National Practice Patterns for Management of Adult Congenital Heart Disease
Operation By Pediatric Heart Surgeons Decreases In-Hospital Death
Background— Surgery for grown-up (age ≥18 years) patients with congenital heart disease (GUCH) is frequently performed by surgeons without specialization in pediatric heart surgery. We sought to define national practice patterns and to determine whether outcomes for GUCH patients are improved if they are treated by specialized pediatric heart surgeons (PHSs) compared with non-PHSs.
Methods and Results— We identified index cardiac procedures in patients with 12 congenital heart disease diagnostic groups using the Nationwide Inpatient Sample 1988 to 2003. PHSs were defined as surgeons whose annual practice volumes were made of >75% annual pediatric heart cases. GUCH operations were defined as operations within these 12 diagnoses occurring in patients ≥18 years of age. We identified 30 250 operations, yielding a national estimate of 152 277±7875 operations. Of these, 111 816±7456 (73%) were pediatric operations, and 40 461±1365 (27%) were GUCH operations. PHSs performed 68% of pediatric operations in all diagnostic groups, whereas non-PHSs performed 95% of GUCH operations within the same diagnostic groups (P<0.0001). In-hospital death rates for GUCH patients operated on by PHSs were lower than death rates for GUCH patients operated on by non-PHSs (1.87% [95% CI, 0.62 to 3.13] versus 4.84% [95% CI, 4.30 to 5.38%]; P<0.0001). Survival advantage increased with increasing surgeon annual pediatric volume (P=0.0031).
Conclusions— Pediatric patients within specific diagnostic groups are more likely to undergo operation by PHSs, whereas GUCH patients within the same diagnostic groups are more likely to undergo operation by non-PHSs. In-hospital death rates are lower for GUCH patients operated on by PHSs. GUCH patients should be encouraged to obtain surgical operation by PHS.
Received March 24, 2008; accepted September 4, 2008.
Because of the improved outcomes for congenital heart surgery in the neonatal period, there are an increasing number of grown-up patients living with congenital heart disease (GUCH).1–5 The best estimates of the current population of GUCH patients came from the 32nd Bethesda Conference held in 2000, which projected that nearly 787 800 patients with congenital heart disease would survive to adulthood.1 Many GUCH patients, however, often require ongoing specialized care and surgical intervention, and despite the notion of adulthood, GUCH patients still offer unique challenges with respect to physiology and anatomic complexity for surgeons unfamiliar with congenital heart disease. Because of the unique challenges that characterize GUCH, there has been a growing focus on developing specialized centers dedicated to the care of this population.1-4,6–11 Many large academic centers have established so-called adult congenital centers, which offer comprehensive care by cardiologists and surgeons who are familiar with the pathology, anatomy, and social issues facing GUCH patients.1-4,10,11 Underlying this initiative to regionalize care is the belief, as pointed out by Webb,4 that surgeons who operate on adult congenital patients should have specialization as congenital heart surgeons as opposed to adult cardiac surgeons without further specialization in congenital heart disease. This belief is predicated on the unproven hypothesis that trained congenital heart surgeons have specialized technical expertise coupled with a greater understanding of the prerepair and postrepair anatomy and physiology that define GUCH patients. The institution of separate congenital heart surgery certification for fellowship graduates as of 2008, which mandates a discrete number of GUCH operations, is further impetus for an organized and uniform approach to “who does what.” We therefore undertook this study to determine whether patients with GUCH who are operated on by surgeons who specialize in congenital heart disease have lower in-hospital rates of death and decreased resource use compared with GUCH patients operated on by cardiac surgeons without congenital specialization.
