Abstract
Objective
Neonates undergoing congenital heart surgery require highly specialized, resource-intensive care. Location of care and degree of specialization can vary between and within institutions. Using a multi-institutional cohort, we sought to determine if location of admission is associated with an increase in health care costs, resource use and mortality.
Methods
We retrospectively analyzed admissions for neonates (<30 days) undergoing RACHS 2–6 procedures between 2004–13 using the Pediatric Health Information Systems database (44 children’s hospitals). Multivariate generalized estimating equations adjusted for center and patient specific risk factors and stratified by age at admission were performed examining the association of admission intensive care unit (ICU) with total hospital costs, mortality and length of stay (LOS).
Results
Of 19,984 neonates (60% male) identified, 39% were initially admitted to a cardiac ICU (CICU), 48% to a neonatal ICU (NICU), and 13% to a pediatric ICU (PICU). In adjusted models, admission to a CICU vs. NICU was associated with a $20,440 reduction in total hospital cost for infants aged 2–7 days at admission (p=0.007) and a $23,700 reduction in total cost for infants aged 8–14 days at admission (p=0.01). Initial admission to a CICU or PICU vs. NICU at <15 days of age was associated with shorter hospital and ICU LOS and fewer days of mechanical ventilation. There was no difference in adjusted mortality by admission location.
Conclusions
Admission to an ICU specializing in cardiac care is associated with significantly decreased hospital costs and more efficient resource use for neonates requiring cardiac surgery.
Introduction
Congenital heart defects have the highest resource use of all birth defects.1 Perioperative care for neonatal congenital heart surgery is especially resource intensive, as it requires highly specialized teams with specific knowledge and skill sets. For this reason, much investigation is directed at optimizing care delivery to provide the highest value care.2–5
Location of perioperative care of infants with congenital heart disease (CHD) varies both between and within institutions for many reasons including center surgical volume, institutional and personnel resources, institutional preferences, financial considerations, historical precedent and/or political factors. Studies evaluating the impact of intensive care unit (ICU) specialization on outcomes, specifically mortality, in pediatric heart surgery have had conflicting results. One multi-institutional study suggested decreased mortality for neonates cared for in a dedicated cardiac intensive care unit (CICU), while another found no association in older children undergoing cardiac surgery.6,7 We recently published our institutional findings that admission to a dedicated CICU was associated with lower hospital costs and resource use compared to a neonatal intensive care unit (NICU) admission for prenatally diagnosed neonates requiring congenital heart surgery.8
The influence of location of care, a proxy for care specialization, on costs and outcomes in congenital surgery across pediatric hospitals with differing structures and surgical volumes requires further study, particularly to determine if our center findings are genenerazible across institutions with differening models of care. The purposes of this study are: 1) to assess the association between admission location and hospital costs and resource use, and 2) to explore a possible association of admission location with mortality in neonates undergoing congenital heart surgery across multiple institutions.
Patients and Methods
Study Design
This study analyzed retrospective observational cohort data from the Pediatric Hospital Information Systems (PHIS) database. PHIS is an administrative dataset that contains information regarding inpatient admission from 44 pediatric hospitals, representing 17 metropolitan areas. When compared to clinical registry data, the PHIS database captures approximately 86% of cardiac surgery cases that fall into a Risk Adjusted Congenital Heart Surgery (RACHS-1) category9, 11 Data available include demographics, dates of admission and discharge, vital status at discharge and billing data for medications, laboratory tests, surgical procedures, imaging procedures, clinical services and supplies. Data are de-identified at the time of data collection and reviewed for reliability and validity prior to inclusion. A collaboration between the Children’s Hospital Association and participating hospitals ensures data quality. Given the nature of this data, institutional board review was not required.10
Study Population
Neonates undergoing cardiac surgery at age <30 days with a RACHS-1 score of 2–611 at a PHIS hospital from January 2004 through December 2013 were included. Cardiac surgeries in infants <30 days by definition must be classified higher than RACHS-1 score 1. Infants initially admitted to any unit other than an ICU were excluded (5%). Patients were identified for inclusion in the study population using cardiac surgery procedure codes.
Data Collection
Study variables were gathered from International Classification of Diseases, Ninth-Revision (ICD-9) and billing codes. Data collected included demographic information, patient characteristics, risk factors, center data and charge/cost data. Demographic data included gender, race and payor. Patient characteristics included admission unit, admission age, admission year, use of prostaglandins and RACHS-1 score. Risk factors including prematurity, low birth weight, the presence of genetic or major non-cardiac abnormality and heterotaxy syndrome were collected using the pre-specified ICD-9 codes listed in Appendix 1 of the online supplement. Complications, including chylothorax, cardiac arrest, use of extracorporeal membrane oxygenation, medical complications, surgical complications and infections were collected using a combination of flags defined in the PHIS system and ICD 9 codes.12 A complete list of the findings flagged as medical complications, surgical complications or infections can be found in Appendix 1 (online supplement).12
Center-level variables included hospital census region and volume of cardiac cases per year. Volume of cardiac cases was calculated using the total number of neonatal cases reported per hospital divided by the number of years reported. Hospital volume was divided into tertiles for the purposes of analysis. Total and category specific charge data were also collected. Data from hospitals missing charge-to-cost ratios (3/44, 7%) were excluded from the cost analysis.
Exposure
The primary exposure of interest was the initial admission location. Admission location was classified as CICU, NICU, or pediatric ICU (PICU) based on the charges documented on the first day of admission at the pediatric hospital. Infants with CICU charges on the initial day of admission were categorized as CICU. Upon detailed examination, some institutions with known CICUs used charge codes for PICU. Therefore, those coded as PICU in PHIS likely represent a mix of institutions with multidisciplinary PICUs and institutions with dedicated CICUs that charge as “PICU.” For this reason, and since “PICU” designated admissions represented only 13% of the cohort, we primarily examined the comparison of the PHIS classified CICU to NICU admissions. We did perform a secondary analysis comparing PHIS classified PICU to NICU admissions separately. While specialization of care actually provided in each unit cannot be known in an administrative dataset, we would generally expect that the presence of cardiac intensive care trained physicians, pediatric cardiology, pediatric cardiothoracic surgical and specially trained nurses and their influence on management would be greatest in a CICU, least in NICUs, and somewhere in between in PICUs, given their heterogeneity in this dataset.13, 14
We know some patients initially admitted to one ICU will transfer preoperatively or postoperatively to another type of ICU. However, the exposure of initial admission location and resulting care decisions will influence downstream outcomes and resource use. For example, a cardiac arrest or infection in the initial unit would likely influence health status and resource use thereafter, even after transfer. Thus, to reduce bias, location of initial admission was treated similar to an “intention to treat” variable in a randomized control trial by assigning patients to their initial admission ICU exposure cohort. Nonetheless, for transparency and completeness we also summarized transfers for the cohort throughout the hospitalization.
