Abstract
Background:
The effect of the consolidation of neonatal pediatric surgical cases to limited surgeons within a hospital is unknown. We elected to model the distribution of complex neonatal procedures using an economic measure of market concentration, the Herfindahl-Hirschmann Index (HHI), and study its effect on outcomes of index pediatric surgical operations.
Methods:
We used data from 49 US children’s hospitals between 2007 and 2017 for the following procedures: congenital diaphragmatic hernia repair (CDH), esophageal atresia and tracheoesophageal fistula repair (EA/TEF), and pull-through for Hirschsprung disease (HD). Mixed effects logistic regression modeling was used to adjust for salient patient characteristics to determine the effect of HHI on in-hospital mortality, condition-specific one-year re-operation, and one-year unplanned readmissions.
Results:
A total of 2270 infants were identified who underwent surgery for the three conditions of interest. On multivariable analysis, increasing HHI was not associated with differences in mortality or condition-specific reoperation within the first year. A decrease in the number of unplanned readmissions at highly concentrated centers was seen for HD (RR 0.8 CI (0.69–0.97), p = 0.02) and CDH (RR 0.4 CI (0.28–0.71), p b 0.001).
Conclusions:
Pediatric surgical specialization did not affect mortality or condition-specific re-operation. However, it did decrease the number of unplanned readmissions following CDH repairs and pull-throughs for HD.
Study design:
Retrospective Cohort Study.
Level of Evidence:
Level II.
Keywords: Specialization, Pediatric surgery, Neonatal, Health services research
Understanding how hospital and surgeon characteristics, including operative volume and specialization, impact patient outcomes and quality of care is of increasing interest and importance. This is especially relevant in the setting of national quality improvement programs such as the American College of Surgeons Children’s Verification Program that set standards for pediatric surgical care and encourage the regionalization of this care to high-volume centers [1]. In adult patients, increasing hospital and surgeon volume have been associated with improved patient outcomes [2, 3]. In a pediatric population, a similar relationship has been shown for biliary atresia and congenital diaphragmatic hernia repair [4, 5], but it has not been demonstrated consistently across a broader range of conditions [6, 7]. Because surgeon volume is not a consistent predictor of surgical quality, another predictor of surgical quality that has been explored is surgeon specialization. One previous study in adults looking at the diversity or specialization of cases done by an individual surgeon demonstrated that more specialized surgeons had improved patient mortality [8]. Another study demonstrated that pediatric patients had improved outcomes if they were cared for by surgeons for whom pediatric patients were an higher percentage of their practice [9].
In order to address concerns about surgeon volume and specialization in pediatric surgery, multiple institutions have formed multidisciplinary specialized centers that focus on a specific set of surgical conditions. Some of these centers also concentrate these specific cases to a small number of surgeons within that practice. The impact of increased consolidation of cases into a small number of surgeons within a hospital on patient outcomes is unknown. The goal of this study was to assess the impact of case consolidation within a hospital on patient outcomes for Hirschsprung disease (HD), esophageal atresia/tracheoesophageal fistula and congenital diaphragmatic hernia including mortality, re-operation and re-admission. We hypothesized that increasing consolidation would not significantly impact patient outcomes after adjusting for patient characteristics and co-morbidities.
1. Methods
1.1. Data source and study population
We conducted a retrospective cohort study using the Pediatric Health Information System Database (PHIS). PHIS is an administrative database that contains inpatient, emergency department, ambulatory surgery, and observation encounter-level data from over 49 not-for-profit, tertiary care pediatric hospitals in the United States. These hospitals are affiliated with the Children’s Hospital Association (Lenexa, KS). Data quality and reliability are assured through a joint effort between the Children’s Hospital Association and participating hospitals. Portions of the data submission and data quality processes for the PHIS database are managed by Truven Health Analytics (Ann Arbor, MI). For the purposes of external benchmarking, participating hospitals provide discharge/encounter data including demographics, diagnoses, and procedures. Nearly all of these hospitals also submit resource utilization data (e.g. pharmaceuticals, imaging, and laboratory) into PHIS. Data are de-identified at the time of data submission, and data are subjected to a number of reliability and validity checks before being included in the database. The study was approved by the University of Utah institutional review board (IRB 00086332).
