative mortality. Instead, studies have included all postoperative deaths occurring pre-discharge and at <30 days if discharged before 30 days as operative deaths, as is standard in the field for infants >2.5 kg.[8] We hypothesized that ignoring the baseline risk attributable to low birth weight itself overestimates the risk specifically attributable to the cardiac surgical procedure in low birth weight infants, which could obscure critical decisions such as cardiac surgical timing. In this study, we sought to employ epidemiologic modeling to 1) to describe the additive risk of performing cardiac surgery in neonates born ≤ 2.0 kg, after accounting for the baseline risks of low birth weight, and 2) to describe the additive risk of being born ≤ 2.0 kg in neonates undergoing cardiac surgery.

Patients and Methods

Data Source

We conducted a retrospective cohort study at Columbia University Irving Medical Center (CUIMC). Data were obtained via manual chart review, including children operated on between January 1, 2006 and December 31, 2016.

Expected mortality estimates were calculated using population-based data over the same time period, derived from The Vermont Oxford Network (VON) Database of Very Low Birth Weight Infants, The VON Expanded Database (https://public.vtoxford.org, Burlington, VT: Vermont Oxford Network. Nightingale Internet Reporting System, accessed on June 10, 2019), and The Society of Thoracic Surgeons-Congenital Heart Surgery (STS-CHS) Database (http://www.sts.org, Chicago, IL: The Society of Thoracic Surgeons. Private communications. August 16, 2019). The VON manages international clinical registries for live born infants in the general population admitted to a participating hospital within the first 28 days of life without having first gone home. The Very Low Birth Weight and Expanded Databases include data on infants >401 grams from 749 neonatal intensive care units in the United States (U.S.). The incidence of congenital heart defects in low birth weight infants requiring either medical or surgical intervention prior to discharge is <1%.[9,10] The STS-CHS Database is the largest congenital heart surgery registry in the world. It includes data on more than 360,000 surgeries conducted at 127 centers in North America. It captures data on approximately 98% of all congenital heart operations in children in the U.S. Children < 2.0 kg comprise approximately 4% of the neonates in this cohort (http://www.sts.org, Chicago, IL: The Society of Thoracic Surgeons. Private communications. August 16, 2019).

Study Population

We included all neonates with birth weights ≤2.0 kg undergoing congenital heart surgery at CUIMC, January 1, 2006 and December 31, 2016. Patients who underwent isolated closure of the ductus arteriosus were excluded. Neonatal heart surgery was defined in this study as any surgery that took place during the initial postnatal admission. These patients were managed pre- and postoperatively in our dedicated cardiac neonatal intensive care unit. During the study period, it was institutional practice to allow extremely low birth infants to grow prior to surgery, if they were hemodynamically stable and if failure to operate was not impeding care progression. In such patients, it was typically institutional practice to wait until they reached 2.0 to 2.5kg. If patients were hemodynamically unstable secondary to their cardiac disease or if failure to operate were otherwise impeding care, institutional practice was typically to operate, regardless of weight, and to perform complete repairs, rather than palliative repairs, when possible.

Data from the VON Databases were included for all neonates, 2006 through 2016, including a small percentage with cardiac disease that might require intervention within the first year of life. In the National Birth Defects Registry, this is estimated to be, however only ~0.2%.[11]

In primary analyses, data from the STS-CHS Database published reports were included only from 2009 through 2016, as the STS-CHS Database adopted its current definition of operative mortality in 2009; prior to 2009, mortality data were not captured post-discharge. In sensitivity analyses, data were included on all patients, 2006 through 2016, using in-hospital mortality as the primary end-point between 2006 through 2008 and operative mortality between 2009 through 2016. The STS data utilized included a small percentage of low birth weight neonates (~4%).

Data Collection

Local data were collected from electronic medical records on patient demographics, preoperative, operative, and postoperative characteristics.

The VON Database was used to abstract the number of patients and the national average in-hospital mortality rates and confidence intervals by birth weight.

