Managing cyanotic neonates with tetralogy of Fallot: A national perspective on surgical approaches

Abstract

Objective

Management of cyanotic neonates with tetralogy of Fallot (ToF) remains an important clinical challenge without Level 1 evidence. The present study employs the Healthcare Cost and Utilization Project Kids Inpatient Dataset to evaluate a national sample of neonates undergoing intervention for ToF.

Methods

A 4-stage algorithm was designed to capture records for neonates with ToF who underwent primary repair, ductal stents, and systemic-to-pulmonary shunts in the Healthcare Cost and Utilization Project Kids Inpatient Dataset (2016, 2019, and 2022). National hospitalization estimates and percent reported reflect weighted results based on survey design. Resource utilization was represented by median hospital length of stay and inflation-adjusted cost in 2023 US dollars.

Results

An estimated 159.5 discharges were identified for primary repair, 145.3 for ductal stents, and 407.4 for surgical systemic to pulmonary shunts. An estimated 256.3 discharges were identified for infants undergoing definitive repairs following ductal stents. From 2016 to 2022, ductal stent utilization increased (linear trend P = .0011), whereas surgical shunt use decreased (trend P = .0012). There was a trend toward a decrease in primary repair (trend P= .12). Over this time frame, there was a significant increase in the median cost for ductal stent (difference, $77,252 [2023 dollars], P = .035) and surgical shunt (difference, $81,111 [2023 dollars]; P = .043) palliations and a decrease in primary repair cost (difference, $76,337 [2023 dollars]; P = .44).

Conclusions

This investigation demonstrated increased utilization of ductal stents for neonatal ToF across the United States. Despite this increase, changes in cost may reflect a complex paradigm shift in patient/center-specific decision making.

Graphical Abstract

Trends in strategy and total hospital cost for neonates with ToF.

Central Message.

Cyanotic neonates with ToF are increasingly treated with ductal stents rather than surgical strategies. Simultaneously, palliation costs have increased while primary repair costs have declined.

Perspective.

This national perspective suggests that the increased utilization of ductal stents has changed preferred pathways for cyanotic neonates with tetralogy of Fallot. Centers may now be triaging higher-risk neonates to palliation, potentially reflecting salutary benefits for primary repair.

Management of cyanotic neonates with tetralogy of Fallot (ToF) remains a clinical challenge without clear guidelines or consensus.¹ Current strategies include neonatal primary repair and staged approaches with either a ductal stent (DS) or surgical shunt (SS). Historically, comparisons were made evaluating primary repair to both palliation strategies with single-center results,² multicenter clinical data,3, 4, 5 administrative datasets,6, 7, 8, 9, 10 and public reporting registries.¹¹ However, results are inconsistent. Steiner and colleagues,⁸ using the pediatric health information system database (2004-2011), found no difference in postoperative mortality, yet a greater cumulative postoperative morbidity and resource utilization for the staged pathway. Conversely, Savla and colleagues,¹⁰ using pediatric health information system 2004-2015 data, found primary repair to be associated with an increased risk of 2-year mortality. In recent years, there has been a dramatic increase in the selection of DS palliation for cyanotic neonates with ductal-dependent pulmonary flow,¹² the subject of the ongoing Comparison of Methods of Pulmonary Blood Flow Augmentation in Neonates: Shunt versus Stent (COMPASS) trial.¹³ To date, no group has characterized how the emerging popularity of DS has influenced care specifically for neonates with ToF and strategy preference.

Our team has established a decision-analytic framework to study this scenario from a health care economic perspective. Our first study utilized inputs from published studies and found that at a 2-year horizon, from a single-patient clinical perspective, a primary repair was preferred over palliative stents (ductal and right ventricular outflow tract) or SS insofar as health care cost and quality-adjusted life years gained.¹⁴ Further, sensitivity analysis found health care costs to be a major factor influencing incremental cost-effectiveness ratios. Our statistical team has extensive experience working with Healthcare Cost and Utilization Project (HCUP) datasets.15, 16, 17 As such, we sought to use this resource to understand the cost implications of these 3 strategies on a national level and whether this would change the results of our prior modeling. The major aims of the present study are to explore an algorithm to study neonatal ToF interventions in the HCUP-Kids Inpatient Database (HCUP-KID); to report national inpatient discharge estimates; and to explore, on a national level, how the paradigm shift has changed cumulative hospital costs.

Methods

A retrospective cohort study was designed using the HCUP-KID. A description of the advantages of HCUP-KID and our approach to the population with ToF can be found in the Appendix E1. The use of this limited dataset was approved by the Cleveland Clinic Institutional Review Board (IRB# 15-139; approved February 3, 2015). Patient consent waived. A 4-stage algorithm was designed to capture patients with ToF as neonates undergoing primary repair, DS, or SS, and ToF infants undergoing complete repair following palliation (Figure 1, Table E1). These are detailed in Appendix E1.

Figure 1.

Schematic of the Healthcare Cost and Utilization Project-Kids Inpatient Dataset algorithm designed to define patients with tetralogy of Fallot (ToF) undergoing neonatal interventions. VSD, Ventricular septal defect; PDA, patent ductus arteriosus.

Study Measures

Descriptive elements, including low birth weight, preterm birth, genetic anomalies, and cardiovascular anomalies, were identified using International Classification of Diseases-10th Revision (ICD-10) codes. Patient family residence was characterized as large metropolitan (≥1 million residents/county), small (50,000 to 999,999 residents/county) metropolitan, or small micropolitan or non-core (<50,000 residents/county) using HCUP-KID residence classification. Location of hospital and payer status were also defined using HCUP-KID elements (Table E2). Outcomes were death, use of extracorporeal membrane oxygenation, prolonged ventilation (>96 hours), and hospital length of stay (LoS) (Tables E3 and E4). We used the procedure day from admission to determine when the repair or palliation was performed. Cost estimates were derived from total charges; HCUP cost-to-charge ratio tables were adjusted to 2023 inflation estimates using the mid-year Consumer Price Index for Medical care.¹⁸

Statistical Analysis

To exceed reportable thresholds per HCUP's data use agreements (n ≥ 11), the 3 years were combined and weighted appropriately. Hospitalization estimates n, weighted, and percent (SE) reported reflect weighted results based on survey design. Rao-Scott likelihood ratio χ² tests were performed to determine statistical significance between groups. Comparison of costs and LoS across years used sampling weight-adjusted regression of log-transformed outcomes. Pairwise comparisons were adjusted using Tukey method for multiple comparisons, and adjusted significance was reported. To test overall changes in surgical strategies, logistic regressions were used to test for overall differences (Wald test) in odds ratios (ORs), followed by orthogonal polynomial contrasts for linear trends across years. Sampling weighted adjusted results are presented unless noted.

Results

We identified an estimated 2196, 2216, and 2267 discharges in 2016, 2019, and 2022, respectively, of neonates with a qualifying ToF diagnosis (Box 3 in Figure 1). From these records, we identified national inpatient discharge estimates of cyanotic neonates requiring surgery. We captured a nationally weighted estimated 159.5 discharges undergoing primary repair (Box 5 in Figure 1), 145.3 discharges with DS (Box 6 in Figure 1), and 407.4 discharges undergoing SS (Box 7 in Figure 1). We identified 256.3 discharges as having definitive repairs following DS (Box 8 in Figure 1 and Table 1). Patients undergoing definitive repair following a SS palliation could not be reliably identified and were excluded.