1.2. Patient selection
Patients were included if they had an ICD 9/10 diagnosis code of Hirschsprung disease, esophageal atresia/ tracheoesophageal fistula (EA/TEF) or congenital diaphragmatic hernia (CDH) and ICD 9/10 procedure code for a pull-through procedure, an EA/TEF repair, or a CDH repair from January 1, 2007 until December 31, 2017. A list of ICD 9/10 diagnosis and procedure codes are provided in the supplementary appendix (Supplementary Table 1). For EA/TEF and CDH, patients were excluded who underwent a repair after 7 days of age. For HD, patients were excluded who underwent a pull-through after 1 year of age.
1.3. Primary exposure and outcomes
In order to calculate the level of case consolidation into a small number of pediatric surgeons for each hospital each year, we used the Herfindahl- Hirschman Index (HHI), which is a tool that is used to assess market concentration in economics [10] and has been used previously to measure surgeon specialization [8]. The HHI is calculated by summing the squares of the proportion of cases for each case type done by each surgeon, the surgeon’s “market share”, at each hospital. HHI would decrease as the neonatal cases of interest are spread over a greater number of surgeons or if a greater number of those cases are done at that hospital. For example, a hospital which performs 20 cases per year equally distributed across 5 surgeons would have an HHI of (4/20)2 + (4/20)2 + (4/20)2 + (4/20)2 + (4/20)2 = 0.2. If one surgeon performed 12 cases and equally distributed the remaining 8 cases to the remaining four surgeons the HHI would be (12/20)2 + (2/20)2 + (2/20)2 + (2/20)2 + (2/20)2 = 0.4. The Department of Justice groups markets into three categories based off of HHI: weakly or unconcentrated markets with an HHI < 0.15, moderately concentrated with an HHI between 0.15 and 0.25, and highly concentrated with an HHI N0.25 [11]. We used these same cutoffs to divide hospitals into two groups by HHI (weakly/moderately, <0.25, and highly, ≥0.25, concentrated) to assess the degree of consolidation of their neonatal case volume.
In order to calculate an HHI for each hospital, each surgical encounter had to contain a unique surgeon identifier, so all patients in our cohort who did not have a unique surgeon identifier associated with their surgical encounter were excluded. To accurately assess the number of practicing pediatric surgeons at each hospital, as the hospital may employ surgeons who did none of the cases of interest in that year, we created a cohort in PHIS of all surgeons performing any of the neonatal procedures of interest or common pediatric surgical cases (appendectomy and inguinal hernia repair) over the 10-year period of interest. All unique surgeon identifiers were identified for each hospital and used to calculate the number of surgeons practicing at that hospital that year. At hospitals with fewer than four surgeons or those which did fewer than four of that specific case in a given year, even with completely equal case distribution, the HHI would be highly concentrated due to the small sample, so these hospitals and all of their patients were also excluded.
The primary outcome of interest was in-hospital mortality following a patient’s index surgical procedure. Our secondary outcomes of interest were condition-specific re-operation in the first year following index surgery, and the number of unplanned readmissions in the first year following index surgery. An unplanned readmission was defined as an admission that was coded as either urgent or emergent. A table of procedure specific reoperation codes is provided in the supplementary appendix (Supplementary Table 1).
1.4. Covariates
The PHIS database provides data regarding patient demographics, including race, ethnicity, payer sources, birthweight, and median household income. Patient gestational age was determined using a combination of the specific gestational age data provided in PHIS and ICD-9 diagnosis codes representing gestational age. We used the Agency for Healthcare Research and Quality Clinical Classification Software (CCS) to model patient comorbidities and other surgical procedures based on encounter ICD-9/10 diagnosis and procedure codes (See Supplementary Table 2). The following covariates were included: congenital cardiac anomalies, neurologic anomalies, chromosomal abnormalities, history of central nervous system infection, cleft lip and palate, congenital diaphragmatic hernia, previous cardiac arrest or hypoxic episode, and ostomy placement prior to definitive pull-through procedure. We also included hospital volume, which was defined as the average number of patients with the condition of interest treated operatively per year at that hospital.