Outcomes
The primary outcome analyzed was total hospital cost. Total hospital cost was calculated by multiplying total hospital charges per patient by hospital specific charge-to-cost ratios. Costs were also adjusted for geographic area and for inflation to 2013 dollars using the Health Care Finance Administration wage/price index and the All Urban Consumers Consumer Price Index.15,16 Charge data from subcategories including imaging, clinical, lab, pharmacy, supply and other charges were collected. These charges were not converted to costs because category specific charge to cost ratios are not available by hospital.
Secondary outcomes of interest included preoperative LOS, total hospital LOS, ICU LOS, days of mechanical ventilation and mortality prior to discharge.
Covariates
Covariates for the adjusted analysis were chosen a priori based on hypothesized clinical significance and potential for confounding and interaction. Covariates included in the multivariable analyses included prematurity, low birth weight, gender, race, prostaglandin use, genetic or major non-cardiac abnormality, the presence of heterotaxy syndrome, RACHS-1 score, center volume (tertiles), payor and admission year.
Statistical analysis
Patient, center and hospital cost data were summarized. Continuous data were expressed as means with standard deviations or medians with interquartile ranges as appropriate. Categorical data were tabulated. Univariate p-values were calculated using chi-squared tests, Kruskal-Wallis test, univariate linear and logistic regression where appropriate. Generalized estimated equations were used to fit a series of multivariable linear regression models to relate the primary and secondary outcomes to the initial admission location after adjusting for patient-level covariates. We specified a gamma distributed outcome model with a log link function for each of these outcome variables. A working exchangeable correlation model was used to account for clustering by hospital. The point estimate of the effect attributable to the exposure compared to the NICU as referent was determined. The secondary outcome of mortality prior to discharge was analyzed using generalized estimating equations under a multivariable logistic regression model for a binary outcome with a logit link, again using a working exchangeable correlation model to account for clustering by hospital.
To better understand the interaction of age at admission and admission location on outcome, each of the above analyses were repeated within each of the following age categories: ≤1 day, 2–7 days, 8–14 days and >15 days at admission. The group admitted at ≤1 day of age was believed to represent a mix of infants diagnosed prenatally or immediately after birth with CHD, as well as a group of infants more likely to be admitted to a NICU for reasons that may have not been fully captured by our covariates, such as uncoded prematurity or growth restriction. We also evaluated the interaction between center volume and admission location. The association of charges by discrete category and admission location were evaluated using a generalized estimating equation. All models were adjusted for predetermined covariates as discussed above.
Three additional sensitivity analyses were performed for the primary outcome of total hospital cost. To examine the robustness of our results and the impact of inherent differences in the population that might impact choice of initial admission unit, first, a propensity score analysis with one to one matching for initial admission to the CICU versus the NICU was performed. Secondly, to specifically exclude the impact of prematurity one of the main factors that would affect admission unit, a sensitivity analysis of total cost excluding patients who categorized as premature was performed. Lastly, while an overall estimate of resource use requires inclusion of all patients regardess of outcome, we also performed a a sensitivity analysis of total cost excluding patients with in-hospital mortality given the variable effect of mortality on cost.
Two sided p-value <0.05 was considered significant. Data analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA) and Stata 12.1 (StataCorp. 2011. Stata Statistical Software: Release 12.1 College Station, TX: StataCorp LP).
Results
A total of 19,984 patients met study criteria; 39% (7,830) were initially admitted to a CICU, 48% (9,493) to a NICU, and 13% (2,661) to a PICU. Descriptive data for the cohort are listed in Table 1. Patients admitted to the NICU were more likely to be premature, have low birth weight and/or have genetic or other major non-cardiac abnormalities. Twenty percent of patients were transferred from their initial admission ICU to another prior to surgery with a larger percentage of primarily non-CICU patients moving after surgery (Table 2). Patients admitted to the NICU spent a median of 55% of their total ICU time in the NICU, compared to a median of 0% NICU time in both CICU admits (p<0.001) and PICU admits (p<0.001). On univariate analysis, complications also varied between units, with a significantly higher percentage of infectious complications and chylothorax found in NICU admissions compared to the CICU admissions (Table 1).
Table 1.
Univariate demographic, covariate variables, and outcome measures by admission location.
| Variable/Outcome | NICU n=9,493 (48%) | CICU N=7,830 (39%) | PICU N=2,661 (13%) | p-value |
|---|---|---|---|---|
| Demographics | ||||
| Male gender, n (%) | 5,669 (60%) | 4,675 (60%) | 1,593 (60%) | 0.9 |
| Race, n (%) | <0.001 | |||
| Non-Hispanic black | 859 (9%) | 870 (11%) | 201 (8%) | |
| Hispanic | 1,451 (15%) | 1,038 (13%) | 556 (21%) | |
| Non-Hispanic white | 4,772 (50%) | 3,964 (51%) | 1,387 (52%) | |
| Asian | 134 (1%) | 65 (1%) | 24 (1%) | |
| Missing or other | 2,277 (24%) | 1,893 (24%) | 493 (19%) | |
| Payor | <0.001 | |||
| Private | 3,734 (39%) | 2,816 (36%) | 975 (37%) | |