Table 1.

Characteristics of patients captured in Healthcare Cost and Utilization Project (HCUP)-Kids Inpatient Dataset 2016, 2019, and 2022^∗

Total	Primary repair	Ductal stent	Definitive repair post ductal stent	Surgical shunt
	n = 116 (Est. visits = 159.5)	n = 107 (Est. visits = 145.3)	n = 186 (Est. visits = 256.3)	n = 296 (Est. visits = 407.4)
Patient clinical characteristics
Low birth weight (<2.5 kg)	14.4 ± 3.4	18.9 ± 4.1	<6%	13.7 ± 1.8
Preterm birth (<37 wk)	16.3 ± 3.4	22.5 ± 4.2	5.9 ± 1.7	17.1 ± 2.2
Genetic anomaly	22.4 ± 4.0	15.0 ± 3.8	17.3 ± 2.9	15.7 ± 2.2
Pulmonary valve/artery atresia	42.0 ± 4.8	52.1 ± 5.2	10.2 ± 2.2	43.7 ± 3.3
Right aortic arch	0	0	0	0
Tracheal airway abnormalities	9.4 ± 2.6	<10%	<6%	<5%
Patient urban/rural residence
Large metro	55.7 ± 5.1	47.5 ± 5.8	59.1 ± 4.1	52.4 ± 3.7
Small metro	31.6 ± 4.8	33.8 ± 5.2	33.4 ± 3.7	34.5 ± 3.4
Micro/noncore	12.7 ± 3.1	18.7 ± 4.2	7.5 ± 1.9	13.2 ± 2.1
Location/teaching status of hospital
Urban teaching	>90%	100.0 ± 0.00	>94%	>95%
Payer status
Private insurance	<45%	42.2 ± 5.9	45.6 ± 3.7	42.1 ± 3.1
Medicaid	49.1 ± 5.6	44.6 ± 6.4	43.0 ± 3.8	52.8 ± 3.2
Self-pay or other	<10%	13.2 ± 3.9	11.4 ± 2.4	5.1 ± 1.4

Values are presented as % ± SE unless otherwise noted. Est., Estimate.

^∗Data suppression applied per HCUP data users agreement.

Select characteristics depicting patients’ co-morbidity and social determinants of health were defined using ICD-10 diagnosis codes and HCUP-KID data elements (Tables E1 and E2). The proportion of neonates with low birthweight (<2.5 kg), pulmonary artery/valve atresia, tracheal anomalies, preterm birth (<37 weeks), or genetic anomalies are reported in Table 1. Overall, patient characteristics were similar, yet due to survey design, comparative statistics were not feasible. Death and significant morbidity were rare (Table 2). Primary repair had the longest LoS and highest inflation-adjusted cost of hospitalization at the initial admission (46.9 days, $342,504), followed by SS (35.1 days, $208,358), DS (21.1 days, $132,259), and definitive repair following DS (7.4 days, $82,788) (Table 2).

Table 2.

Outcomes, hospital length of stay, and resource utilization derived from Healthcare Cost and Utilization Project (HCUP)-Kids Inpatient

Total	Primary repair	Ductal stent	Definitive repair following ductal stent	Surgical shunt
	n = 116 Est. visits = 159.5	n = 107 Est. visits = 145.3	n = 186 Est. visits = 256.3	n = 296 Est. visits = 407.4
Death and significant morbidity for each strategy
Died during hospitalization	<10%	<8%	<6%	5.4 ± 1.2
ECMO	<10%	<8%	<6%	8.1 ± 1.5
Prolonged vent >96 h	<10%	<8%	<6%	4.9 ± 1.3
Total resource utilization of each strategy
Length of stay	46.9 (22.9-88.8)	21.1 (13.4-47.5)	7.4 (5.0-13.8)	35.1 (20.5-58.9)
Total cost adjusted to 2023 US dollars	342,504 (147,799-666,220)	132,259 (82,524-278,881)	82,788 (60,581-131,322)	208,358 (132,801-361,213)

Values are presented as % ± SE or median [Q1, Q3] unless otherwise noted. Statistics are weighted. Data suppression applied per HCUP data users agreement. SAS Survey Procedures were used for all analyses. Est., Estimate; ECMO, extracorporeal membrane oxygenation.

From 2016 to 2022, there were substantial changes in strategy utilization. For palliative approaches, the estimated number of DS discharges increased (n = 26.7 in 2016 to n = 77.1 in 2022), and the estimated number of SS decreased (n = 184.9 in 2016 to n = 103.5 in 2022). Moreover, neonates admitted for ToF were almost 3 times more likely to have had a DS in 2022 versus 2016 (OR, 2.86; 95% CI, 1.52-5.39). Although there was not a significant difference in likelihood of a DS in 2019 versus 2016 (OR, 0.55; 95% CI, 0.78-3.09), a linear trend was observed (P = .0011). The estimated number of SS discharges decreased from 184.9 in 2016 to 119.0 in 2019 and 103.5 in 2022. Neonates with ToF in 2022 were far less likely to receive SS compared with 2016 or 2019 (linear trend P = .0012) (Figure 2). The sampling weight-adjusted number of discharges for neonates with ToF undergoing early primary repair was 59.5 in 2016, 61.9 in 2019, and 38.1 in 2022. The odds of undergoing primary repair in 2022 versus 2016 were not statistically significant (linear trend P = .12) (Figure 2).

Figure 2.

Odds of neonatal tetralogy of Fallot intervention in Healthcare Cost and Utilization Project-Kids Inpatient Dataset since 2016.

With changes in strategy across the 3 years, there were changes in median hospital LoS and inflation-adjusted total cost of hospitalization. From 2016 to 2022, median LoS and cost for primary repair decreased by 8.6 days (P = .67) and $76,337 2023 US dollars (21%), respectively. This difference was primarily between 2019 and 2022, but assessing a linear trend, these differences in cost were not statistically significant (P = .44, overall). For DS, median total cost increased significantly (difference in medians, $77,252 2023 US dollars, overall P = .043, pairwise P = .035), paralleled by an observed 7.4-day increase in median LoS (P = .46). The total cost for SS also increased significantly (difference in medians, $81,111 2023 US dollars, overall P = .0446, pairwise P = .043), with a 12.5-day increase in hospital LoS (P = .058). Between 2016 and 2022, there was no significant change in cost for definitive repair following DS (difference in medians, $11,584 2023 US dollars [P = .65] and .4 days [P = .80]) (Table 3). Further inquiry into ductal stent and SS hospitalizations found that these palliative strategies were consistently performed early in the hospitalization (Table E5). This assessment demonstrates a significant increase in resource utilization for palliative strategies.

Table 3.

Model-derived median values depicting trends in hospital length of stay (LoS) and cost between strategies, by year

Variable	2016	2019	2022
Median hospital LoS (d)
Early primary repair	54.3	48.3	45.7
Ductal stent	21.7	25.6	29.1
Definitive repair following ductal stent	11.4	10.4	11.1
Surgical shunt	34.2	37.6	46.7
Median cost (2023 US dollars)
Early primary repair	362,869	360,723	286,532
Ductal stent	97,051∗	163,158	174,303∗
Definitive repair following ductal stent	91,394	96,162	102,978
Surgical shunt	208,414†	213,481	289,525†

Statistical significance derived by Tukey adjusted P values.