STS-CHS Database published reports were used to abstract the number of neonates undergoing cardiac surgery and the national average operative mortality rates by surgical risk category.

Outcomes

The primary outcome was operative mortality, defined as any death occurring before hospital discharge or within 30 postoperative days in patients discharged prior to 30 days. In sensitivity analyses, in-hospital mortality was also examined.

Statistical Analyses

Expected Mortality Rates

Patient and operative characteristics were described using standard summary statistics. Indirect standardization was used to calculate expected mortality rates and 95% confidence intervals (CIs), using the VON (standardized for birth weight) and STS-CHS (standardized for surgical risk category) databases as separate standard populations. The expected mortality rate standardized for birth weight represents the proportion of patients in our cohort that would be expected to die based on their birth weights alone, in the absence of congenital heart surgery. The expected mortality rate standardized for surgical risk category represents the proportion of patients in our cohort that would be expected to die based on the complexity of their congenital heart surgery, were they of appropriate birth weight. Expected mortality rates, R_ST, were calculated according to the following equation:

$${\text{R}}_{\text{ST}}=\frac{{\sum }_i{R}_i{N}_i}{{\sum }_i{N}_i},$$

where R is the expected mortality for each weight or surgical complexity category, i, and N is the number of patients in our cohort observed in each of the i categories. As an example using fictional data, if the observed mortality rate in one reference population weight category (i.e. VON, 1.5–2.0 kg) is 20%, the number of deaths expected in a sample population of 53 patients would be calculated as 0.2 multiplied by 53, equaling 11 patients. The expected death rate in this category would then be calculated as 11 divided by 53, equaling 21%. The overall expected mortality rate could be calculated as the sum of the expected deaths across all categories divided by the total number of patients. Weight categories were defined by 500 g increments (500–1000 g, 1001–1500 g, and 1501–2000 g). Surgical complexity categories were defined by The Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery (STAT) Mortality Risk Categories, empirically derived measures of operative mortality based on congenital heart surgical procedure type.[12]

Risk Differences

Risk differences were calculated by subtracting expected (R_ST) from observed mortality rates in our cohort. The risk difference standardized for birth weight represents the extra risk of mortality attributable to congenital heart surgery, given the birth weight distribution in our cohort, for infants born at ≤2.0 kg. The risk difference standardized for STAT category represents the extra risk of mortality attributable to low birth weight, given the surgical complexity in our cohort, when performing congenital heart surgery. Stratified risk differences standardized for birth weight were analyzed by weight category for patients >1.0 kg only, given the small numbers of patients <1 kg and within each surgical complexity category. Confidence intervals (CIs) were calculated using the exact binomial method.

Analyses were conducted using Stata (version 14.1; StataCorp, College Station, Texas, USA). The CUIMC Institutional Review Board approved the study with waiver of informed consent.

Results

Patient Characteristics

Over the 10-year study period, 118 neonates ≤2.0 kg with hemodynamically significant congenital heart disease were admitted to the Infant Cardiac Intensive Care Unit at Columbia University Irving Medical Center. Comfort care was pursued in 6 neonates with Trisomy 13 or 18. Five neonates were discharged after trans-catheter balloon valvuloplasties without surgical intervention. In total, 107 neonates with birth weights ≤2.0 were determined to require cardiac surgery prior to discharge and were included in our analyses. Two neonates died while awaiting surgical intervention. One death occurred as the result of fulminant necrotizing enterocolitis in a 20-day old 1.0 kg infant with pulmonary atresia / intact ventricular septum. The second death occurred as the result of septic shock in a 6-month-old former 29-week gestation infant (birth weight 1.3 kg), with a ventricular septal defect, aortic coarctation, chronic lung disease, and pulmonary hypertension.