| Government | 4,595 (48%) | 3,532 (45%) | 1,423 (53%) | |
| Other | 1,164 (12%) | 1,482 (19%) | 263 (10%) | |
| Patient Characteristics | ||||
| Age at admission, days, mean ± SD | 1.9±3.9 | 3.2±5.5 | 5.9±7.59 | <0.001 |
| Age at admission, days, Median (IQR) | 0 (0,2) | 0 (0,3) | 0 (0,9) | <0.001 |
| Age at admission, n (%)* | <0.001 | |||
| 0–1 days | 6,996 (74%) | 4,858 (62%) | 1,284 (48%) | |
| 2–7 days | 1,706 (18%) | 1,778 (23%) | 543 (20%) | |
| 8–14 days | 556 (6%) | 687 (9%) | 412 (15%) | |
| 15+ days | 235 (2%) | 506 (6%) | 422 (16%) | |
| Admission year, n (%) | <0.001 | |||
| 2003 | 85 (1%) | 43 (1%) | 12 (0.5%) | |
| 2004 | 858 (9%) | 665 (8%) | 234 (9%) | |
| 2005 | 934 (10%) | 736 (9%) | 295 (11%) | |
| 2006 | 1,022 (11%) | 891 (11%) | 305 (11%) | |
| 2007 | 1,097 (12%) | 830 (11%) | 283 (11%) | |
| 2008 | 1,176 (12%) | 795 (10%) | 211 (8%) | |
| 2009 | 1,037 (11%) | 658 (8%) | 213 (8%) | |
| 2010 | 967 (10%) | 677 (9%) | 192 (7%) | |
| 2011 | 875 (9%) | 815 (10%) | 318 (12%) | |
| 2012 | 756 (8%) | 861 (11%) | 301 (11%) | |
| 2013 | 686 (7%) | 859 (11%) | 297 (11%) | |
| Prostaglandin use, n (%) | 7,428 (78%) | 6,146 (78%) | 2,076 (78%) | 0.9 |
| RACHS-1 score | <0.001 | |||
| 2 | 1,979 (21%) | 1,566 (20%) | 736 (28%) | |
| 3 | 3,505 (37%) | 2,920 (37%) | 880 (33%) | |
| 4 | 2,620 (28%) | 1,958 (25%) | 696 (26%) | |
| 5 | 55 (1%) | 44 (1%) | 21 (1%) | |
| 6 | 1,334 (14%) | 1,342 (17%) | 328 (12%) | |
| Risk Factors | ||||
| Prematurity†, n (%) | 1,428 (15%) | 720 (9%) | 195 (7%) | <0.001 |
| Low birth weight†, n (%) | 1,196 (12%) | 567 (7%) | 149 (6%) | <0.001 |
| Other genetic or major non-cardiac defect, n(%) | 1,409 (15%) | 884 (11%) | 263 (10%) | <0.001 |
| Heterotaxy syndrome, n (%) | 601 (6%) | 358 (5%) | 117 (4%) | <0.001 |
| Center data | ||||
| Census region, n (%) | <0.001 | |||
| Midwest | 2,866 (30%) | 886 (11%) | 784 (29%) | |
| Northeast | 1,436 (15%) | 1,912 (24%) | 237 (9%) | |
| South | 2,713 (29%) | 2,853 (36%) | 1,058 (40%) | |
| West | 2,478 (26%) | 2,179 (28%) | 582 (22%) | |
| Center volume, n (%) | <0.001 | |||
| Lowest tertile, <50 cases/ yr | 3,146 (33%) | 2,104 (27%) | 760 (29%) | |
| Middle tertile, 50–75 cases/yr | 3,095 (33%) | 4,958 (63%) | 372 (14%) | |
| Highest tertile, > 75 cases/ yr | 3,252 (34%) | 768 (10%) | 1,529 (57%) | |
| Complications/Infections | ||||
| Chylothorax, n (%) | 558 (6%) | 290 (4%) | 123 (5%) | <0.001 |
| Medical complications, n (%) | 165 (2%) | 108 (1%) | 24 (1%) | 0.004 |
| Surgical complications, n (%) | 4,720 (50%) | 3,811 (49%) | 1,213 (46%) | 0.001 |
| Infectious complications, n (%) | 3,678 (39%) | 2,186 (28%) | 787 (30%) | <0.001 |
| ECMO, n (%) | 630 (7%) | 556 (7%) | 210 (8%) | 0.07 |
| Outcomes | ||||
| Total hospital cost, median (IQR)‡ | $106,470 ($57,830, $185,500) | $100,202 ($63,890, 169,410 | $118,480 ($75,110, $201,410) | <0.001 |
| Total hospital LOS, median (IQR) | 24 (15,42) | 17 (11,30) | 18 (10,33) | <0.001 |
| ICU LOS, median (IQR) | 18 (11,34) | 11 (6,20) | 13 (7,23) | <0.001 |
| Days of mechanical ventilation, median (IQR) | 6 (3,13) | 6 (3,10) | 6 (3,11) | <0.001 |
| Preoperative LOS, median (IQR) | 5 (2,8) | 3 (1,5) | 3 (1,6) | <0.001 |
| Postoperative LOS, median (IQR) | 19 (10, 35) | 14 (8, 26) | 14 (7,28) | <0.001 |
| Mortality, n (%) | 830 (9%) | 600 (8%) | 224 (8%) | 0.04 |
age of admission not available for 1 CICU subject.
definition available in Appendix 1.
total cost available for 18,065 (90%) of total cohort
NICU - Neonatal intensive care unit, CICU - cardiac intensive care unit, PICU - Pediatric intensive care unit, SD – standard deviation. RACHS-1 – risk adjusted congenital heart surgery, ECMO- extracorporeal membrane oxygenation, IQR – interquartile range, ICU – intensive care unit, LOS – length of stay
Table 2.
Movement of patients from initial ICU prior to and after cardiac surgery.
| Initial ICU of admission | N (%) Transfer to another ICU in first 48 hours | N (%) Transfer to another ICU any time prior to surgery | N (%) Transfer to another ICU after surgery |
|---|---|---|---|
| NICU | 949 (10%) | 1385 (15%) | 7,464 (79%) |
| CICU | 57 (1%) | 124 (2%) | 616 (8%) |
| PICU | 74 (3%) | 93 (3%) | 543 (19%) |
ICU – intensive care unit, NICU - Neonatal intensive care unit, CICU - cardiac intensive care unit, PICU - Pediatric intensive care unit
Total Hospital Cost
Cost data were available for 18,065 (90%) patients. Those with cost data were more likely to have government insurance, to be admitted at a later era, to be from the south, admitted to a lower volume center and admitted to a NICU. With regard to resource use, patients with cost data available had slightly longer hospital and ICU LOS and days of mechanical ventilation compared to those without. (Appendix 2, Table 1, online supplement). Median total hospital cost for the surgical hospitalization was $110,240 (Interquartile range (IQR) $68,790 – 187,920). In adjusted models stratified by age at admission, initial admission at ≤1 day of age to a CICU compared to a NICU was associated with no difference in cost. Initial admission to the CICU was associated with lower total costs compared to the NICU at 2–7 days of age (↓$20,440) and 8–14 days of age (↓$23,700) with no significant difference at >15 days of age (Figure 1). The full multivariate model is available in the online supplement, Appendix 2, Table 2. The propensity score model confirmed the results of the primary model (Appendix 2, Table 3A, online supplement). When premature infants were excluded, the results were similar for infants admitted at age 2–7 days with change in the same direction for those 8–14 days (Appendix 2, Table 3Bonline supplement).
Figure 1.
Adjusted estimated mean of total hospital cost by unit of admission stratified by age at admission. Blue – CICU, Green – PICU, Red – NICU admission. NS = Not significant, *p<0.05, †p<0.01, ‡p≤0.001 for comparison to NICU as referent.