Editorial p 2321
Clinical Perspective p 2352
Patient and Surgeons
The Nationwide Inpatient Sample (NIS) is a stratified, cross-sectional database that includes ≈20% of all community (nonfederal) hospital discharges in the United States. The NIS is managed under the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality.12 To ensure the representative nature of the database, the NIS is stratified by geographical region, hospital bed size, teaching status, urban versus rural location, and hospital ownership. The data for this study were derived from the combination of the NIS databases from 1988 to 2003, and our analysis used sampling weights provided within the NIS to derive national estimates. The sampling frame of the NIS has been nearly constant over time (eg, a stratified sample of 20% of US nonfederal hospital discharges), except for a modification in 1998, which excluded short-term rehabilitation hospitals from the sampling frame. To account for this small change, we used the revised NIS trend weights published by the Agency for Healthcare Research and Quality,12 which allow us to compare across years with the same effective sampling scheme. The NIS was searched for the individual years 1988 to 2003, selecting hospital discharges for which the primary International Classification of Diseases, 9th revision, clinical modification (ICD-9-CM) procedure codes were for all open heart or thoracic aortic (eg, coarctation of the aorta) operations (Table 1⇓). Using the methodology described by Elixhauser et al,13 we also abstracted comorbidities. Hospital discharges were excluded if the surgeon identifier was missing.
Using this initial procedural data set, we then incorporated 12 ICD-9-CM–coded diagnoses (Table 1⇑), adjudicated by consensus among the authors, that capture the vast majority of GUCH. GUCH operations were then defined as those operations within these specified 12 diagnoses performed in patients ≥18 years of age.
We then profiled individual surgeons by 2 methods. The first expressed concentration in pediatric heart surgery as a percentage of any individual surgeon’s annual cardiac case load. Pediatric heart surgeons were defined as those surgeons whose annual practice consisted of ≥75% pediatric heart surgery (eg, operations undertaken in patients <18 years of age). The second method quantified the absolute annual number of pediatric heart operations performed by individual surgeons as a continuous measure of increasing specialist experience. Next, to confirm our initial assumption that the nonpediatric heart surgeons were in fact surgeons performing adult cardiac surgery rather than surgeons who were performing primarily GUCH operations (eg, GUCH specialists), we reanalyzed our data set to subset surgeons on the basis of annual percent of GUCH-only operations and the annual percent of pediatric plus GUCH operations. Absolute GUCH case numbers and absolute numbers of GUCH plus pediatric operations were recorded. Because of the nature of the NIS, whereby different hospitals are sampled in any given year, surgeon classification was based on annual volume for each individual year (ie, year specific).
The primary outcome was in-hospital death; our secondary outcomes were length of stay (LOS) and total hospital charges. For each outcome, logistic and linear regression models adjusted for patient- and hospital-level characteristics were fit to estimate the attributable risk as a result of operation by a pediatric heart surgeon versus a nonpediatric heart surgeon and as a result of increasing pediatric heart surgery volume (coded as percent of pediatric operations). C indexes and receiver-operating characteristics curves were generated for all logistic regression models to provide insight into model discrimination. SAS software, version 9.1 (SAS Institute, Inc, Cary, NC), was used to fit the models to account for the survey design of the NIS, the potential clustering of outcomes within a hospital, and the decrease in clustering occurring over increasing time intervals. Our method of risk adjustment included fitting an initially saturated logistic regression model using the following variables: all comorbidity measures, female gender, patient age, surgeon category, hospital location and teachings status, and patient primary diagnostic group (tetralogy of Fallot [ToF], double-outlet right ventricle [DORV], atrial septal defect (ASD), ventricular septal defect [VSD], and transposition of the great arteries [TGA]). Manual stepwise backward selection was used, excluding those variables with nonsignificant probability values. Minimization of the information criteria (Akaike information criteria and Bayesian information criteria) was used to evaluate model fit after exclusion of each variable; the full-model c statistic also was assessed. Interaction terms were considered, and transformations of scale were used to improve calibration. Institutional Review Board approval was gained under expedited review for this study.
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
Among the 1 874 421 cardiac cases initially identified from the NIS, 41% were missing a surgeon identifier. For congenital diagnosis and GUCH cases, this percentage was 48% and 44%, respectively. Using only those cases with a surgeon identifier meeting our initial criteria, we identified 30 250 operations identified between 1988 and 2003, yielding a national estimate of 152 227±7875 operations. Of these, 111 816±7456 (73%) were pediatric heart operations, and 40 461±1365 (27%) were adult congenital operations. Demographic data regarding patients and hospitals are shown in Table 2. There was a large discrepancy based on age: 68% of pediatric heart operations (n=76 141±6855) were performed by surgeons with specialization in pediatric heart surgery; however, only 4.4% (n=1797±224) of GUCH operations were performed by surgeons with specialization in pediatric heart surgery (P<0.001).