In total, 105 neonates with birth weights ≤ 2.0 kg underwent cardiac surgery at our institution, 2006 to 2016. There was a median of 8 surgeries per year (IQR 7–12). Patient characteristics are described in Table 1. In brief, 8.6% of infants (n=9) weighed 0.5–1.0 kg; 32.4% (n=34) weighed 1.0–1.5 kg, and 59.0% (n=62) weighed 1.5–2.0 kg. The median gestational age at birth was 33 weeks (IQR 31–35 weeks; range 25–40). Birth weights were similar across STAT categories. Slightly more than half of infants were prenatally diagnosed (57%; n=60). The median age at surgery was 38 days (IQR 19–68; range 3–159 days) for all infants. Age at surgery was inversely correlated with birth weight; age at surgery was 62 days (IQR 38–80) for infants ≤1.0 kg, 50 days (IQR 31–80) for infants 1.0–1.5 kg, and 33 days (IQR 13–49) for infants 1.5–2.0 kg (R=−0.44, p=<0.01). Median gestational age at surgery was 39 weeks (IQR 36–41; range 30–53). There was no significant difference in gestational age by weight category; median gestational age was 38 weeks (IQR 32–40) for infants ≤1.0 kg, 39 weeks (IQR 36–42) for infants 1.0–1.5 kg, and 39 weeks (IQR 37–41) for infants 1.5–2.0 kg. The median weight at the time of surgery was 2.1 kg (IQR 1.8–2.5 kg; range 0.8–4.3 kg). The lowest birth weight infants had the lowest weights at surgery; 1.7 kg (IQR 1.0–2.1 kg) for infants born at ≤1.0 kg, 2.2 kg (IQR 1.6–2.9 kg) for infants born at 1.0–1.5 kg, and 2.1 kg (IQR 1.9–2.5 kg) for infants born at 1.5–2.0 kg (R=0.10, p=0.32). Table 2 details the operative procedures and the numbers of neonates undergoing corrective versus palliative repairs.

Table 1.

Patient and operative characteristics

Variable	Median (IQR) (N = 105)
Male sex	55 (52.4)
Birth weight
All patients, kg	1.6 (1.3–1.8)
STAT category 1, kg (n=21)	1.6 (1.3–1.9)
STAT category 2, kg (n=25)	1.5 (1.1–1.7)
STAT category 3, kg (n=10)	1.7 (1.3–1.8)
STAT category 4, kg (n=45)	1.7 (1.5–1.8)
STAT category 5, kg (n=4)	1.4 (1.2–1.6)
Gestational age, weeks	33 (31–35)
Small for gestational age, n (%)	56 (53.3)
Prenatal diagnosis, n (%)	60 (57.1)
Genetic syndrome, n (%)	18 (17.1)
Major extra-cardiac malformation, n (%)	27 (25.7)
Weight at surgery, kg	2.1 (1.8–2.5)
Age at surgery, days	38.0 (19.0–68.0)
Corrected gestational age at surgery, weeks	39.0 (36.4–41.0)
Preoperative ventilation, n (%)	57 (54.3)
Cardiopulmonary bypass, n (%)
≤1.0 kg	6 (66.7)
1.0–1.5 kg	29 (85.3)
1.5–2.0 kg	52 (81.3)

Data are number (%) or median (IQR).

IQR interquartile range, STAT Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery Mortality Risk

Table 2.

Distribution of cardiac surgeries in neonates <2.0 kg (N=105)

Procedure	n	%
Single Ventricle Repairs
Shunt	9	8.6
Pulmonary Artery Band	5	4.8
Norwood	2	1.9
Biventricular Repairs
TOF Repair, for TOF with Pulmonary Stenosis or Membranous Pulmonary Atresia	22	21.0
Complete Repair	12
Palliative Shunt	10
Coarctation Repair	11	10.5
VSD Repair	20	19.0
VSD	14
VSD with Arch Reconstruction	5
VSD with Vascular Ring Repair	1
Arterial switch	8	7.6
Arterial switch without VSD	4
Arterial switch with VSD Repair	4
Complete Atrioventricular Canal Repair	5	4.8
Truncus Arteriosus Repair	4	3.8
Complete Repair	3
Palliative Shunt	1
Aortopulmonary Window Repair	3	2.9
Pacemaker Placement	3	2.9
TAPVC Repair	3	2.9
Repair of Hypoplastic/Interrupted Aortic Arch without VSD	3	2.9
DORV Repair	2	1.9
Tumor Resection	2	1.9
ALCAPA Repair	1	1.0
Repair of Supravalvar Aortic Stenosis	1	1.0
Unifocalization, TOF with multiple aortopulmonary collaterals	1	1.0