When patients who died prior to discharge were excluded from the analysis, the findings were similar with the exception of the following: there was no longer a significant difference in cost between admission to the CICU and NICU at 2–7 days of age (online supplement, Appendix 2, Table 3C).
We found a significant interaction between admission location and center volume with respect to total hospital cost (Table 3). When stratified by age at admission and center volume, admission to the CICU was associated with a lower total hospital cost for all ages at admission compared to the NICU for the highest center volume tertile. At the lowest volume tertile, there was no significant difference in costs between the CICU and NICU except for an isolated finding of higher total hospital costs with admission to the CICU for those ≤1 day of age at admission.
Table 3.
Adjusted assessment of the interaction between center volume, age of admission and admission location on total cost. *NICU referent
| Lowest center volume tertile <50 cases/year | Middle center volume tertile 50–75 cases/year | Highest center volume tertile >75 cases/year | |||||
|---|---|---|---|---|---|---|---|
| Age at admission | Adjusted difference | p-value | Adjusted difference | p-value | Adjusted difference | p-value | |
| 0–1 days | CICU | $87,790 | <0.001 | −$5,720 | 0.6 | −$29,070 | <0.001 |
| 2–7 days | CICU | $40,470 | 0.2 | −$19,140 | 0.2 | −$28,330 | <0.001 |
| 8–14 days | CICU | $7,180 | 0.8 | −$37,470 | 0.03 | −$23,480 | 0.02 |
| >15 days | CICU | −$1,050 | >0.9 | $17,000 | 0.4 | −$32,310 | 0.008 |
NICU - Neonatal intensive care unit, CICU - cardiac intensive care unit
Charges by category were evaluated in order to assess the source of cost differences. Adjusted differences in charges by category and admission unit and age are shown in Figure 2. Room charges had the largest magnitude difference, with lower charges for those admitted to the CICU compared to the NICU. Other statistically significant differences were found in pharmacy, imaging and clinical categories (Figure 2).
Figure 2.
Adjusted comparison of difference in charge by category and admission location. For each category, data for CICU and PICU relative to NICU (referent) is displayed. Solid bars- CICU, Open bars – PICU. Blue 0–1 days of age at admission, Red 2–7 days of age at admission, Green 8–14 days of age at admission
Secondary outcome: Measures of hospital resource utilization
Median hospital LOS for all 19,984 patients was 21 days (IQR 13–36) with a median ICU stay of 14 days (IQR 3–12). Median days of mechanical ventilation for the cohort were 6 days (IQR 3–12), and median preoperative LOS was 4 days (IQR 1–6). In univariate analysis, those admitted to the CICU had shorter postoperative LOS than those admitted to the NICU. In adjusted models, admission to the CICU was associated with shorter hospital LOS and ICU LOS compared to the NICU in all age groups (Table 4). In addition, admission to the CICU was associated with a shorter preoperative LOS than admission to the NICU in the three youngest age groups. Those admitted to the CICU had fewer days of mechanical ventilation at all age groups except those admitted at >15 days of age.
Table 4.
The adjusted effect of admission location on secondary measures of resource utilization. *NICU referent
| Total Hospital LOS | Total ICU LOS | Preoperative LOS | Days of mechanical ventilation | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Age at admission |
Adjusted difference |
95% CI | p-value | Adjusted difference |
95% CI | p- value |
Adjusted difference |
95% CI | p- value |
Adjusted difference |
95% CI | p- value |
|
| 0–1 days | |||||||||||||
| CICU | −5.4 | −7.4, −3.5 | <0.001 | −5.4 | −7.2, −3.6 | <0.001 | −1.9 | −2.2, −1.6 | <0.001 | −2.8 | −4.3, −1.4 | <0.001 | |
| PICU | 1.0 | −1.6, 3.6 | 0.4 | 0.2 | −2.0, 2.5 | 0.8 | −1.2 | −1.6, −1.0 | <0.001 | 0.7 | −1.2, 2.5 | 0.5 | |
| 2–7 days | |||||||||||||
| CICU | −8.1 | −10.9, −5.7 | <0.001 | −7.7 | −10.1, −5.3 | <0.001 | −2.1 | −2.6, −1.7 | <0.001 | −2.5 | −3.9, −1.1 | 0.001 | |
| PICU | −6.7 | −9.5, −3.9 | <0.001 | −6.8 | −9.4, −4.3 | <0.001 | −1.7 | −2.2, −1.3 | <0.001 | −2.7 | −4.3, −1.2 | 0.001 | |
| 8–14 days | |||||||||||||
| CICU | −8.0 | −11.1, −4.8 | <0.001 | −9.2 | −12.5, −6.0 | <0.001 | −1.6 | −2.1, −1.1 | <0.001 | −2.7 | −4.7, −0.7 | 0.008 | |
| PICU | −6.4 | −9.8, −3.0 | <0.001 | −5.5 | −8.8, −2.2 | 0.001 | −1.3 | −1.9, −0.8 | <0.001 | −1.9 | −4.0, 0.3 | 0.1 | |
| >15 days | |||||||||||||
| CICU | −9.2 | −14.3, −4.1 | <0.001 | −6.4 | −11.0, −1.9 | 0.006 | −0.5 | −1.0, 0.0 | 0.06 | −2.8 | −6.5, 1.0 | 0.2 | |
| PICU | −8.6 | −13.8, −3.4 | 0.001 | −4.9 | −9.5, −0.3 | 0.04 | −0.8 | −1.3, −0.3 | 0.001 | −2.7 | −6.4, 1.1 | 0.2 | |
LOS – length of stay, NICU - Neonatal intensive care unit, CICU - cardiac intensive care unit, PICU - Pediatric intensive care unit
Secondary outcome: Hospital Mortality
Overall hospital mortality for the cohort was 1,654/19,984 (8.3%). In adjusted models stratified by age, there was no difference in hospital mortality between admission to the CICU or NICU at any age (p=0.4–>0.9).
PICU analyses
When the PICU group was compared to the NICU for the primary and secondary outcomes, findings were similar to the CICU vs. NICU results described above. Initial admission to the PICU was associated with lower total costs compared to the NICU at 2–7 days of age (↓$26,430) and 8–14 days of age (↓$25,910) with no significant difference at >15 days of age (Figure 1). In adjusted models, admission to the PICU was associated with shorter hospital LOS and ICU LOS compared to the NICU in all age groups except those <1 day of age at admission (Table 4). Similar to CICU admission, admission to the PICU was associated with a shorter preoperative LOS than admission to the NICU in the three youngest age groups. Admission to the PICU was associated with a higher mortality than admission to the NICU in those admitted ≤1 day of age (Adjusted odds ratio 1.4, 95% CI 1.04–1.82) with no significant difference in mortality when compared to the NICU at the other age groups.