Even within the identified adult congenital cardiac diagnoses, the diagnosis-based case mix for pediatric heart surgeons differed from that of nonpediatric heart surgeons (P<0.001; Table 3). In particular, ToF, dextro-TGA, and DORV made up a much larger proportion of operations done by pediatric heart surgeons, collectively 26.7% versus 2.4%. Conversely, ostium secundum ASD dominated the case mix of nonpediatric heart surgeons: 77.4% compared with only 36.8% for pediatric heart surgeons. All other diagnoses differed by <4 percentage points in their proportion within the case mix, and all made up a greater proportion of the pediatric heart surgeon’s case mix (Figure 1). Note, however, that the designations of TGA, TOF, and DORV represent the patient’s primary congenital diagnosis but that the vast majority of these adult patients underwent postrepair procedures (eg, pulmonary valve replacements, ablation of arrhythmias, VSD closure) rather than primary repair. Specific procedures within each diagnostic group are listed in Table 2.
The differences in practice patterns among pediatric and nonpediatric heart surgeons were then analyzed by patient age. Pediatric patients within the same diagnostic subgroups undergo operation by surgeons whose primary focus is pediatric heart surgery, whereas GUCH patients undergo operation primarily by nonpediatric heart surgeons (Figure 2).
In-Hospital Mortality Rates
National estimated overall in-hospital mortality rates for GUCH patients was 1904±118 (95% CI, 1673 to 2135) (4.71%; 95% CI, 4.19 to 5.23). Estimated in-hospital mortality rates for GUCH patients operated on by pediatric heart surgeons was significantly lower than the estimated mortality rates for GUCH patients operated on by nonpediatric heart surgeons (34±12; 95% CI, 11 to 57) (1.87%; 95% CI, 0.62 to 3.13%) versus 1870±117 (95% CI, 1640 to 2100) (4.84%; 95% CI, 4.30 to 5.38; P<0.0001).
In the first multivariable logistic regression model adjusted for patient and hospital characteristics, GUCH patients operated on by nonpediatric heart surgeons had significantly increased in-hospital death rates (odds ratio, 4.50; 95% CI, 2.12 to 9.53; P<0.0001; Table 4). Interestingly, VSD also was associated with an increased risk of in-hospital death, perhaps because of the increased proportion of VSD repairs performed by nonpediatric heart surgeons. Model discrimination (c statistic) with this set of covariables was 0.790. The subsequent logistic regression models explored the relationship between increasing pediatric heart annual experience and death and increasing annual GUCH plus pediatric heart experience and death. For these models, we used the annual percentage of pediatric heart surgery operations and the annual percentage of GUCH plus pediatric heart surgery operations as continuous variables. In agreement with our previous results, we found that increased pediatric heart annual experience and increased GUCH plus pediatric heart annual experience decreased in-hospital mortality rates for GUCH patients after adjustment for other factors (Table 4). Model discriminations were similar, with c statistics of 0.793 and 0.785, respectively.
Factors Associated With Increased LOS and Increased Total Hospital Charges
Multivariable linear regression models adjusted for patient and hospital characteristics were fit to evaluate the impact of specialist care on 2 metrics of resource use: LOS and total hospital charges. Not surprisingly, there was a strong linear relationship between increasing LOS and increased hospital charges. Such models demonstrated no beneficial effect when operation was performed by pediatric heart surgeons. However, when increasing annual percentage of pediatric heart surgery operations per surgeon was analyzed as a continuous explanatory variable, there was a significant reduction in resource use associated with higher individual surgeon percentage of pediatric heart operations (Table 5). Surgeons who performed a greater percentage of pediatric heart operations had decreased LOS and decreased total hospital charges. Interestingly, teaching hospitals also were associated with increased total hospital charges.
We have characterized the national practice patterns of surgeons performing GUCH surgery in the United States and compared in-hospital mortality rates, LOS, and total hospital charges among surgeons with specialization in pediatric heart surgery and surgeons without such specialization. Our study documented a striking dichotomy whereby patients with GUCH undergo operation by surgeons whose primary focus is adult cardiac surgery. In contradistinction, pediatric patients with congenital heart disease within the same diagnostic subgroups undergo operation by surgeons whose primary focus is congenital heart surgery. Therefore, the dichotomy appears motivated primarily by an arbitrary designation of adulthood rather than a fundamental difference in the underlying disease process.