^∗P = .0347 for the 2016 versus 2022 comparison.

^†P = .0429 for the 2016 versus 2022 comparison.

Discussion

The present study focuses on the evolving practice patterns on a national level and investigates changes in cumulative cost for these interventions over 6 years. We show that:

• •HCUP-KID with the ICD-10 can be used to define a national dataset of cyanotic ToF neonates requiring surgery (See Appendix E1).
• •From 2016 to 2022, DS significantly increased in utilization while surgical palliation decreased. We also observed a trend toward a decline in primary repair.
• •There was an observed reduction in cost for primary repair and a significant increase in the cost of palliative strategies. These results could suggest a complex paradigm shift in the neonatal ToF paradigm.

This work additionally builds on our prior work, where we used decision analytic modeling to create a cost-effectiveness comparison for cyanotic neonates with ToF. In our initial investigation, we implemented inflation-adjusted cost values from O'Byrne and colleagues.³ Initially, we found that at baseline, primary repair was the dominant strategy; however, sensitivity analysis revealed that as the cumulative cost of DS decreased (<$134,581 2023 US dollars) or the primary repair cost increased (>213,335 2023 US dollars), the most cost-effective strategy would become DS.¹⁹ These results demonstrate that the cost of primary repair exceeds the cost threshold from our prior sensitivity analysis and therefore may be less cost-effective than ductal stents. This corresponds with an observed shift in management strategies for neonates with ToF. However, because the ductal stent pathway has been selected more frequently, its cumulative costs have also increased.

Paradigm Shift With Ductal Stents

Overall, our results suggest that DS are being more frequently selected over other surgical strategies. However, they also infer that between 2016 and 2019, DS utilization may not have had an impact on primary repair preference. Then, between 2019 and 2022, DS may have been selected more frequently over primary repair. Given the limitations of a cross-sectional administrative dataset, it is challenging to delineate on a national scale why this shift is occurring.

There are many potential benefits to a ductal stent that may influence decision making, as well as several persistent benefits for primary repair. Ductal stents avoid a median sternotomy, cardiopulmonary bypass, and have a shorter cumulative hospital LoS. Centers that routinely perform DS may also be comfortable with routine interstage catheterization studies and balloon expansion of the stent, as neointimal proliferation occurs. This strategy has been advocated by the Congenital Cardiac Research Collaborative as a way to reduce the risk of interstage cyanosis.²⁰ Finally, DS for ToF are currently not in the Society of Thoracic Surgeons Congenital Heart Surgery Database STAT mortality system; therefore, this trend may reflect an evolving risk diversion. Alternatively, primary repair may still be an important strategy for centers that are not facile with ductal stent technology, for neonates with a severely tortuous ductus, and those whose families reside in a remote or rural location where it may be challenging to complete robust interstage monitoring. With the change in national practice patterns and these considerations in mind, for a cyanotic neonate with ToF, the comparison between ductal stents and primary repair may soon become invalid.

Evolving Costs for Neonatal Strategies

Initially, we hypothesized that as the ductal stent learning curve was superseded, the utilization of hospital resources would decrease and be reflected by a decreased total cost and hospital LoS. Surprisingly, because DS were favored, we observed an increased cost for both palliative strategies, and similarly, costs of primary repair decreased, coincident with decreased primary repair utilization. In this limited dataset, such trends could not be explained by differences in rudimentary patient characteristics defined using ICD-10 diagnoses or a longer clinical preamble to the index procedure. Moreover, because HCUP-KID is derived from claims data and costs from a composite charge value, we were unable to determine if these costs were driven by additional transcatheter studies before discharge. Therefore, further investigations into the drivers of cost were not feasible. We speculate that because DSs have been broadly implemented, more centers may have offered this strategy to higher-risk neonates, whereas others may have developed additional ductal monitoring protocols. For instance, in the experience of Glatz and colleageus²⁰ and the Congenital Cardiac Research Collaborative, some centers perform additional interstage catheterization studies to monitor neointimal proliferation and, when necessary, dilate the ductus to optimize pulmonary flow. An adoption of this strategy and additional monitoring strategies may be a driver of LoS and costs. Further investigation into the evolution of monitoring for palliation is necessary to improve resource utilization and care for these high-risk patients.

Limitations

Characterizing death and significant morbidity

Quantification of death and significant morbidity was limited due to the reportable thresholds required by HCUP-KID. Compared with prior studies, Ghimire and colleagues²¹ in an assessment of neonatal TOF primary repair in HCUP-KID 2003-2012, the incidences of death, extracorporeal membrane oxygenation, and respiratory failure were <10%, consistent with our suppressed statistics. In HCUP-KID for a small population of high-risk neonates, quantification of death and significant morbidity is challenging and requires a larger sample that is feasible in 3-years of neonatal ToF.

Identification of definitive repair after palliation

HCUP-KID sample is cross-sectional. Lacking a longitudinal nature, identification of serial hospitalizations was not feasible. As such, we characterized infants receiving definitive repairs who had received a prior ductal stent using procedure codes that would reflect the removal of the stent. We attribute this success to the ICD-10 transition, where codes for ductal stent placement/removal were new and clearly articulated. However, attempts to define definitive repair for patients following SS were unsuccessful. These efforts are discussed at length in Appendix E1.

Limitations of HCUP-KID for patients requiring congenital heart surgery

This is a limited dataset, advertised for promise evaluating patients requiring congenital heart surgery.²² As a nationwide sample of discharges, while focused on 80% of “complicated” pediatric discharges, some cyanotic ToF neonates may have been missed. Future studies into HCUP-KID may benefit from larger, less specific populations of patients requiring congenital heart surgery. Moreover, HCUP-KID is neither able to assess interstage monitoring or reintervention, nor the subsequent interventions. Such considerations are important for future HCUP-KID investigations.

Future investigations into neonatal tetralogy

As is expected with all administrative studies, HCUP-KID hospital discharge estimates are a snapshot of care for these complex patients. Often, for ToF, future interventions are dependent upon the surgeon's ability to preserve the pulmonary valve, while alleviating outflow tract obstruction. The Society of Thoracic Surgeons Congenital Heart Surgery Database demonstrated that nearly 50% of ToF repairs are achieved with a transannular patch (TAP), and although not statistically significant, a greater proportion of neonates require a TAP.¹¹ This being said, the pulmonary valve z score may be more important than age at repair.²³ Neither the use of a TAP nor the z score can be assessed in HCUP-KID. Longitudinal clinical data and future studies leveraging new models or prospective cohorts may be necessary to further examine value-based care for these patients.

Conclusions

The present study builds on our prior work defining value-based care for neonates with cyanosis who require surgery for ToF. Utilization of DS has increased, whereas SS and primary repair decreased. This trend parallels an increase in the resource utilization for palliative strategies and a potential decrease in the resource utilization for primary repair. These findings suggest that centers may be choosing individualized benchmarks, and more investigations are necessary to understand this complex paradigm shift. Future investigations of this neonatal paradigm should utilize more granular clinical data as well as qualitative assessments of the entire care team to understand the changes in practice at the patient level, which may be contributing to the increase in ductal stent use and resource utilization.