TOF Tetraolgy of Fallot, VSD Ventricular Septal Defect, TAPVC Total Anomalous Pulmonary Venous Connection, DORV Double Outlet Right Ventricle, ALCAPA Anomalous Left Coronary Artery from the Pulmonary Artery.

Operative Mortality

Observed overall operative mortality for infants born at ≤2.0 kg was 14.3% (n = 15 of 105); 0.0% (n = 0) for infants ≤1.0 kg (CI 0–33.6%), 20.6% (n = 7) for infants 1.0–1.5 kg (CI 8.7–37.9%), and 12.9% (n = 8) for infants 1.5–2.0 kg (CI 5.7–23.9%). Table 3 describes the number of deaths observed and death rates that would have been expected, standardized for VON birth weights and STAT procedural risk category.

Table 3:

Observed and expected death rates from the Vermont Oxford Network (VON), Society of Thoracic Surgeons-Congenital Heart Surgery (STS-CHS), and Columbia University Irving Medical Center (CUIMC)

	Observed Cases in the VON, by Weight			Observed Cardiac Surgical Cases at CUIMC, ≤2.0 kg			Expected Death Rate among Low Weight Patients (95% CI)b	Added Risk of Cardiac Surgery (95% CI)c
Weight	Number of Cases	Deaths Observed	Weight-Specific Death Rate	Number of Cases	Deaths Observed	Death Rate Observed (95% CI)
≤1.0 kg	94,542	26,416	27.9%	9	0	0.0% (0.0 to 33.6)	33.3% (7.5 to 70.1)	NId
1.0–1.5 kg	116,637	4,977	4.3%	34	7	20.1% (8.7 to 37.9)	2.9% (0.1 to 15.3)	17.6% (2.9 to 32.4)
1.5–2.0 kg	199,015	3,971	2.0%	62	8	12.9% (5.7 to 23.9)	1.6% (0.0 to 8.7)	11.3% (2.4 to 20.2)
Standardized Total	--	--	--	105	15	14.3% (8.2 to 22.5)	4.8% (1.6 to 10.8)	9.5% (1.7 to 17.4)
	Observed Cases in the STS-CHS, by Case Complexity			Observed Cardiac Surgical Cases at CUIMC, ≤2.0 kg			Expected Death Rate among Cardiac Surgical Patients (95% CI)b	Added Risk of Low Weight (95% CI)c
STAT category	Number of Cases	Deaths Observed	STAT-Specific Death Rate	Number of Cases	Deaths Observed	Death Rate Observed (95% CI)
1	1,425	70	4.9%	21	1	4.8% (0.1 to 23.8)	4.8% (0.1 to 23.8)	0.0% (−12.9 to 12.9)
2	5,177	214	4.1%	25	4	16.0% (4.5 to 36.1)	4.0% (0.1 to 20.4)	12.0% (−4.3 to 28.3)
3	4,715	185	3.9%	10	2	20.0% (2.5 to 55.6)	NId	NId
4	15,364	1,451	9.4%	45	7	15.6% (6.5 to 29.5)	8.8% (2.5 to 21.2)	6.7% (−6.8 to 20.1)
5	7,780	847	10.9%	4	1	25.0% (0.6 to 80.6)	NId	NId
Standardized Total	--	--	--	105	15	14.3% (8.2 to 22.5)	5.7% (2.1 to 12.0)	8.6% (0.5 to 16.6)

VON Vermont Oxford Network, CUIMC NICU Columbia University Irving Medical Center Neonatal Intensive Care Unit, LBW low birth weight, CI confidence interval, STS-CHS Society of Thoracic Surgeons-Congenital Heart Surgery database, STAT Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery Mortality Risk.