Discussion
In a multicenter analysis of pediatric hospitals across the nation, admission to an ICU specialized in delivering cardiac care is associated with significantly decreased total hospital costs and more efficient resource use for neonates undergoing cardiac surgery. These findings remained significant even after adjusting for hospital surgical volume as a potential confounder, since high volume centers are more likely to have specialized CICUs. Adjusted analysis revealed no significant difference in mortality.
This multicenter study builds on our previous single center findings. In the single center study, we found that in term, prenatally diagnosed infants, CICU admission was associated with shorter hospital and ICU LOS and fewer days of mechanical ventilation, but no difference in total cost when compared to the NICU.8 In this multicenter analysis with larger sample size, we found a decreased total cost between CICU and NICU admissions. When the effect size is compared to the median total cost for our population, admission to the CICU was associated with a savings of approximately 18%. This represents a significant opportunity for streamlining resource use in this resource-heavy population.1 While determining the specific drivers of these cost differences is challenging and beyond the scope of this analysis, differences in lengths of stay as reflected in differences in room charges likely contribute. We also saw differences in other clinical and pharmacy charges. Furthermore, other factors such as variation seen in the complication rates between units may contribute as well. Interestingly, our current findings contrast those of Gupta et al, which suggested that total charges are higher in patients cared for in a CICU versus NICU.6 They also found a significant difference in mortality favoring the CICU while we did not. The difference in these findings may be due to their inclusion of only propensity-matched pairs of patients who did not transfer from one unit to another preoperatively, limiting their analysis to ~12% of the population of neonates undergoing cardiac surgeries at PHIS hospitals. Because transfers may have been influenced by changes in the patients’ medical conditions following their initial ICU admission, this restriction may have introduced selection biases that affected their results. Of note, in the Gupta paper, 20% of patients transferred ICUs prior to surgery, similar to our findings. However, these patients were excluded in the Gupta analysis. Furthermore, the inclusion of only 1:1 matched patients excluded >85% of the neonatal CHD surgery population and significantly limits the ability to interpret these findings and use apply the findings to the typical clinical setting. In our analysis, we included a larger selection of the PHIS database hospitals (44/48) and patients (95% of those undergoing congenital heart surgery), allowing for a more generalizable sample. In addition, in our cohort, patients who were initially admitted to the NICU spend the majority of their ICU time in the NICU and since there were no transfers from another unit into the NICU after admission, these patients represent a group of patients with time in the NICU as their unique exposure.
We describe significant differences in resource use by admission location that we speculate are due to specialization of care. Admission to a CICU was associated with shorter hospital stays and fewer days of mechanical ventilation in the previous studies.6,8 Our study further suggests that intensive care specialization may be advantageous by reducing cost and resource use in neonates undergoing congenital heart surgery. Specialization at the hospital level has been shown to decrease costs17, so it is not surprising that unit-specific specialization would streamline resource use. While the natural question is whether the NICU patient population differed in ways that we could not account for statistically leading to some selection bias (or confounding by indication), is remains notable that our findings remained robust in the propensity matched analysis and when premature infants were excluded. Finally, the fact that these findings are aligned with our previous single center work, where the granularity of our data allowed us to adjust for these confounding indications more adequately, add to the likelihood that this difference in cost is not solely related to inherent differences in the patient populations cared for by these different units. Additionally, the fact that some cost differences based on admission unit were no longer present when patients who died prior to discharge were excluded suggests that, in some instances, cost savings may be due to a difference in timing of deaths between units or a difference in resources used in the highly resource intensive period prior to death.
Since high volume centers often those more likely to have specialized CICUs, understanding interaction between center volume and age at admission is important. The association between center volume and cost in CHD and between higher center volume and improved outcomes has been reported.3,18–20 The interaction between hospital volume and admission location may reflect not just the hospital volume but a difference in the specialization of ICU care related to volume. At lower volume centers, ICU specialization may be systematically different than at higher volume centers. This may explain the finding of resource savings for CICU admissions within the highest volume centers but the lack of similar findings in lower volume centers. Likely, as our data suggest, in higher volume centers, ICU specialization contributes to achieving lower cost care and improved outcomes. Alternatively, higher volume may result in superior specialization that cannot be achieved or sustained at a similar level when the volume of CHD patients results in fewer CICU admissions. It may not be possible, for all centers, however to support an individual specialized unit. We would speculate that our study supports specialization of care within any unit, not necessarily only initial admission to a CICU. For example, in smaller centers that cannot support a separate CICU, the best answer may be providing cardiac specific training to a subset of NICU nurses and physicians to attempt to approximate the specialization provided in a CICU.
As with any retrospective study, the possibility of uncontrolled confounding exists. Specifically, we do not have information regarding the physician decision-making process with respect to admission unit, or any other measures of quality such as accreditation by professional centers. However, we did control for known covariates that increase likelihood of admission to the NICU that also impact outcome and our findings remained robust on multiple sensitivity analyses designed to further address this confounding. Nonetheless, given our use of administrative data, unmeasured confounders may still exist. As birth weight in grams and gestational age in weeks are frequently missing or inaccurate in PHIS, ICD-9-DM codes were used, as has been done in previous rigorous studies using PHIS.21 It is possible that transfer from the initial unit of admission may affect our analysis. Our intent was to determine the effect of initial location of admission on resource, and so patients were analyzed by their initial location of admission as in an intent-to-treat analysis. As transfers from one ICU to another are likely heavily influenced by the patient’s condition at the time of transfer, adjusting for those time-dependent factors is difficult, and can also introduce bias. Additionally, the direction of transfer was almost exclusively from the NICU to another ICU. This supports that initial admission to a NICU has a significant effect on cost and resource use, and that the location of initial admission and preoperative care has a significant impact on the remainder of the hospital course. In addition, isolating cost by each location of care after the initial exposure to an ICU and adjusting for those time-dependent factors is difficult, and would introduce further bias. Most likely, the lack of significance in the >15 days age group is related to a relatively small number of patients in this group. Another limitation is that as the level of cardiac specialization within the group billed as PICU is not completely known, it likely represents a mix of general pediatric ICUs and dedicated cardiac ICUs. For that reason, we analyzed this group separately with only slight differences in the results as noted above that likely are secondary to their mixed level of specialization. However, it does make makes the comparison of this group to the NICU difficult to interpret. Additionally, we do not have any information on care that occurred prior to admission or transfer to the PHIS hospital.While PHIS data quality is rigorously confirmed through the Children’s Hospital Association and submitting hospitals, the possibility of misclassification exists and cannot be quantified.