Conflict of Interest Statement

The authors reported no conflicts of interest.

The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest.

Appendix E1. Methods and Discussion

Healthcare Cost and Utilization Project-Kids Inpatient Dataset for the Study of Congenital Heart Surgery

Healthcare Cost and Utilization Project-Kids Inpatient Dataset (HCUP-KID) is the largest publicly available pediatric database in the United States derived from discharge billing abstracts, including 48 states and the District of Columbia. It samples 4000 US hospitals prioritizing “complicated” newborns and other pediatric discharges ages 0 to 20 years at admission. Unweighted, each year contains approximately 3 million pediatric discharges and weighted more than 6 million hospitalizations. It is designed to sample 10% of normal newborns and 80% of “other pediatric discharges.” The large sample size enables analysis of “rare conditions (eg, congenital anomalies) as well as, uncommon treatments (eg, cardiac surgery).”^E1 This dataset also features many procedural elements (International Classification of Diseases-10th Edition-Procedure Coding System [ICD-10-PCS] and procedure day since admission) and nonclinical data elements, including payer status and family residence, to enable an assessment of nonclinical patient factors. HCUP-KID was available on a triennial basis, and datasets from years 2016, 2019, and 2022 were utilized to assess the present paradigm.

Four-Stage Algorithm for Capturing Neonatal Tetralogy of Fallot in HCUP-KID

Stage 1 captured patients using ICD-10 diagnosis codes with either a primary tetralogy of Fallot (ToF) diagnosis or associated codes (Figure 1, Table E1). Stage 2 differentiated neonates from infants using the HCUP-KID neonate identifier. Stage 3 identified surgical ToF neonates and infants based on the combination and order of procedure codes using procedure day (Table E1). Initially, the study was designed in 3 stages, yet center-specific validation is prohibited for HCUP datasets; therefore, Stage 4 was developed to improve our model specificity. Using the cohorts from Stage 3, reports with counts of all ICD-10 diagnosis codes were generated. These results found several diagnosis codes inconsistent with a diagnosis of surgical repair for cyanotic ToF neonates, as adjudicated by 3 of the investigators (S.M.H., D.B., T.K.) (Table E1). Therefore, these codes were incorporated as exclusion criteria.

Characterizing Definitive Repair Following Surgical Shunt in HCUP-KID

Three strategies were attempted to capture infant definitive repairs following surgical shunt (SS). First, emulating the capture of ductal stent infants, procedural codes that might indicate the removal of an SS from the pulmonary arteries were applied with codes for a definitive repair (Table E4 and Figure E1). Although unsuccessful, presumed shunt take-down codes were substituted for diagnosis codes that might indicate prior cardiac surgery or repair of the sternum or chest wall (Table E4). Both strategies resulted in the capture of patients either disproportional to surgically palliated ToF patients or below the reportable threshold (n < 11). Out of options, our team assessed reports of ICD-10 procedure and diagnosis codes for neonates and infants captured in Stage 2 of the algorithm. We were surprised to find a large incidence of codes describing a thymectomy. We rationalized that the absence of a thymectomy code could indicate a prior median sternotomy for SS in this population (Figure E2, Table E4). Before the application of this theory, we assessed the incidence of thymectomy reporting for ToF neonates previously captured in Stages 1 through 3. These codes were only identifiable in a fraction of patients and present for some with a DiGeorge diagnosis. These preliminary findings indicated that this strategy had low sensitivity and specificity, indicating that a thymectomy code cannot be used to indicate a prior sternotomy. Our team has 3 main explanations for these challenges. We have 3 main explanations for our challenges in defining serial hospitalizations in previously shunted patients.

• •HCUP-KID contains 40 diagnosis codes and 25 procedural codes per record. Therefore, the sampling schema may miss important codes that must be consistently included for cohort capture.
• •Codes for prior cardiac surgery, or prior median sternotomy, may not have been a billing priority for complete ToF repairs and, therefore, omitted.
• •The codes for the removal of a Blalock-Thomas-Taussig shunt in the ICD-10 were converted from the ICD-9 and may not have been a priority for retraining with the ICD transition. This final explanation is logical because we consistently identified patients receiving a ductal stent at definitive repair. These removal codes were intentionally introduced to ICD-10 billing schema, and coders likely received appropriate training. With these challenges, our team further investigated the HCUP-National Readmission Dataset (NRD), which possessed linkage codes and showed promise for identifying previously shunted patients.

Figure E1.

Healthcare Cost and Utilization Project-Kids Inpatient Dataset algorithm for identifying serial interventions for patients with tetralogy of Fallot (ToF) with prior surgical palliation. VSD, Ventricular septal defect.

Figure E2.

Schematic depiction of logic for identifying patients with prior surgical shunts based on thymectomy codes, which might indicate prior median sternotomy.

Capturing a National Sample

Initially, we chose to investigate HCUP-KID rather than other datasets because we hoped to capture a national sample that pulled from 4000 US hospitals, rather than the 49 tertiary care centers included in PHIS. We integrated ICD-10 diagnosis and procedure codes, as well as the order of events based on procedure day (reported in 98% of ToF study records) or column order, to capture a national sample of ToF neonates requiring intervention. Similar ICD strategies have been employed in the Pediatric Health Information System and HCUP-National Inpatient Sample and the Society of Thoracic Surgeons Congenital Heart Surgery Database clinical registry.E2, E3, E4, E5, E6 However, assessment of neonatal palliative strategies has not been attempted in HCUP-KID. We also combined ICD-10 codes that reflect ligation or removal of a patent ductus arteriosus with codes for definitive repair to differentiate those undergoing definitive repair after ductal stent placement from infants undergoing an index definitive repair. Therefore, across this dataset, major steps of the ductal stent pathway could be captured without linkage codes that could identify serial admissions.

Investigations into HCUP-NRD

HCUP-NRD: Data source and study design

Derived from the State Inpatient Database, HCUP-NRD was designed to address health care readmissions for all ages on a national scale. It is derived from 30 geographically dispersed states encompassing 61.1% of the US population and 60% of all US hospitalizations. This dataset is advantageous as each patient contains a unique identifier code to track readmissions within the state and the calendar year.^E7 This dataset includes patient demographics, including age, median household income quartile, urban/rural location of patients residence, and expected payment source, yet it does not include hospital factors or a neonate identifier code. The NRD is available on an annual basis and the years 2016 through 2021 were queried.

A similar algorithm to HCUP-KID was designed to capture the cyanotic neonatal ToF surgical population. Because this dataset lacked a neonatal identifier code, stage 2 utilized discharges with an associated HCUP-NRD birth code (Table E1) an additional arm to capture newborn infants who were readmitted or transferred to a different hospital within 14 days of birth discharge (Table E1). Patients in boxes 5 through 7 were captured from January through June to enable a 6-month readmission analysis, the time frame for an expected definitive repair in the calendar year. A fourth stage was added to query for readmission procedural codes associated with reintervention or definitive repair in that calendar year. Using the NRD identifier codes patients following index procedure were queried to assess whether reintervention or definitive repair, as defined by box 5 through 7 codes, occurred in the same calendar year. Initial queries for reinterventions and definitive repair following palliation did not exceed the HCUP data use agreement reportable threshold (n > 11) and this dataset was not pursued further.