1.5. Statistical analysis
All statistical analysis was performed using R version 3.5.2. All statistical tests were two-tailed and a p < 0.05 was considered significant. We first performed a univariate analysis to assess the relationship between HHI category and the primary and secondary outcomes. Chi squared tests were used for categorical variables and ANOVA for continuous variables. To assess the effect of HHI on our primary and secondary outcomes, we developed mixed-effect logistic regression models to adjust for salient patient-level covariates while clustering within hospital as a random effect [12]. For each model, forward stepwise selection was performed and patient characteristics with a p < 0.05 were selected to remain in the model. Discrimination was assessed using the area under the receiver operating curve (AUC) and calibration was assessed using the Brier Score [13, 14].
2. Results
2.1. Patient and hospital demographics
From 2007 to 2017, 4766 children underwent one of the three procedures of interest: 2030 for HD, 1648 for EA/TEF, and 1086 for CDH. Of those 2270 were treated at a hospital where an HHI could be calculated: 1097 HD patients at 33 hospitals, 673 EA/TEF patients at 31 hospitals, and 500 CDH patients at 20 hospitals. For each condition, hospitals were divided into two categories based on their HHI categorization: weakly/moderately or highly concentrated, and baseline patient characteristics were compared across HHI categories (Table 1). For HD, patients of a lower gestational age were more likely to be treated at a hospital with a weakly/moderately concentrated HHI (p < 0.001). Hispanic patients were more likely to be treated at a hospital with a highly concentrated HHI for both EA/TEF and CDH (p < 0.001 for both). For both HD (p = 0.03) and EA/TEF (p = 0.02), patients with congenital chromosomal abnormalities were more likely to be treated at facilities with weakly/moderately concentrated HHI. There were no statistically significant trends between HHI category and gender, black race, public insurance, birthweight, congenital cardiac anomalies, or congenital neurologic anomalies across any of the three conditions.
Table 1.
Patient characteristics by hospital HHI category.
| Characteristic | Weakly/Moderately Concentrated (HHI <0.25) | Highly Concentrated (HHI ≥ 0.25) | p-Value |
|---|---|---|---|
| Hirschsprung disease (n = 1097) | (n = 459) | (n = 638) | |
| HHI range | 0.13–0.24 | 0.25–0.68 | <0.001 |
| Female sex (%) | 24 | 24 | 1 |
| Black race (%) | 17 | 14 | 0.2 |
| Hispanic or Latino ethnicity (%) | 10 | 8 | 0.2 |
| Public insurance (%) | 49 | 53 | 0.4 |
| Birthweight (kg, mean ± SD) | 3.3 (±0.55) | 3.3 (±0.54) | 0.8 |
| Gestational age (wk, mean ± SD) | 38 (±6.4) | 39 (±2.2) | <0.001 |
| Congenital cardiac anomalies (%) | 4 | 2 | 0.1 |
| Chromosomal anomalies (%) | 2 | 1 | 0.02 |
| Neurologic anomalies (%) | 1 | 1 | 0.9 |
| EA/TEF (n = 673) | (n = 331) | (n = 342) | |
| HHI range | 0.13–0.24 | 0.26–0.68 | <0.001 |
| Female sex (%) | 45 | 42 | 1 |
| Black race (%) | 12 | 10 | 0.6 |
| Hispanic or Latino ethnicity (%) | 14 | 15 | <0.001 |
| Public insurance (%) | 46 | 49 | 0.1 |
| Birthweight (kg, mean ± SD) | 2.5 (±0.77) | 2.6 (±0.73) | 0.1 |
| Gestational age (wk, mean ± SD) | 37 (±2.8) | 37 (±3.0) | 0.3 |
| Congenital cardiac anomalies (%) | 11 | 10 | 1 |
| Chromosomal anomalies (%) | 5 | 2 | 0.03 |
| Neurologic anomalies (%) | 8 | 8 | 1 |
| CDH (n = 500) | (n = 129) | (n = 371) | |
| HHI Range | 0.17–0.24 | 0.26–0.76 | <0.001 |
| Female sex (%) | 41 | 43 | 0.7 |