Conclusions
Our results demonstrate that in a national cohort of pediatric hospitals, initial admission to an ICU specialized in delivering cardiac care results in significantly decreased hospital costs and more efficient resource use for neonates requiring cardiac surgery. While mortality was not shown to be different, in the current era of capitated payments and efforts to provide value-based care, cost reduction and more efficient resource use is key while optimizing outcome. This study indicates the importance of increased cardiac specialization and sufficient patient volume to maintain specialization within the constraints of physical location in the care of these patients.
Central picture.
Resource use decreases with CICU admission in neonatal heart surgery (modeled figure).
Video.
Dr. Joyce Johnson, MD MS describes the key findings and relevance of this multicenter observational study.
Central Message.
Admission to an ICU specializing in cardiac care is associated with significantly decreased hospital costs and more efficient resource use for neonates requiring cardiac surgery.
Perspective Statement.
This study shows that a modifiable factor, admission to a unit with cardiac specialization, was associated with decreased resource use in infants undergoing congenital heart surgery in an analysis of an administrative, multi-center dataset. By specializing care for these infants, there is an opportunity to improve outcomes and limit cost, increasing value in this population.
Acknowledgments
The authors would like to acknowledge the contributions of Dr. Tom Greene for his statistical expertise and Dr. Lajja Desai for her editorial assistance.
Funding sources: This investigation was supported by the University of Utah Study Design and Biostatistics Center (Salt Lake City, UT), with funding in part from the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant 8UL1TR000105 (formerly UL1RR025764).
Abbreviations
- CHD
congenital heart disease
- ICU
intensive care unit
- CICU
cardiac intensive care unit
- PICU
pediatric intensive care unit
- NICU
neonatal intensive care unit
- PHIS
Pediatric Health Information Systems
- RACHS-1
Risk Adjusted Congenital Heart Surgery
- ICD-9
International Classification of Diseases, Ninth-Revision
- LOS
length of stay
- IQR
interquartile range
Appendix 1
ICD9 codes:
Chylothorax: 457.8
Prematurity: for extreme immaturity (765.0), other preterm infants (765.1), or weeks of gestation <37 weeks (765.26, 765.28).
Low birth weight was defined as the presence of ICD-9 codes for slow fetal growth and fetal malnutrition (764) and low birth weight status (V21.3).
Genetic or other major non-cardiac abnormality was defined as the presence of ICD-9 codes for known and other genetic syndromes and congenital anomalies (279.11, 279.12, 279.13, 756.16, 758.0–9, 759.7–9, 259.4, 737.3, 756.0–7, 553.3).
Heterotaxy syndrome was defined as the presence of ICD-9 codes for congenital anomalies of the spleen (759.0), situs inversus (759.3), or other anomalies of the great veins (747.49).
Surgical, medical and infection complications flags as defined by PHIS can be found through the children’s hospital association website: https://www.childrenshospitals.org/~/media/Files/Groups/PHIS/Reference%20Resources/Data%20Content/Other%20Materials/FlagsInfection%20Med%20Comp%20Surg%20Comp%20Current%20List.pdf
Figure 1.
Diagram of cohort inclusion
Appendix 2
Table 1.
Comparison of those with and without cost data.
| Covariates and outcomes | Observations with hospital cost (n=18,065, 90.4%) | Observations without hospital cost (n=1,919, 9.6%) | P value |
|---|---|---|---|
| Patient demographics | |||
| Male , n (%) | 10,808 (60) | 1129 (59) | 0.40 |
| Race, n (%) | <0.0001 | ||
| Non-Hispanic White | 9,170 (51) | 953(50) | |
| Non-Hispanic Black | 1,623(9) | 307(16) | |
| Hispanics | 2,818 (16) | 227(12) | |
| Asian | 210 (1) | 13 (1) | |
| Missing/other | 4,244 (23) | 419 (22) | |
| Payer [(n (%)] | <0.0001 | ||
| Government | 8,824 (49) | 726 (38) | |
| Private | 6,628 (37) | 897 (47) | |
| Other | 2,613 (14) | 296 (15) | |
| Patient characteristics | |||
| Age at admitted (days) median, (range) | 1(0,29) | 0 (0, 29) | <0.0001 |
| Admitted year, n (%) | <0.0001 | ||
| 2003 | 107 (1) | 33 (2) | |
| 2004 | 1,205 (7) | 552 (29) | |
| 2005 | 1,742 (10) | 223 (12) | |
| 2006 | 2,048 (11) | 170 (9) | |
| 2007 | 2,123 (12) | 87 (5) | |
| 2008 | 2,055 (11) | 127 (7) | |
| 2009 | 1,736 (10) | 172 (9) | |
| 2010 | 1,694 (9) | 142(7) | |
| 2011 | 1,876 (10) | 132 (7) | |
| 2012 | 1,787 (10) | 131 (7) | |
| 2013 | 1,692 (9) | 150 (8) | |
| Prostaglandin use, n (%) | 12,093(67) | 1,322(69) | 0.08 |
| RACHS score, n (%) | 0.0001 | ||
| 2 | 3,919 (22) | 362 (19) | |
| 3 | 6,576 (36) | 729 (38) | |
| 4 | 4,804 (27) | 470 (24) | |
| 5 | 108 (1) | 12(1) | |
| 6 | 2,658 (15) | 346 (18) | |
| Risk factors | |||
| Premature, n (%) | 2,148 (12) | 195 (10) | 0.03 |
| Low birth weight, n (%) | 1,736 (10) | 176 (9) | 0.53 |
| Other congenital defect, n (%) | 2,329 (13) | 227 (12) | 0.18 |
| Heterotaxy, n (%) | 989 (5) | 87 (5) | 0.08 |
| Complications/Infections | |||
| Chylothorax, n (%) | 915 (5) | 56 (3) | <0.0001 |
| Medical complication, n (%) | 266 (1) | 31(2) | 0.62 |
| Surgery complication, n (%) | 8,796 (49) | 948 (49) | 0.55 |
| Infection, n (%) | 6,048 (33) | 603 (31) | 0.07 |
| EMCO, n (%) | 1,258 (7) | 138 (7) | 0.71 |
| Center data | |||
| Region, n (%) | <0.0001 | ||
| West | 5,018 (28) | 221 (12) | |
| South | 6,335 (35) | 289 (15) | |
| Northeast | 2,238 (12) | 1,347(70) | |
| Midwest | 4,474 (25) | 62 (3) | |
| Central volume, n (%) | <0.0001 | ||
| <50.1 | 5,187 (29) | 362 (19) | |
| 50.1 to 74.8 | 5,941 (33) | 69 (4) | |
| ≥74.8 | 6, 937(38) | 1,488 (78) | |
| Outcomes | |||
| Length of stay (days), median (IQR) | 21 (13,36) | 18 (11,36) | <0.0001 |
| Mechanical ventilation (days), median (IQR) | 6 (3,12) | 4 (2,10) | <0.0001 |
| ICU LOS (days), median (IQR) | 15 (8,27) | 13 (7,25) | <0.0001 |
| Pre-operative LOS, median (IQR) | 4 (1,7) | 3 (1,5) | <0.0001 |
| Mortality, n (%) | 1,476 (8) | 178 (9) | 0.09 |
| Admission locations, n (%) | <0.0001 | ||
| CICU | 6,430 (36) | 1,400 (73) | |
| NICU | 9,100 (50) | 393 (20) | |
| PICU | 2,535 (14) | 126 (7) | |
Table 2.