Working with HCUP-NRD

HCUP-NRD was pursued as an alternative to HCUP-KID to identify definitive repairs following surgical palliations and reinterventions. This dataset does not prioritize pediatric patients in its sampling, nor does it oversample complicated patients such as those with congenital heart disease. This was a known limitation that we had hoped would be surpassed by collating 2018 through 2021 years of the dataset. Through initial queries deploying a similar methodology from HCUP-KID, we identified several other important challenges and limitations that are important for future investigations into the NRD.

• •HCUP-NRD does not have a neonatal identifier code challenging the identification of newborn infants for their initial intervention. Hospital transfers are frequent in this population, and in some states, they occur in the majority of neonates with congenital heart disease undergoing surgery.^E8 We utilized readmission codes within 14 days of birth discharge to account for patients with ToF who might have an early readmission for cyanosis or be transferred from an adult hospital to a dedicated children's hospital or a hospital that could perform an intervention. This strategy allowed us to improve our capture for the initial intervention.
• •HCUP-NRD only includes serial readmission codes for patients within a single calendar year. To avoid potential bias for reinterventions based on annual incidence, the samples were fractionated to the first 6 months of a calendar year. This enabled ample time for interstage reinterventions and a definitive repair to occur by age 6 months, the standard timeframe for the ToF paradigm. This strategy was necessary for our initial queries and readmission analysis. However, US birth trends favor births in the latter half of the year, predominantly July through October, reflecting conceptions October through February.^E9 Initial queries reflected this pattern which was not investigated further and may be important for future investigations into case volume for patients with congenital heart disease.
• •Because HCUP-NRD is derived from individual State Inpatient Database groupings, readmissions between hospitals in different states cannot be tabulated. This is especially a challenge for the population with congenital heart disease and their families who travel between state lines to centers of surgical excellence. This trans-state strategy is common practice, has been supported by legislature,^E10 and is a part of theoretical models for regionalization^E11 and improved care for congenital heart disease. This limitation certainly influences our ability to capture readmitted patients and use of this dataset. With these considerations, other datasets are likely to be more adept at understanding readmissions and the complete care paradigm for patients with ToF and congenital heart disease undergoing surgery.

Table E1.

Healthcare Cost and Utilization Project-Kids (HCUP-KID) Inpatient Dataset algorithm codes