^aExpected number of deaths is calculated based on the weight and STAT specific death rates from VON and STS-CHS, respectively.

^bExpected death rate is the standardized rate based on the weight and STAT specific death rates from VON and STS-CHS, respectively.

^cRisk difference between the observed and expected death rates.

^dNI, denotes values for which the estimates within the subpopulations are not interpretable given the small sample size.

Among all infants ≤2.0 kg not undergoing cardiac surgery, the standardized expected mortality rate was 4.8% (CI 1.6–10.8%). Cardiac surgery for these low birth weight neonates was estimated to increase the risk of mortality 9.5% (CI 1.7–17.4%) (Figure 1a). The additional risk of cardiac surgery was similar across weight categories. The differences in risk met statistical significance among infants 1.0–1.5 kg and 1.5–2.0 kg (Figure 1b), though not among the 9 infants ≤1.0 kg.

Fig. 1a.

Observed and expected risk of mortality among neonates ≤2.0 kg undergoing cardiac surgery at Columbia University Irving Medical Center; b Risk of mortality attributable to cardiac surgery among neonates ≤2.0 kg. This represents the difference between the observed mortality in neonates <2.0 kg undergoing surgery and the mortality that would have been expected given their birth weight alone if they had not undergone surgery. The expected mortality is the rate expected in the Columbia cohort based on the weight-specific death rates observed in low birth weight neonates in the Vermont Oxford Network (not cardiac specific neonates).

Overall, the expected risk of mortality for neonates of normal birth weight undergoing congenital heart surgery was 5.7% (CI 2.1–12.0). Low birth weight, for these neonates was estimated to increase the risk of mortality 8.6% (CI 0.5–16.6). Sensitivity analyses focused on in-hospital (instead of operative) mortality had minimal effect on our results.

Discussion

In this retrospective analysis of low birth weight neonates undergoing congenital heart surgery, we isolated the risk associated with performing congenital heart surgery in infants born at ≤2.0 kg, after accounting for the baseline risks attributable to birth weight, and the risk of being born at ≤2.0 kg, among infants undergoing cardiac surgeries. We found that performing congenital heart surgery in infants born at ≤2.0 kg at our institution was associated with approximately a 10% average increase in operative mortality, when compared to similar-weight infants not requiring congenital heart surgery. We found this to be roughly double (5 percentage points higher than) the mortality that would have been expected following congenital heart surgery, were these patients of average birth weight. Approximately half of the total risk of mortality (8.6 percentage points) in this low weight surgical cohort could be explained by the infants’ birth weight alone and may be independent of their surgery.

Low weight infants are at a higher risk of mortality, even in the absence of heart disease.[13–17] These infants have high rates of respiratory failure, bronchopulmonary dysplasia, predisposition to intestinal ischemia and necrotizing enterocolitis, limited tubular and medullary renal function, and immature immune systems.[18] The Vermont Oxford Network has documented consistent decline in mortality for low birth weight infants over the last several decades.[13–17] In the current era, on average, infants born in the United States weighing 1.5–2.0 kg have a >98% chance of survival—>90% for born weighing 0.5–1.5 kg—but that still represents absolute risks of 2% and 10% respectively. The Vermont Oxford Network has also further shown significant and persistent variation in center-level outcomes for low weight infants. Nationally, there remains a two-percentage point absolute difference in mortality for neonates 0.5 to 1.5 kg across neonatal intensive care units (20% relative risk difference).[13] These effects appear to be even more profound for infants at the lower end of the weight and gestational age spectrum.[16,19]