| Black race (%) | 3 | 6 | 0.4 |
| Hispanic or Latino ethnicity (%) | 15 | 21 | <0.001 |
| Public insurance (%) | 47 | 54 | 0.3 |
| Birthweight (kg, mean ± SD) | 3.1 (±0.62) | 3.0 (±0.57) | 0.5 |
| Gestational age (wk, mean ± SD) | 38 (±2.2) | 38 (±2.0) | 0.9 |
| Congenital cardiac anomalies (%) | 5 | 4 | 1.0 |
| Chromosomal anomalies (%) | 0 | 2 | 0.2 |
| Neurologic anomalies (%) | 2 | 1 | 1.0 |
2.2. Consolidation of cases
Cases for all three conditions were similarly concentrated in a small group of surgeons; median HHI for CDH was 0.3 (IQR 0.28–0.36), for EA/TEF was 0.3 (IQR 0.2–0.33), and for HD was 0.3 (IQR 0.22–0.31). For HD, the mean percentage of cases done by the highest volume surgeon was 27% for weakly/moderately concentrated hospitals and 44% for highly concentrated hospitals. For EA/TEF, the mean percentage of cases done by the highest volume surgeon was 26% for weakly/moderately concentrated hospitals and 45% for highly concentrated hospitals. For CDH, the mean percentage of cases done by the highest volume surgeon was 27% for weakly/moderately concentrated hospitals and 47% for highly concentrated hospitals. Combining all three conditions, there was strong correlation between the HHI and the percentage of cases done by the highest volume surgeon at each hospital (r = 0.94, p < 0.001).
2.3. Unadjusted outcomes
The overall mortality rate was 1% in HD, 7% in EA/TEF, and 15% in CDH (Table 2). On univariate analysis, no significant differences were seen in mortality based on HHI category for any of the 3 conditions (Supplementary Table 3). The one-year condition-specific reoperation rate was 8% in HD, 4% in EA/TEF, and 4% in CDH (Table 2). On univariate analysis, no significant differences were seen in re-operation based on HHI category for any of the 3 conditions (Supplementary Table 3). The mean number of unplanned readmissions in the first year was 0.8 (SD ±1.7) for HD, 0.8 (±1.4) for EA/TEF, and 0.5 (±1.3) for CDH (Table 2). For HD, compared to weakly/moderately concentrated hospitals, patients treated at highly concentrated hospitals were at decreased risk of unplanned readmissions (p = 0.01). For EA/TEF and CDH, HHI category had no statistically significant impact on risk of unplanned readmissions (Supplementary Table 3).
Table 2.
Mortality, re-operation, and unplanned readmission unadjusted rates.
| Condition | Weakly/Moderately Concentrated (HHI <0.25) | Highly Concentrated (HHI ≥ 0.25) |
|---|---|---|
| HD | ||
| Mortality rate (%) | 1 | 1 |
| Re-operation rate (%) | 7 | 9 |
| Unplanned readmissions (mean ± SD) | 0.9 ± 2.1 | 0.7 ± 1.3 |
| EA/TEF | ||
| Mortality rate (%) | 8 | 5 |
| Re-operation rate (%) | 5 | 4 |
| Unplanned readmissions (mean ± SD) | 0.8 ± 1.6 | 0.7 ± 1.2 |
| CDH | ||
| Mortality rate (%) | 16 | 15 |
| Re-operation rate (%) | 7 | 4 |
| Unplanned readmissions (mean ± SD) | 0.6 ± 1.6 | 0.5 ± 1.2 |
2.4. Multivariable adjusted outcomes
The patient and hospital characteristics included in each model are listed in Supplementary Table 4. With our sample size and number of covariates we have 95% power to detect a medium effect size for all of our outcomes and conditions. With regards to the primary outcome of in-hospital mortality, after adjusting for patient-level characteristics, there was no association between HHI category and in-hospital mortality: HD (highly compared to weakly/moderately concentrated p = 0.9, EA/TEF (highly compared to weakly/moderately concentrated p = 0.09), and CDH (highly compared to weakly/moderately concentrated p = 0.6, Table 3).