Estimated beta coefficients of the covariates in the generalized estimating equation multivariate model for the primary outcome of total hospital cost by age group of admission.
| Covariate | 0–1 days of age | p value | 2–7 days of age | p-value | 8–14 days of age | p-value | >15 days of age | p-value |
|---|---|---|---|---|---|---|---|---|
| Male gender | 0.04 | 0.04 | 0.09 | 0.01 | 0.1 | 0.02 | 0.2 | 0.003 |
| Race | <0.001 | 0.6 | 0.7 | 0.5 | ||||
| Non-Hispanic black | 0.2 | <0.001 | 0.08 | 0.2 | −0.04 | 0.7 | 0.2 | 0.1 |
| Hispanic | 0.05 | 0.2 | 0.006 | 0.9 | −0.08 | 0.3 | −0.09 | 0.5 |
| Non-Hispanic white | ref | ref | ref | ref | ||||
| Asian | −0.02 | 0.9 | 0.09 | 0.6 | 0.3 | 0.3 | −0.2 | 0.6 |
| Missing or other | 0.08 | 0.003 | −0.03 | 0.5 | 0.01 | 0.9 | 0.009 | 0.9 |
| Payor | 0.04 | 0.004 | <0.01 | 0.7 | ||||
| Private | −0.06 | 0.01 | −0.08 | 0.06 | −0.2 | 0.003 | −0.09 | 0.4 |
| Government | ref | ref | ref | ref | ||||
| Other | −0.03 | 0.03 | −0.2 | 0.002 | −0.1 | 0.1 | −0.06 | 0.6 |
| Admission year | <0.001 | <0.001 | <0.001 | 0.07 | ||||
| 2003 | ref | ref | ref | ref | ||||
| 2004 | −0.2 | 0.05 | −0.5 | 0.08 | −0.7 | 0.05 | −0.9 | 0.3 |
| 2005 | −0.3 | 0.006 | −0.5 | 0.05 | −0.6 | 0.08 | −0.9 | 0.2 |
| 2006 | −0.2 | 0.08 | −0.5 | 0.08 | −0.5 | 0.2 | −0.6 | 0.4 |
| 2007 | −0.2 | 0.2 | −0.4 | 0.1 | −0.6 | 0.09 | −0.8 | 0.3 |
| 2008 | −0.2 | 0.2 | −0.6 | 0.05 | −0.6 | 0.07 | −1 | 0.2 |
| 2009 | −0.09 | 0.5 | −0.5 | 0.08 | −0.5 | 0.2 | −0.9 | 0.3 |
| 2010 | 0.05 | 0.7 | 0.4 | 0.2 | −0.3 | 0.3 | −0.8 | 0.3 |
| 2011 | 0.06 | 0.6 | −0.1 | 0.6 | −0.2 | 0.6 | −0.7 | 0.4 |
| 2012 | 0.2 | 0.1 | −0.2 | 0.5 | −0.2 | 0.6 | −0.5 | 0.5 |
| 2013 | 0.1 | 0.3 | −0.2 | 0.4 | −0.3 | 0.4 | −0.5 | 0.5 |
| Prostaglandin use | −0.08 | 0.005 | 0.05 | 0.2 | 0.08 | 0.2 | 0.05 | 0.5 |
| RACHS-1 score | 0.1 | <0.001 | 0.2 | <0.001 | 0.3 | <0.01 | 0.4 | <0.001 |
| Prematurity | 0.1 | 0.004 | 0.1 | 0.06 | 0.4 | 0.002 | 0.3 | 0.2 |
| Low birth weight | 0.2 | <0.001 | 0.2 | 0.03 | −0.07 | 0.6 | −0.07 | 0.7 |
| Other genetic or major non-cardiac defect | 0.4 | <0.001 | 0.4 | <0.001 | 0.4 | <0.01 | 0.40 | .001 |
| Heterotaxy syndrome | 0.2 | <0.001 | 0.09 | 0.4 | 0.04 | 0.8 | 0.1 | 0.7 |
| Center volume | <0.001 | <0.001 | <0.001 | 0.03 | ||||
| Lowest tertile, <50 cases/ yr | ref | ref | ref | ref | ||||
| Middle tertile, 50–75 cases/yr | 0.1 | 0.2 | 0.03 | 0.7 | 0.08 | 0.5 | 0.009 | 0.9 |
| Highest tertile, > 75 cases/ yr | −0.4 | <0.001 | −0.4 | <0.001 | −0.4 | 0.001 | −0.4 | 0.02 |
Table 3.
Sensitivity analysis of the generalized estimating equation for the outcome of total hospital cost. A) using a propensity matching analysis with 1:1 matching between NICU and CICU (NICU referent) B) excluding patients with in-hospital mortality adjusted for the prespecified covariates, C) excluding patients with prematurity adjusted for the prespecified covariates.