Importantly, we found that for GUCH patients, in-hospital death rates are significantly lower when operation is performed by specialized pediatric heart surgeons. The survival advantage increases as an individual surgeons’ annual percentage of pediatric heart operations increases. Furthermore, our data showed a reduction in resource use associated with operation by surgeons performing a higher annual percentage of pediatric heart operations. These data suggest that selective referral of GUCH patients to centers with specialized pediatric heart surgeons will reduce in-hospital death rates and improve the value of delivered care.
Many studies have documented improved results for pediatric heart surgery in neonates in high-volume centers compared with low-volume centers.14–17 Literature from other surgical specialties has similarly shown a survival benefit for patients undergoing operation by surgeons with specialization in the associated field.18 Current guidelines4,6–10 recommend that >50% of GUCH patients should be evaluated every 12 to 24 months by a cardiologist with specific expertise at a regional center, yet similar initiatives supporting the provision for specialized care for GUCH patients by surgeons with expertise in congenital heart surgery have not been proposed. Considering the statement by Somerville2 that reoperation and invasive investigations were the most frequent reasons for admission to an established GUCH unit in the United Kingdom, such a disparity among treating physicians seems unreasonable.
Pertinent to our present study is the increasing body of evidence that volume criteria alone are a crude proxy for surgical quality and that such criteria ought to be augmented by incorporation of specific processes of care and appropriate risk-adjusted patient outcome analyses.19–21 Birkmeyer and Dimick20 found that the addition of a process measure (perioperative β-blockade) to the original Leapfrog volume standards would have averted 33% of the 23 790 deaths among patients undergoing high-risk procedures in 2000. Additionally, these authors found that for patients undergoing coronary artery bypass graft operation and percutaneous coronary intervention, referral based on risk-adjusted mortality rates would result in 5 times more lives saved than volume criteria alone.20 We believe, from the data in the present study, that decreased in-hospital death rates are likely the result of refined processes of care (eg, the environment) at institutions possessing specialized congenital heart surgeons in addition to the expertise of the surgeons themselves. That is, patients not only are evaluated and operated on by surgeons with expertise in congenital heart disease but also are evaluated and cared for by a multidisciplinary team that includes cardiologists, electrophysiologists, nuclear medicine physicians, radiologists, pediatricians, and nursing and ancillary staff who function within a highly evolved and cohesive system. Recent literature, mainly from the United Kingdom and Canada, has emphasized the importance of developing dedicated GUCH centers.1–4,11 Appropriate and timely transfer of care from a pediatric center to a GUCH center on reaching adulthood also can be coordinated more successfully once both environments have been established.3,4 Such environments are likely to be particularly important for more complex or rare morphological entities.1–5 Indeed, we have shown that selective referral already exists for patients with TGA and DORV. GUCH patients with these diagnoses are preferentially operated on by specialized pediatric heart surgeons (likely within a congenital cardiac center); therefore, despite increased complexity, we found no increased risk of death in these subgroups.
However, the vast majority (94%) of adult congenital patients treated in the United States are operated on for repair of ASD and VSD, and these repairs are performed by surgeons whose practices are focused on adult cardiac surgery rather than congenital heart surgery. Our data demonstrate, however, that despite the apparent simplicity of a VSD repair, patients with this diagnosis had a significantly higher risk of death after adjustment for other factors. It is probable that this increased risk of death is attributable to inappropriate referral to surgeons without expertise in congenital heart disease or secondary to treatment in a center without evolved processes of care specific to GUCH patients.