Table 3.
Adjusted impact of HHI on mortality, re-operation and unplanned readmissions.
| Weakly/moderately concentrated (HHI <0.25) | Highly concentrated (HHI ≥ 0.25) | p-Value | AUC, Brier Score | |
|---|---|---|---|---|
| HD | ||||
| Mortality (OR, CI) | Ref | 0.9 (0.23–3.6) | 0.9 | 0.75, 0.008 |
| Re-operation in 1st year (OR, CI) | Ref | 1.4 (0.85–2.2) | 0.2 | 0.65, 0.07 |
| Number of unplanned readmissions (RR, CI) | Ref | 0.8 (0.69–0.97) | 0.02 | 0.80, −0.54 |
| EA/TEF | ||||
| Mortality (OR, CI) | Ref | 0.5 (0.27–1.1) | 0.09 | 0.79, 0.06 |
| Re-operation in 1st year (OR, CI) | Ref | 0.9 (0.35–2.2) | 0.8 | 0.77, 0.04 |
| Number of unplanned readmissions (RR, CI) | Ref | 0.9 (0.73–1.1) | 0.4 | 0.74, −0.15 |
| CDH | ||||
| Mortality (OR, CI) | Ref | 0.9 (0.46–1.5) | 0.6 | 0.63, 0.02 |
| Re-operation in 1st year (OR, CI) | Ref | 0.5 (0.19–1.2) | 0.1 | 0.68, 0.002 |
| Number of unplanned readmissions (RR, CI) | Ref | 0.4 (0.28–0.71) | <0.001 | 0.78, 0.48 |
AUC represents “area under the receiver operating curve”.
With regards to the secondary outcome of condition-specific reoperation in the first year after the initial operation, after adjusting for patient-level characteristics, there was no association between HHI category and re-operation: HD (highly compared to weakly/moderately concentrated p = 0.2, EA/TEF (highly compared to weakly/moderately concentrated p = 0.8), and CDH (highly compared to weakly/moderately concentrated p = 0.1, Table 3).
With regards to the secondary outcome of number of unplanned readmissions in the first year after the initial operation, after adjusting for patient-level characteristics, there was an association between HHI category and readmission seen in HD and CDH patients. HD patients treated at highly concentrated compared to weakly/moderately concentrated hospitals were at decreased risk of unplanned readmissions (adjusted risk ratio, [aRR]: 0.8, CI 0.69–0.97, p = 0.02). CDH patients treated at highly concentrated compared to weakly/moderately concentrated hospitals were also at decreased risk of unplanned readmissions (adjusted risk ratio, [aRR]: 0.4, CI 0.28–0.71, p < 0.001). There was no association between HHI category and number of unplanned readmission for EA/TEF (highly compared to weakly/moderately concentrated p = 0.4, Table 3).
3. Discussion
3.1. Summary of results
Using an economic measure of market concentration to assess consolidation of cases in a small subset of surgeons in a dataset of infants undergoing three complex surgical procedures at US Children’s Hospitals, we found no consistent relationship between case consolidation and in-hospital mortality or condition-specific re-operation. A decrease in unplanned readmissions was seen in patients with HD and CDH treated at highly concentrated hospitals; however, this effect was not observed in EA/TEF.