Code type	Code	Definition
Stage 1
ICD-10-CM	Q21.3	Tetralogy of Fallot
ICD-10-CM	Q21.0	Ventricular Septal Defect
ICD-10-CM	Q22.1	Congenital Pulmonary Valve Stenosis
ICD-10-CM	Q22.0	Pulmonary Valve Atresia (Congenital)
Stage 2
HCUP-KID Core File	AGE_NEONATE = 1	Neonatal Age (first 28 d after birth) Indicator
HCUP-NRD Core File	I10_BIRTH	ICD-10-CM Birth Indicator
Stage 3
Box 5 early primary repair
ICD-10-PCS	02QM0ZZ	Repair Ventricular Septum, Open Approach
ICD-10-PCS	02RM07Z	Replacement of Ventricular Septum with Autologous Tissue Substitute, Open Approach
ICD-10-PCS	02RM0JZ	Replacement of Ventricular Septum with Synthetic Substitute, Open Approach
ICD-10-PCS	02RM0KZ	Replacement of Ventricular Septum with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	02RM08Z	Replacement of Ventricular Septum with Zooplastic Tissue, Open Approach
ICD-10-PCS	02UM07Z	Supplement Ventricular Septum with Autologous Tissue Substitute, Open Approach
ICD-10-PCS	02UM0JZ	Supplement Ventricular Septum with Synthetic Substitute, Open Approach
ICD-10-PCS	02UM0JZ	Supplement Ventricular Septum with Synthetic Substitute, Open Approach
ICD-10-PCS	02UM0KZ	Supplement Ventricular Septum with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	02UM08Z	Supplement Ventricular Septum with Zooplastic Tissue, Open Approach
ICD-10-PCS	02NH0ZZ	Release Pulmonary Valve, Open Approach
ICD-10-PCS	02RH07Z	Replacement of Pulmonary Valve with Autologous Tissue Substitute, Open Approach
ICD-10-PCS	02RH08Z	Replacement of Pulmonary Valve with Zooplastic Tissue, Open Approach
ICD-10-PCS	02RH0JZ	Replacement of Pulmonary Valve with Synthetic Substitute, Open Approach
ICD-10-PCS	02RH0KZ	Replacement of Pulmonary Valve with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	02RP0JZ	Replacement of Pulmonary Trunk with Synthetic Substitute, Open Approach
ICD-10-PCS	02RP07Z	Replacement of Pulmonary Trunk with Autologous Tissue Substitute, Open Approach
ICD-10-PCS	02RP08Z	Replacement of Pulmonary Trunk with Zooplastic Tissue, Open Approach
ICD-10-PCS	02RP0KZ	Replacement of Pulmonary Trunk with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	02QH0ZZ	Repair Pulmonary Valve, Open Approach
Box 6 ductal stent
ICD-10-PCS	027R34T	Dilation of Ductus Arteriosus with Drug-eluting Intraluminal Device, Percutaneous Approach
ICD-10-PCS	027R3DT	Dilation of Ductus Arteriosus with Intraluminal Device, Percutaneous Approach
ICD-10-PCS	027R44T	Dilation of Ductus Arteriosus with Drug-eluting Intraluminal Device, Percutaneous Endoscopic Approach
ICD-10-PCS	027R4DT	Dilation of Ductus Arteriosus with Intraluminal Device, Percutaneous Endoscopic Approach
Box 7 systemic to pulmonary shunt
ICD-10-PCS	021P08A	Bypass Pulmonary Trunk from Innominate Artery with Zooplastic Tissue, Open Approach
ICD-10-PCS	021P08B	Bypass Pulmonary Trunk from Subclavian with Zooplastic Tissue, Open Approach
ICD-10-PCS	021P08D	Bypass Pulmonary Trunk from Carotid with Zooplastic Tissue, Open Approach
ICD-10-PCS	021P09A	Bypass Pulmonary Trunk from Innominate Artery with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021P09B	Bypass Pulmonary Trunk from Subclavian with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021P09D	Bypass Pulmonary Trunk from Carotid with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021P0AA	Bypass Pulmonary Trunk from Innominate Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021P0AB	Bypass Pulmonary Trunk from Subclavian with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021P0AD	Bypass Pulmonary Trunk from Carotid with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021P0JA	Bypass Pulmonary Trunk from Innominate Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	021P0JB	Bypass Pulmonary Trunk from Subclavian with Synthetic Substitute, Open Approach
ICD-10-PCS	021P0JD	Bypass Pulmonary Trunk from Carotid with Synthetic Substitute, Open Approach
ICD-10-PCS	021P0KA	Bypass Pulmonary Trunk from Innominate Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021P0KB	Bypass Pulmonary Trunk from Subclavian with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021P0KD	Bypass Pulmonary Trunk from Carotid with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021P0ZA	Bypass Pulmonary Trunk from Innominate Artery with No Device, Open Approach
ICD-10-PCS	021P0ZB	Bypass Pulmonary Trunk from Subclavian with No Device, Open Approach
ICD-10-PCS	021P0ZD	Bypass Pulmonary Trunk from Carotid with No Device, Open Approach
ICD-10-PCS	021Q08A	Bypass Right Pulmonary Artery from Innominate Artery with Zooplastic Tissue, Open Approach
ICD-10-PCS	021Q08B	Bypass Right Pulmonary Artery from Subclavian with Zooplastic Tissue, Open Approach
ICD-10-PCS	021Q08D	Bypass Right Pulmonary Artery from Carotid with Zooplastic Tissue, Open Approach
ICD-10-PCS	021Q09A	Bypass Right Pulmonary Artery from Innominate Artery with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021Q09B	Bypass Right Pulmonary Artery from Subclavian with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021Q09D	Bypass Right Pulmonary Artery from Carotid with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021Q0AA	Bypass Right Pulmonary Artery from Innominate Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021Q0AB	Bypass Right Pulmonary Artery from Subclavian with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021Q0AD	Bypass Right Pulmonary Artery from Carotid with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021Q0JA	Bypass Right Pulmonary Artery from Innominate Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	021Q0JB	Bypass Right Pulmonary Artery from Subclavian with Synthetic Substitute, Open Approach
ICD-10-PCS	021Q0JD	Bypass Right Pulmonary Artery from Carotid with Synthetic Substitute, Open Approach
ICD-10-PCS	021Q0KA	Bypass Right Pulmonary Artery from Innominate Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021Q0KB	Bypass Right Pulmonary Artery from Subclavian with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021Q0KD	Bypass Right Pulmonary Artery from Carotid with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021Q0ZA	Bypass Right Pulmonary Artery from Innominate Artery with No Device, Open Approach
ICD-10-PCS	021Q0ZB	Bypass Right Pulmonary Artery from Subclavian with No Device, Open Approach
ICD-10-PCS	021Q0ZD	Bypass Right Pulmonary Artery from Carotid with No Device, Open Approach
ICD-10-PCS	021R08A	Bypass Left Pulmonary Artery from Innominate Artery with Zooplastic Tissue, Open Approach
ICD-10-PCS	021R08B	Bypass Left Pulmonary Artery from Subclavian with Zooplastic Tissue, Open Approach
ICD-10-PCS	021R08D	Bypass Left Pulmonary Artery from Carotid with Zooplastic Tissue, Open Approach
ICD-10-PCS	021R09A	Bypass Left Pulmonary Artery from Innominate Artery with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021R09B	Bypass Left Pulmonary Artery from Subclavian with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021R09D	Bypass Left Pulmonary Artery from Carotid with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021R0AA	Bypass Left Pulmonary Artery from Innominate Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021R0AB	Bypass Left Pulmonary Artery from Subclavian with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021R0AD	Bypass Left Pulmonary Artery from Carotid with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021R0JA	Bypass Left Pulmonary Artery from Innominate Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	021R0JB	Bypass Left Pulmonary Artery from Subclavian with Synthetic Substitute, Open Approach
ICD-10-PCS	021R0JD	Bypass Left Pulmonary Artery from Carotid with Synthetic Substitute, Open Approach
ICD-10-PCS	021R0KA	Bypass Left Pulmonary Artery from Innominate Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021R0KB	Bypass Left Pulmonary Artery from Subclavian with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021R0KD	Bypass Left Pulmonary Artery from Carotid with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021R0ZA	Bypass Left Pulmonary Artery from Innominate Artery with No Device, Open Approach
ICD-10-PCS	021R0ZB	Bypass Left Pulmonary Artery from Subclavian with No Device, Open Approach
ICD-10-PCS	021R0ZD	Bypass Left Pulmonary Artery from Carotid with No Device, Open Approach
ICD-10-PCS	021W08P	Bypass Thoracic Aorta, Descending to Pulmonary Trunk with Zooplastic Tissue, Open Approach
ICD-10-PCS	021W08Q	Bypass Thoracic Aorta, Descending to Right Pulmonary Artery with Zooplastic Tissue, Open Approach
ICD-10-PCS	021W08R	Bypass Thoracic Aorta, Descending to Left Pulmonary Artery with Zooplastic Tissue, Open Approach