Importantly, we demonstrated that specialist care resulted in reduced in-hospital mortality rates without an increase in 2 common metrics of resource use: LOS and total hospital charges. The issue of reduction in the cost of care and hospital LOS has been addressed in other surgical disciplines.22,23 Hollenbeck and colleagues,22 in an elegant study, found that volume-based referral for common low-risk procedures such as prostatectomy provided an important reduction in LOS. The authors make the important point that benefits from volume-based referral can be extended beyond in-hospital death, an idea that may offset the drawbacks of regionalization in terms of travel time and healthcare access. Accordingly, Dudley et al14 found that 76% of patients treated at a low-volume institution in California for a variety of surgical conditions, including pediatric heart surgery, could have reached a high-volume institution without traveling >40 km (25 miles) farther. Birkmeyer and colleagues24 examined directly the impact of regionalization of pancreatic and esophageal resection on patient travel times. Their data showed that travel burden could be mitigated by adopting low- or medium-volume standards as the regionalization standard while still providing a 4% to 5% reduction in risk-adjusted mortality rates.24
What are the implications of the present study? Motivation for a paradigm shift in referral patterns depends largely on the dynamic between patient perspective and policy perspective. At what point should surgery for GUCH patients be regionalized to centers possessing dedicated congenital heart surgeons? Clearly, from a patient’s perspective, reducing the risk of each individual death is paramount; therefore, initiatives ought to be articulated in terms of lives saved. However, from a policy perspective, the issues of regionalization are more complex. Increased survival must be balanced with pragmatic concerns and with what is financially feasible. Reducing resource use may offer a more provocative political target for conditions such as GUCH in which overall death is low and therefore the crude survival benefit of regionalization may be minimal. Further prospective studies should be undertaken to confirm our findings.
Our study has several limitations. Administrative data are limited in their ability to differentiate patients according to the severity of illness.19,25 Confounding variables not accounted for in the NIS may have biased our results. Furthermore, it is not possible to determine whether comorbidities identified in the NIS are admission diagnoses or postadmission (or postprocedure) complications.
Additionally, we acknowledge the widely disparate environments in which dedicated congenital surgeons and adult acquired surgeons work. A hospital-level analysis exploring the contribution of the institution itself to our outcome measures would help clarify whether the congenital surgeon is truly a surrogate for the “systems” that he/she brings with him/her. However, we chose to focus this article on the individual surgeon, recognizing that further studies are recommended. Our cut point of 75% for the designation of a surgeon with specialization in pediatric heart surgery was adjudicated by consensus. Sensitivity analysis suggested a cut point of 93%, a figure that would have provided unreasonable polarization of groups (ie, an all-or-none phenomenon).
Surgeon volume analysis could be confounded by use of a separate surgeon identifier for an individual surgeon operating at 2 different hospitals, especially if the surgeon has polarized practices at the different hospitals (ie, a children’s hospital and a hospital that is not a children’s hospital). However, the result would be to bias the results toward the null because these surgeons would be labeled noncongenital surgeons and therefore a smaller difference would be detected among dedicated congenital and adult-acquired surgeons. Additionally, 2 states may list surgeon practice groups under a single identifier, which may bias the results if the individual surgeon case mix within the practice varies. However, we do not believe that this occurs often enough to be a significant concern.
Missing data also limit our analysis and the inferences that can be drawn from our study. Among the 1 874 421 cardiac cases initially identified from the NIS, 41% were missing a surgeon identifier. For congenital diagnosis and GUCH cases, this percentage was 48% and 44%, respectively. Much of the missing surgeon information is due to the fact that entire states did not report this information. California, Georgia, Hawaii, Illinois, Indiana, Kansas, Massachusetts, Ohio, Utah, and Vermont do not provide surgeon or physician identifiers, and cases from these states make up 70% of all the cases with missing surgeon. We assume that these states are not materially different with regard to relative outcomes of pediatric heart surgeons and nonpediatric heart surgeons; thus, the data can be considered “missing at random,” meaning that the results should not be biased as a result of missing data. Additionally, because our national estimate of 152 227 cases is derived from cases within the NIS containing a surgeon identifier, it is possible that our data importantly underestimate the actual national case load.
Certainly, the use of in-hospital death as our main outcome metric severely restricts the applicability of our findings to many important long-term issues in the field of congenital heart disease.26
Despite these limitations, our study is the first to demonstrate that specialist care for GUCH patients reduced in-hospital mortality rates without a corresponding increase in hospital LOS and total hospital charges. Although specialized tertiary care facilities have developed globally in response to the increasing number of patients living with GUCH, many of them have evolved on the basis of local disease prevalence, estimated need, or simply available resources. There is certainly no consensus because of the paucity of data guiding the optimum framework and prerequisites for an ideal GUCH center. Our study, however, clearly suggests that initiatives to develop dedicated GUCH centers should incorporate surgeons with specialization in pediatric heart surgery to improve outcomes. The additional financial burden of regionalization may be mitigated in part by the decreased resource use inherent in the provision of specialized care.