In 1976, the March of Dimes called for a tiered system for neonatal intensive care units, to address variation in outcomes. In 2004, the American Academy of Pediatrics Committee on Fetus and Newborn issued a policy statement establishing a system based on the capabilities available at each center.[20] In the years since, data have repeatedly confirmed that infrastructure and personnel in neonatal intensive care units explains variation in outcomes and that adherence to this capacity-based, tiered system improves neonatal outcomes at a national level.[21,22]

A similar debate has taken place in congenital heart community over the last several decades, accelerated in recent years by a series of lay-press articles.[23–25] Many have called for regionalization of care by center-level volumes.[26–32] Others have maintained that volumes explain only a small fraction of center-level variation in cardiac surgical outcomes and have pointed to some smaller cardiac centers that have exceptionally good outcomes and some larger centers with poorer outcomes.[33,34]

As some of the observed risk can be attributed to being of low birth weight, this aspect of care should be considered when evaluating between-center variation in outcomes. Our data raise the possibility that some of the observed variation in cardiac surgical outcomes for low weight infants may be the result of centers’ capacity to care for low birth weight children at large, rather than the surgeons’ technical capacity to perform the operation on physically small hearts or the intensivists’ familiarity with managing particular cardiac conditions. Our data are in-line with previous studies on gestational age, which have pointed to a parallel relationship between gestational age and mortality among neonates with congenital heart defects and the neonatal population at large.[35,36] As cardiologists and surgeons consider the wisdom of offering surgical repair on increasingly smaller children at their local institutions, this ought to be considered.

Unfortunately, our data cannot answer questions regarding optimal surgical timing, types of surgery with lower risk, or the impact of other factors such as genetic or extra-cardiac anomalies for low weight infants. We do not have data on the morbidities these children experienced while awaiting surgery nor on the reasons for surgical delays. and our power to perform such calculations is limited. That having been said, our data do not suggest that children operated on at lower weights had increased additive risks related to cardiac surgery, as the majority of the difference in observed mortality between infants greater or less than 1.5 kg would be appropriately attributable to differences in expected mortality by birth-weight alone, rather than to differences in children’s responses to cardiac surgery. Future investigations on surgical timing ought to include consideration of the additive effects of being born small.

Due to the low number of infants in the group <1.0 kg, we were unfortunately unable to provide reliable evidence for the risk difference in infants born at the lowest weight. Intuitively, we would expect the <1.0kg cohort to have the highest mortality risk among the three weight groups; we know from the VON database nationally this group has a baseline 30% risk of mortality before surgery. None of our nine children <1.0 kg died. These patients were managed pre- and postoperatively in our dedicated cardiac neonatal intensive care unit. Our study is further limited by the fact that the VON data do include some children with cardiac defects and the STS-CHS data do include some infants ≤2.0 kg. In both instances, these represent small fractions of cases in the reference databases and we do not believe this to be large enough to meaningfully increase the calculated mortality rates. Additionally, any potential bias due to the inclusion of these infants would likely lead to an underestimation of the risk difference rather than overestimation.

The VON database also reports in-hospital—rather than in-hospital or 30 day—mortality. There may be some infants in the VON ≤2.0 kg and discharged before 30 days who died at home, and there may be infants in VON who died early and would not have reached surgery if they had been cardiac surgery patients. That having been said, sensitivity analyses using in-hospital—rather than operative—mortality yielded the exact same results, as all deaths in our cohort occurred in hospital and no children born at <2.0 kg were discharged alive at <30 days.

Conclusion

As surgeons advance the limits of complex cardiac surgeries in low weight neonates, and as we evaluate them for doing so, it is helpful to understand that a) the rate of mortality observed in the general neonatal cardiac surgical population is not a realistic target and b) the risk attributable to the surgery remains higher among low birth weight patients. As surgeons advance the limits of complex cardiac surgeries in low weight neonates, and as we evaluate them for doing so, it is helpful to understand that a) the rate of mortality observed in the general neonatal cardiac surgical population is not a realistic target for <2.0kg infants and b) the risk attributable to the surgery remains higher among low birth weight patients.