3.2. Comparison to previous literature
Previous studies of surgeon specialization and patient outcomes have demonstrated a benefit to increasing specialization. Hall et al. demonstrated that when treating an adult patient population surgeons who were more specialized, as measured by a consolidation in the types of cases that each surgeon performed, patient outcomes improved [8]. In pediatric surgery specifically, Rhee et al., developed an index of pediatric surgeon specialization based on the proportion of cases a surgeon performed on pediatric cases compared to their total case volume [9]. They also demonstrated a benefit in being treated by a surgeon who performed a larger proportion of their cases on pediatric patients [9]. Our study used a different marker of surgeon specialization, consolidation of specific case types to a small number of surgeons within a group of practicing pediatric surgeons, and we failed to show any benefit in mortality or re-operation of increased specialization. Nevertheless, for highly concentrated hospitals, there was a decreasing number of unplanned readmissions in patients treated for HD and CDH. This may reflect that more specialization centers have increased resources such as post-discharging teaching and support that help them to reduce readmissions. One limitation of using the PHIS database is that we are unable to capture patients readmitted to another hospital, which limits the generalizability of this finding.
In the setting of increasing concerns about the impact of the increasing pediatric surgical workforce, multiple studies have attempted to understand the modern pediatric surgeon case volume particularly with regards to complex neonatal cases. Fonkalsrud et al. demonstrated that from 1970 to 2010, during which time there was a dramatic increase in the pediatric surgical workforce, the number of index cases performed per surgeon declined both at their hospital and nationally using an administrative database [15]. Abdullah et al. demonstrated a similar trend from 1980 to 2013 by surveying the case logs of recertifying pediatric surgeons [16]. The most effective way to address this decline in surgeon volume needs to be further studied, however, one proposed solution, which has already been implemented in some centers, is to consolidate rare cases into a handful of surgeons. This increases their annual case volume with the goal of maintaining competency. Our study of case consolidation within freestanding children’s hospitals fails to demonstrate a clinical benefit of this strategy other than a modest reduction in unplanned readmissions for two of the three conditions studied.
3.3. Implications
This study failed to demonstrate a consistent and meaningful change in patient outcomes following complex neonatal surgery with increasing consolidation of cases to a small number of surgeons within a hospital. This finding is particularly relevant in the setting of concerns about decreasing surgeon volume of index cases and concerns about what volume of cases is required to maintain competency [16]. Consolidation of cases within a group of practicing surgeons, resulting in formal or informal sub-specialization, is one proposed mechanism to address this problem, however, we were unable to demonstrate a consistent clinical benefit. While this remains a viable option for increasing individual surgeon volume, further study is needed to better understand how to maximize the benefit to patients. In addition to the manner it was used here, HHI may be used in the future to better understand consolidation by linking it to directly with case distribution data from specialized centers or by studying changes in HHI over time as specific hospitals change their practice.
3.4. Limitations
There are several limitations to our study. Mortality is rare following pediatric surgical procedures, and this may not be the most appropriate quality metric, however, it is the standard outcome used in studies of hospital and surgeon volume. In future studies, condition-specific metrics such as enterocolitis in HD, survival of patients to surgical repair in CDH, or recurrent dilations in EA/TEF may be more meaningful outcomes. This was a retrospective study using an administrative database, so we are reliant on the completeness and accuracy of the data particularly with regards to the procedure and diagnosis codes. Additionally, one of the limitations of the PHIS database is that we cannot accurately capture what happens to these patients after discharge unless they are readmitted to the same hospital. Furthermore, we had to eliminate all hospitals that did not report surgeon identifiers or did too few cases to accurately calculate an HHI, which may bias our sample. While consolidation within a high-volume center may not have a measurable impact, it is possible that the impact of the distribution of cases at a low-volume center is more meaningful. Furthermore, the impact of being treated at a highly-specialized center may be greater for patients who are complex and require re-operative procedures or procedures later in life, which would be excluded by our age criteria. While HHI is a well-established metric in economics, its previous use to describe surgical specialization is limited and there are limited comparisons available to validate our results.