Reid GJ, Irvine MJ, McCrindle BW, Sananes R, Ritvo PG, Siu SC, Webb GD. Prevalence and correlates of successful transfer from pediatric to adult health care among a cohort of young adults with complex congenital heart defects. Pediatrics. 2004; 113: e197–e205.
Therrien J, Dore A, Gersony W, Iserin L, Lithberson R, Meijboom F, Colman JM, Oechslin E, Taylor D, Perloff J, Somerville J, Webb GD. Canadian Cardiovascular Society Consensus Conference 2001 update: recommendations for the management of adults with congenital heart disease: part I. Can J Cardiol. 2001; 17: 940–959.
Therrien J, Gatzoulis M, Graham T, Bink-Boelkens M, Connelly M, Niwa K, Mulder B, Pyeritz R, Perloff J, Somerville J, Webb GD. Canadian Cardiovascular Society Consensus Conference 2001 update: recommendations for the management of adults with congenital heart disease: part II. Can J Cardiol. 2001; 17: 1029–1050.
Therrien J, Warns C, Daliento L, Hess J, Hoffman A, Marelli A, Thilen U, Presbitero P, Perloff J, Somerville J, Webb GD. Canadian Cardiovascular Society Consensus Conference 2001 update: recommendations for the management of adults with congenital heart disease: part III. Can J Cardiol. 2001; 17: 1135–1158.
Gatzoulis MA, Hechter S, Stu SC, Webb GD. Outpatient clinics for adults with congenital heart disease: increasing workload and evolving patterns in referral. Heart. 1999; 81: 57–61.
Agency for Healthcare Research and Quality. HCUP nationwide inpatient sample (NIS): Healthcare Cost and Utilization Project (HCUP): 1988–2005. Available at: www.hcup-us.ahrq.gov/nisoverview.jsp. Accessed August 25, 2006.
Hannan EL, Racz M, Kavey RE, Quaegebeur JM, Williams R. Pediatric cardiac surgery: the effect of hospital and surgeon volume on in-hospital mortality. Pediatrics. 1998; 101: 963–969.
Welke KF, Diggs BS, Karamlou T, Ungerelider RM. Overall hospital surgical case volumes and mortality rates do not accurately represent quality in pediatric heart surgery. Paper presented at: Southern Thoracic Surgical Association 54th Annual Meeting; Bonita Springs, Fla; November 2007.
Hollenbeck BK, Dunn RL, Miller DC, Daignault S, Taub DA, Wei JT. Volume-based referral for cancer surgery: informing the debate. J Clin Oncol. 2007; 25: 91–96.
Williams WG. Surgical outcomes in congenital heart disease: expectations and realities. Eur J Cardiothorac Surg. 2005; 27: 937–944.
Because of the improved outcomes for congenital heart surgery in the neonatal period, there are an increasing number of grown-up patients living with congenital heart disease (GUCH). Many GUCH patients, however, often require ongoing specialized care and surgical intervention, and despite the notion of adulthood, GUCH patients still offer unique challenges with respect to physiology and anatomic complexity for surgeons unfamiliar with congenital heart disease. Because of these unique challenges that characterize GUCH, there has been a growing focus on developing so-called adult congenital centers dedicated to the care of this growing population. Despite these initiatives, however, surgery for GUCH patients is still frequently performed by surgeons without specialization in pediatric heart surgery. Our study defines the national practice patterns for GUCH care in the United States and demonstrates that in-hospital mortality rates are significantly lower when operation is performed by specialized pediatric heart surgeons. The survival advantage increases as an individual surgeon’s annual percentage of pediatric heart operations increases. Furthermore, our study shows a reduction in resource use associated with operation by surgeons performing a higher annual percentage of pediatric heart operations. These data suggest that selective referral of GUCH patients to centers with specialized pediatric heart surgeons will reduce in-hospital death rates and improve the value of delivered care.