ICD-10-PCS	021W09P	Bypass Thoracic Aorta, Descending to Pulmonary Trunk with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021W09Q	Bypass Thoracic Aorta, Descending to Right Pulmonary Artery with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021W09R	Bypass Thoracic Aorta, Descending to Left Pulmonary Artery with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021W0AP	Bypass Thoracic Aorta, Descending to Pulmonary Trunk with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021W0AQ	Bypass Thoracic Aorta, Descending to Right Pulmonary Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021W0AR	Bypass Thoracic Aorta, Descending to Left Pulmonary Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021W0JP	Bypass Thoracic Aorta, Descending to Pulmonary Trunk with Synthetic Substitute, Open Approach
ICD-10-PCS	021W0JQ	Bypass Thoracic Aorta, Descending to Right Pulmonary Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	021W0JR	Bypass Thoracic Aorta, Descending to Left Pulmonary Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	021W0KP	Bypass Thoracic Aorta, Descending to Pulmonary Trunk with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021W0KQ	Bypass Thoracic Aorta, Descending to Right Pulmonary Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021W0KR	Bypass Thoracic Aorta, Descending to Left Pulmonary Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021W0ZP	Bypass Thoracic Aorta, Descending to Pulmonary Trunk, Open Approach
ICD-10-PCS	021W0ZQ	Bypass Thoracic Aorta, Descending to Right Pulmonary Artery, Open Approach
ICD-10-PCS	021W0ZR	Bypass Thoracic Aorta, Descending to Left Pulmonary Artery, Open Approach
ICD-10-PCS	021X08P	Bypass Thoracic Aorta, Ascending/Arch to Pulmonary Trunk with Zooplastic Tissue, Open Approach
ICD-10-PCS	021X08Q	Bypass Thoracic Aorta, Ascending/Arch to Right Pulmonary Artery with Zooplastic Tissue, Open Approach
ICD-10-PCS	021X08R	Bypass Thoracic Aorta, Ascending/Arch to Left Pulmonary Artery with Zooplastic Tissue, Open Approach
ICD-10-PCS	021X09P	Bypass Thoracic Aorta, Ascending/Arch to Pulmonary Trunk with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021X09Q	Bypass Thoracic Aorta, Ascending/Arch to Right Pulmonary Artery with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021X09R	Bypass Thoracic Aorta, Ascending/Arch to Left Pulmonary Artery with Autologous Venous Tissue, Open Approach
ICD-10-PCS	021X0AP	Bypass Thoracic Aorta, Ascending/Arch to Pulmonary Trunk with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021X0AQ	Bypass Thoracic Aorta, Ascending/Arch to Right Pulmonary Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021X0AR	Bypass Thoracic Aorta, Ascending/Arch to Left Pulmonary Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	021X0JP	Bypass Thoracic Aorta, Ascending/Arch to Pulmonary Trunk with Synthetic Substitute, Open Approach
ICD-10-PCS	021X0JQ	Bypass Thoracic Aorta, Ascending/Arch to Right Pulmonary Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	021X0JR	Bypass Thoracic Aorta, Ascending/Arch to Left Pulmonary Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	021X0KP	Bypass Thoracic Aorta, Ascending/Arch to Pulmonary Trunk with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021X0KQ	Bypass Thoracic Aorta, Ascending/Arch to Right Pulmonary Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021X0KR	Bypass Thoracic Aorta, Ascending/Arch to Left Pulmonary Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	021X0ZP	Bypass Thoracic Aorta, Ascending/Arch to Pulmonary Trunk, Open Approach
ICD-10-PCS	021X0ZQ	Bypass Thoracic Aorta, Ascending/Arch to Right Pulmonary Artery, Open Approach
ICD-10-PCS	021X0ZR	Bypass Thoracic Aorta, Ascending/Arch to Left Pulmonary Artery, Open Approach
ICD-10-PCS	031309N	Bypass Right Subclavian Artery to Left Pulmonary Artery with Autologous Venous Tissue, Open Approach
ICD-10-PCS	03130AM	Bypass Right Subclavian Artery to Right Pulmonary Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	03130AN	Bypass Right Subclavian Artery to Left Pulmonary Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	03130JM	Bypass Right Subclavian Artery to Right Pulmonary Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	03130JN	Bypass Right Subclavian Artery to Left Pulmonary Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	03130KM	Bypass Right Subclavian Artery to Right Pulmonary Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	03130KN	Bypass Right Subclavian Artery to Left Pulmonary Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	03130ZM	Bypass Right Subclavian Artery to Right Pulmonary Artery, Open Approach
ICD-10-PCS	03130ZN	Bypass Right Subclavian Artery to Left Pulmonary Artery, Open Approach
ICD-10-PCS	031409M	Bypass Left Subclavian Artery to Right Pulmonary Artery with Autologous Venous Tissue, Open Approach
ICD-10-PCS	031409N	Bypass Left Subclavian Artery to Left Pulmonary Artery with Autologous Venous Tissue, Open Approach
ICD-10-PCS	03140AM	Bypass Left Subclavian Artery to Right Pulmonary Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	03140AN	Bypass Left Subclavian Artery to Left Pulmonary Artery with Autologous Arterial Tissue, Open Approach
ICD-10-PCS	03140JM	Bypass Left Subclavian Artery to Right Pulmonary Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	03140JN	Bypass Left Subclavian Artery to Left Pulmonary Artery with Synthetic Substitute, Open Approach
ICD-10-PCS	03140KM	Bypass Left Subclavian Artery to Right Pulmonary Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	03140KN	Bypass Left Subclavian Artery to Left Pulmonary Artery with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	03140ZM	Bypass Left Subclavian Artery to Right Pulmonary Artery, Open Approach
ICD-10-PCS	03140ZN	Bypass Left Subclavian Artery to Left Pulmonary Artery, Open Approach
ICD-10-PCS	031309M	Bypass Right Subclavian Artery to Right Pulmonary Artery with Autologous Venous Tissue, Open Approach
Box 8 complete repair following ductal stent
ICD-10-PCS	02LR0CT	Occlusion of Ductus Arteriosus with Extraluminal Device, Open Approach
ICD-10-PCS	02LR0DT	Occlusion of Ductus Arteriosus with Intraluminal Device, Open Approach
ICD-10-PCS	02LR0ZT	Occlusion of Ductus Arteriosus, Open Approach
ICD-10-PCS	02VR0CT	Restriction of Ductus Arteriosus with Extraluminal Device, Open Approach
ICD-10-PCS	02VR0DT	Restriction of Ductus Arteriosus with Intraluminal Device, Open Approach
ICD-10-PCS	02VR0ZT	Restriction of Ductus Arteriosus, Open Approach
ICD-10-PCS	02QM0ZZ	Repair Ventricular Septum, Open Approach
ICD-10-PCS	02RM07Z	Replacement of Ventricular Septum with Autologous Tissue Substitute, Open Approach
ICD-10-PCS	02RM0JZ	Replacement of Ventricular Septum with Synthetic Substitute, Open Approach
ICD-10-PCS	02RM0KZ	Replacement of Ventricular Septum with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	02RM08Z	Replacement of Ventricular Septum with Zooplastic Tissue, Open Approach
ICD-10-PCS	02UM07Z	Supplement Ventricular Septum with Autologous Tissue Substitute, Open Approach
ICD-10-PCS	02UM0JZ	Supplement Ventricular Septum with Synthetic Substitute, Open Approach
ICD-10-PCS	02UM0JZ	Supplement Ventricular Septum with Synthetic Substitute, Open Approach
ICD-10-PCS	02UM0KZ	Supplement Ventricular Septum with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	02UM08Z	Supplement Ventricular Septum with Zooplastic Tissue, Open Approach
ICD-10-PCS	02NH0ZZ	Release Pulmonary Valve, Open Approach
ICD-10-PCS	02RH07Z	Replacement of Pulmonary Valve with Autologous Tissue Substitute, Open Approach
ICD-10-PCS	02RH08Z	Replacement of Pulmonary Valve with Zooplastic Tissue, Open Approach
ICD-10-PCS	02RH0JZ	Replacement of Pulmonary Valve with Synthetic Substitute, Open Approach
ICD-10-PCS	02RH0KZ	Replacement of Pulmonary Valve with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	02RP0JZ	Replacement of Pulmonary Trunk with Synthetic Substitute, Open Approach
ICD-10-PCS	02RP07Z	Replacement of Pulmonary Trunk with Autologous Tissue Substitute, Open Approach
ICD-10-PCS	02RP08Z	Replacement of Pulmonary Trunk with Zooplastic Tissue, Open Approach
ICD-10-PCS	02RP0KZ	Replacement of Pulmonary Trunk with Nonautologous Tissue Substitute, Open Approach
ICD-10-PCS	02QH0ZZ	Repair Pulmonary Valve, Open Approach
Stage 4 HCUP-KID exclusion codes
ICD-10-CM	Q24.0	Dextrocardia
ICD-10-CM	Q25.1	Coarctation of aorta
ICD-10-CM	Q26.2	Total anomalous pulmonary venous connection
ICD-10-CM	Q23.4	Hypoplastic left heart syndrome
ICD-10-CM	Q20.4	Double inlet ventricle
ICD-10-CM	Q21.2	Atrioventricular septal defect
ICD-10-CM	Q20.3	Discordant ventriculoarterial connection