4. Conclusions
In the context of concerns about minimum volume of index cases required to maintain pediatric surgeons’ competence one proposed solution that has already been implemented at some centers is further sub-specialization and consolidation of these rare cases. Our study demonstrates that for three neonatal conditions, HD, EA/TEF, and CDH increasing consolidation has no impact on in-hospital mortality or re-operation. For HD and CDH consolidation did result in a modest reduction in the number of unplanned readmissions.
Supplementary Material
Acknowledgments
Funding: This study was funded by the Utah-Intermountain Healthcare Surgical Research Fellowship (LCCP), and a grant from the Agency for Healthcare Research and Quality (1K08HS025776, BTB).
Footnotes
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.jpedsurg.2020.02.044.
References
- [1].American College of Surgeons. Optimal Resources for Children’s Surgical Care v.1: Children’s Surgery Vertification Quality Improvement Program. Chicago, IL: American College of Surgeons; 2015. [Google Scholar]
- [2].Birkmeyer JD, Siewers AE, Finlayson EVA, et al. Hospital volume and surgical mortality in the United States. N Engl J Med 2002;346(15):1128–37. [DOI] [PubMed] [Google Scholar]
- [3].Birkmeyer JD, Stukel TA, Siewers AE, et al. Surgeon volume and operative mortality in the United States. N Engl J Med 2003;349(22):2117–27. [DOI] [PubMed] [Google Scholar]
- [4].McKiernan P, Baker AJ, Lloyd C, et al. British Paediatric surveillance unit study of biliary atresia: outcome at 13 years. J Pediatr Gastroenterol Nutr 2009;48(1):78–81. [DOI] [PubMed] [Google Scholar]
- [5].Bucher BT, Guth RM, Saito JM, et al. Impact of hospital volume on in-hospital mortality of infants undergoing repair of congenital diaphragmatic hernia. Ann Surg 2010; 252(4):635–42. [DOI] [PubMed] [Google Scholar]
- [6].Dubrovsky G, Sacks GD, Friedlander S, et al. Understanding the relationship between hospital volume and patient outcomes for infants with gastroschisis. J Pediatr Surg 2017;52(12):1977–80. [DOI] [PubMed] [Google Scholar]
- [7].Lawrence AE, Minneci PC, Deans KJ, et al. Relationships between hospital and surgeon operative volumes and outcomes of esophageal atresia/tracheoesophageal fistula repair. J Pediatr Surg 2019;54(1):44–9. [DOI] [PubMed] [Google Scholar]
- [8].Hall BL, Hsiao EY, Majercik S, et al. The impact of surgeon specialization on patient mortality. Ann Surg 2009;249(5):708–16. [DOI] [PubMed] [Google Scholar]
- [9].Rhee D, Papandria D, Yang J, et al. Comparison of pediatric surgical outcomes by the surgeon’s degree of specialization in children. J Pediatr Surg 2013;48(8):1657–63. [DOI] [PubMed] [Google Scholar]
- [10].Besanko D Economics of strategy. 4th ed Hoboken, NJ: John Wiley & Sons; 2007. [Google Scholar]
- [11].U. S. Department of Justice and the Federal Trade Commission. In: Justice USDo, editor. Horizontal Merger Guidelines; 2010. [Google Scholar]
- [12].Burton P, Gurrin L, Sly P. Extending the simple linear regression model to account for correlated responses: an introduction to generalized estimating equations and multi-level mixed modelling. Stat Med 1998;17(11):1261–91. [DOI] [PubMed] [Google Scholar]
- [13].Murphy AH. A new vector partition of the probability score. J Appl Meteorol 1973;12 (4):595–600. [Google Scholar]
- [14].Robin X, Turck N, Hainard A, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics 2011;12:77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Fonkalsrud EW, O’Neill JA, Jabaji Z, et al. Changing relationship of pediatric surgical workforce to patient demographics. Am J Surg 2014;207(2):275–80. [DOI] [PubMed] [Google Scholar]
- [16].Abdullah F, Salazar JH, Gause CD, et al. Understanding the operative experience of the practicing pediatric surgeon: implications for training and maintaining competency. JAMA 2016;151(8):735–41. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.