| A. Total Hospital cost (propensity score model) | B. Total Hospital cost (excluding prematurity) | C. Total Hospital cost (excluding mortality) | ||||
|---|---|---|---|---|---|---|
| Age at admission/admission location | Adjusted difference | p-value | Adjusted difference | p-value | Adjusted difference | p-value |
| 0–1 days | ||||||
| CICU | $5,541 | 0.3 | $4,760 | 0.5 | $1,650 | 0.8 |
| PICU | -- | -- | $16,030 | 0.04 | $12,510 | 0.07 |
| 2–7 days | ||||||
| CICU | −$17,446 | 0.02 | −$20,300 | 0.002 | −$10,490 | 0.12 |
| PICU | -- | -- | −$25,590 | 0.01 | −$17,630 | 0.02 |
| 8–14 days | ||||||
| CICU | −$42,748 | <0.01 | −$14,670 | 0.1 | −$20,850 | 0.02 |
| PICU | -- | -- | −$21,560 | 0.04 | −$24,440 | <0.01 |
| >15 days | ||||||
| CICU | −$16,432 | 0.4 | $3,530 | 0.8 | −$3,290 | 0.8 |
| PICU | -- | -- | −$14,340 | 0.3 | −$14,140 | 0.3 |
NICU - Neonatal intensive care unit, CICU - cardiac intensive care unit, PICU - Pediatric intensive care unit
Footnotes
Potential Conflicts of Interest: The authors have no conflicts of interest relevant to this article to disclose.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Russo CA, Elixhauser A. Hospitalizations for birth defects, 2004. HCUP statistical brief #24. 2007 [PubMed] [Google Scholar]
- 2.Pasquali SK, Jacobs JP, Bove EL, Gaynor JW, He X, Gaies MG, et al. Quality-cost relationship in congenital heart surgery. Ann Thorac Surg. 2015;100:1416–1421. doi: 10.1016/j.athoracsur.2015.04.139. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Connor JA, Gauvreau K, Jenkins KJ. Factors associated with increased resource utilization for congenital heart disease. Pediatrics. 2005;116:689–695. doi: 10.1542/peds.2004-2071. [DOI] [PubMed] [Google Scholar]
- 4.Smith AH, Gay JC, Patel NR. Trends in resource utilization associated with the inpatient treatment of neonatal congenital heart disease. Congenit Heart Dis. 2014;9:96–105. doi: 10.1111/chd.12103. [DOI] [PubMed] [Google Scholar]
- 5.Chan T, Kim J, Minich LL, Pinto NM, Waitzman NJ. Surgical volume, hospital quality, and hospitalization cost in congenital heart surgery in the united states. Pediatr Cardiol. 2015;36:205–213. doi: 10.1007/s00246-014-0987-2. [DOI] [PubMed] [Google Scholar]
- 6.Gupta P, Beam BW, Noel TR, Dvorchik I, Yin H, Simsic JM, et al. Impact of preoperative location on outcomes in congenital heart surgery. Ann Thorac Surg. 2014;98:896–903. doi: 10.1016/j.athoracsur.2014.04.123. [DOI] [PubMed] [Google Scholar]
- 7.Burstein DS, Jacobs JP, Li JS, Sheng S, O’Brien SM, Rossi AF, et al. Care models and associated outcomes in congenital heart surgery. Pediatrics. 2011;127:e1482–1489. doi: 10.1542/peds.2010-2796. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Johnson JT, Tani LY, Puchalski MD, Bardsley TR, Byrne JL, Minich LL, et al. Admission to a dedicated cardiac intensive care unit is associated with decreased resource use for infants with prenatally diagnosed congenital heart disease. Pediatr Cardiol. 2014;35:1370–1378. doi: 10.1007/s00246-014-0939-x. [DOI] [PubMed] [Google Scholar]
- 9.Pasquali SK, Jacobs JP, Jacobs ML, Gaies MG, Shah SS, Hall M, et al. Measuring Hospital Performance in Congenital Heart Surgery: Administrative Versus Clinical Registry Data. Ann Thorac Surg. 2015;99:932–938. doi: 10.1016/j.athoracsur.2014.10.069. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Pediatric Health Information Systems (PHIS) 2014 [Google Scholar]
- 11.Jenkins KJ, Gauvreau K, Newburger JW, Spray TL, Moller JH, Iezzoni LI. Consensus-based method for risk adjustment for surgery for congenital heart disease. J Thorac Cardiovasc Surg. 2002;123:110–118. doi: 10.1067/mtc.2002.119064. [DOI] [PubMed] [Google Scholar]
- 12.Children’s Hospital Association. [Accessed on May 30, 2014]; http://www.childrenshospitals.org.
- 13.Daenen W, Lacour-Gayet F, Aberg T, Comas JV, Daebritz SH, Di Donato R, et al. Optimal Structure of a Congential Heart Surgery Department in Europe: by EACTS Congenital Heart Disease Committee. Eur J Cardiothorac Surg. 2003;24:343–351. doi: 10.1016/s1010-7940(03)00444-5. [DOI] [PubMed] [Google Scholar]
- 14.Baden HP, Zimmerman JJ, Brilli RJ, Wong H, Wetzel RC, Burns JP, et al. Intensivist-led team approach to critical care of children with heart disease. Pediatrics. 2006;117:1854–1856. doi: 10.1542/peds.2006-0353. [DOI] [PubMed] [Google Scholar]
- 15.O’Byrne ML, Gillespie MJ, Shinohara RT, Dori Y, Rome JJ, Glatz AC. Cost comparison of transcatheter and operative pulmonary valve replacement (from the pediatric health information systems database) Am J Cardiol. 2016;117:121–126. doi: 10.1016/j.amjcard.2015.10.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. [Accessed on May 30, 2014];All Urban consumers consumer price index. http://www.bls.gov/cpi.
- 17.Eastaugh SR. Hospital specialization and cost efficiency: benefits of trimming product lines. Hosp Health Serv Adm. 1992;37:223–235. [PubMed] [Google Scholar]
- 18.Kansy A, Ebels T, Schreiber C, Tobota Z, Maruszewski B. Association of center volume with outcomes: Analysis of verified data of european association for cardio-thoracic surgery congenital database. Ann Thorac Surg. 2014;98:2159–2164. doi: 10.1016/j.athoracsur.2014.07.065. [DOI] [PubMed] [Google Scholar]
- 19.Karamlou T, Jacobs ML, Pasquali S, He X, Hill K, O’Brien S, et al. Surgeon and center volume influence on outcomes after arterial switch operation: Analysis of the sts congenital heart surgery database. Ann Thorac Surg. 2014;98:904–911. doi: 10.1016/j.athoracsur.2014.04.093. [DOI] [PubMed] [Google Scholar]
- 20.Pasquali SK, Sun JL, d’Almada P, Jaquiss RD, Lodge AJ, Miller N, et al. Center variation in hospital costs for patients undergoing congenital heart surgery. Circ Cardiovasc Qual Outcomes. 2011;4:306–312. doi: 10.1161/CIRCOUTCOMES.110.958959. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lillehei CW, Gauvreau K, Jenkins KJ. Risk adjustment for neonatal surgery: A method for comparison of in-hospital mortality. Pediatrics. 2012;130:e568–574. doi: 10.1542/peds.2011-3647. [DOI] [PubMed] [Google Scholar]