ICD-10-CM, International Classification of Diseases, Tenth Revision, Clinical Modification; ICD-10-PCS, International Classification of Diseases, Tenth Revision, Procedure Coding System.

Table E2.

Codes employed for descriptors

Code type	Code	Definition
Genetic anomalies
ICD-10-CM	Q999	Chromosomal abnormality, unspecified
ICD-10-CM	Q9388	Other microdeletions
ICD-10-CM	Q9389	Other deletions from the autosomes
ICD-10-CM	Q998	Other specified chromosome abnormalities
ICD-10-CM	Q917	Trisomy 13, unspecified
ICD-10-CM	Q913	Trisomy 18, unspecified
ICD-10-CM	D821	Di George's syndrome
ICD-10-CM	Q909	Down syndrome, unspecified
Prematurity (<37 wk gestation)
ICD-10-CM	P0510	Newborn small for gestational age, unspecified weight
ICD-10-CM	P0732	Preterm newborn, gestational age 29 completed wk
ICD-10-CM	P0734	Preterm newborn, gestational age 31 completed wk
ICD-10-CM	P0736	Preterm newborn, gestational age 33 completed wk
ICD-10-CM	P0735	Preterm newborn, gestational age 32 completed wk
ICD-10-CM	P0737	Preterm newborn, gestational age 34 completed wk
ICD-10-CM	P0738	Preterm newborn, gestational age 35 completed wk
ICD-10-CM	P0739	Preterm newborn, gestational age 36 completed wk
Low birth weight (<2.5 kg)
ICD-10-CM	P0718	Other low birth weight newborn, 2000-2499 g
ICD-10-CM	P0518	Newborn small for gestational age, 2000-2499 g
ICD-10-CM	P0717	Other low birth weight newborn, 1750-1999 g
ICD-10-CM	P0716	Other low birth weight newborn, 1500-1749 g
ICD-10-CM	P0715	Other low birth weight newborn, 1250-1499 g
ICD-10-CM	P0714	Other low birth weight newborn, 1000-1249 g
ICD-10-CM	P0517	Newborn small for gestational age, 1750-1999 g
ICD-10-CM	P0516	Newborn small for gestational age, 1500-1749 g
ICD-10-CM	P0703	Extremely low birth weight newborn, 750-999 g
Right aortic arch
ICD-10-CM	Q2547	Right aortic arch
Pulmonary valve/artery atresia
ICD-10-CM	Q22.0	Pulmonary valve atresia
ICD-10-CM	Q25.5	Atresia of pulmonary artery
Airway abnormalities
ICD-10-CM	J398	Other specified diseases of upper respiratory tract
ICD-10-CM	Q32.0	Congenital tracheomalacia
ICD-10-CM	Q32.2	Congenital bronchomalacia
ICD-10-CM	Q32.1	Other congenital malformations of trachea
Hospital descriptors
HCUP-KID Hospital File	HOSP_BEDSIZE	Bed size of hospital (Small, Medium, Large)
HCUP-KID Hospital File	HOSP_LOCTEACH	Location/teaching status of hospital (rural, Urban nonteaching, Urban teaching)
HCUP-KID Hospital File	HOSP_REGION	Region of hospital (Northeast, Midwest, South, West)
Social determinants of health
HCUP-KID Core File	PL_NCHS	Patient Location: NCHS Urban-Rural Code (Central, Fringe, Metropolitan, Micropolitan, non-metropolitan or micropolitan)
HCUP-KID Core File	ZIPINC_QRTL	Median household income for patient's ZIP Code (based on current year), Quartile
HCUP-KID Core File	PAY1	Expected primary payer, uniform (Medicare, Medicaid, Private Insurance, Self-pay, No charge, Other, Missing, Invalid)
HCUP-KID Core File	RACE	Race/ethnicity of patient (White, Black, Hispanic, Asian or Pacific Islander, Native American, Other)

ICD-10-CM, International Classification of Diseases, Tenth Revision, Clinical Modification; ICD-10-PCS, International Classification of Diseases, Tenth Revision, Procedure Coding System; HCUP-KID, Healthcare Cost and Utilization Project-Kids Inpatient Dataset.

Table E3.

Codes employed to define outcomes

Code type	Code	Definition
HCUP-KID Core File	DIED	Died during hospitalization
ICD-10-PCS	5A1522F	Extracorporeal Oxygenation, Membrane, Central
ICD-10-PCS	5A1522G	Extracorporeal Oxygenation, Membrane, Peripheral Veno-arterial
ICD-10-PCS	5A15223	Extracorporeal Membrane Oxygenation, Continuous
ICD-10-PCS	5A09557	Assistance with Respiratory Ventilation, Greater than 96 Consecutive Hours, Continuous Positive Airway Pressure
HCUP-KID Core File	LOS	Length of stay, cleaned
HCUP-KID Core File	TOTCHG	Total charges, cleaned (rounded dollars)

HCUP-KID, Healthcare Cost and Utilization Project-Kids Inpatient Dataset; ICD-10-PCS, International Classification of Diseases, Tenth Revision, Procedure Coding System.

Table E4.

Challenges with definitive repair following shunt

Code type	Code	Definition
Surgical shunt removal
ICD-10-PCS	02PY0JZ	Removal of Synthetic Substitute from Great Vessel, Open Approach
ICD-10-PCS	02PY0KZ	Removal of Nonautologous Tissue Substitute from Great Vessel, Open Approach
ICD-10-PCS	02PY0YZ	Removal of Other Device from Great Vessel, Open Approach
ICD-10-PCS	02CP0ZZ	Extirpation of Matter from Pulmonary Trunk, Open Approach
ICD-10-PCS	02CQ0ZZ	Extirpation of Matter from Right Pulmonary Artery, Open Approach
ICD-10-PCS	02CR0ZZ	Extirpation of Matter from Left Pulmonary Artery, Open Approach
Prior cardiac surgery
ICD-10-CM	Z8774	Personal history of (corrected) congenital malformations of heart and circulatory system
ICD-10-CM	Z95818	Presence of other cardiac implants and grafts
ICD-10-CM	T82857A	Stenosis of other cardiac prosthetic devices, implants and grafts, initial encounter
ICD-10-CM	T82868A	Thrombosis due to vascular prosthetic devices, implants and grafts, initial encounter
ICD-10-CM	Z95828	Presence of other vascular implants and grafts
ICD-10-CM	T82858A	Stenosis of other vascular prosthetic devices, implants and grafts, initial encounter
ICD-10-CM	Y848	Other medical procedures as the cause of abnormal reaction of the patient, or of later complication, without mention of misadventure at the time of the procedure
ICD-10-CM	Y832	Surgical operation with anastomosis, bypass or graft as the cause of abnormal reaction of the patient, or of later complication, without mention of misadventure at the time of the procedure
ICD-10-CM	Z95	Presence of Cardiac and vascular implants and grafts
ICD-10-CM	Z95.9	Presence of cardiac implant or graft
ICD-10-PCS	0PQ00ZZ	Repair Sternum, Open Approach
ICD-10-PCS	0WQ80ZZ	Repair Chest Wall, Open Approach
ICD-10-PCS	0WQ8XZZ	Repair Chest Wall, External Approach
Codes for thymectomy and validation
ICD-10-PCS	07TM0ZZ	Resection of Thymus, Open Approach
ICD-10-PCS	07BM0ZZ	Excision of Thymus, Open Approach
ICD-10-CM	D821	Di George's syndrome

ICD-10-PCS, International Classification of Diseases, Tenth Revision, Procedure Coding System.

Table E5.

Trends in the day of neonatal interventional strategy between the 3 cross-sectional time points

Strategy	Year	n	Estimated discharges∗	Day intervention occurred
				Median	Quartile 1	Quartile 3	Min	Max
Early primary repair	2016	42	59.5	4.8	3.09	8.25	1	13
	2019	46	61.9	4.2	2.20	8.60	1	13
	2022	28	38.1	5.7	3.49	8.73	1	20
Ductal stents	2016	19	26.7	3.1	1.42	4.66	1	15
	2019	31	41.5	3.6	1.65	6.04	1	19
	2022	57	77.1	2.9	1.15	5.74	1	18
Surgical shunts	2016	130	184.9	2.8	1.00	5.27	1	15
	2019	90	119	3.5	1.80	6.55	1	18
	2022	76	103.5	3.9	1.43	7.66	1	20

^∗n weighted.