BACKGROUND: Low case volume has been associated with poorer surgical outcomes in a multitude of surgical procedures. We studied the association among low case volume, outcomes, and the likelihood of pediatric liver transplantation.
METHODS: We studied a cohort of 6628 candidates listed in the Organ Procurement and Transplantation Network for primary pediatric liver transplantation between 2002 and 2012; 4532 of the candidates went on to transplantation. Candidates were divided into groups according to the average volume of yearly transplants performed in the listing center over 10 years: >15, 10 to 15, 5 to 9, and <5. We used univariate and multivariate Cox regression analyses with bootstrapping on transplant recipient data and identified independent recipient and donor risk factors for wait-list and posttransplant mortality.
RESULTS: 38.5% of the candidates were listed in low-volume centers, those in which <5 transplants were performed annually. These candidates had severely reduced likelihood of transplantation with only 41% receiving a transplant. For the remaining candidates, listed at higher volume centers, the transplant rate was 85% (P < .001). Being listed at a low-volume center was a significant risk factor in multivariate Cox regression analysis for both wait-list mortality (hazard ratio, 3.27; confidence interval, 2.53–4.23) and posttransplant mortality (hazard ratio, 2.21; confidence interval, 1.43–3.40).
CONCLUSIONS: 38.5% of pediatric transplant candidates are listed in low-volume transplant centers and have lower likelihood of transplantation and poorer outcomes. If further studies substantiated these findings, we would advocate consolidating pediatric liver transplantation in higher volume centers.
- CI —
- confidence interval
- HR —
- hazard ratio
- MELD —
- Model for End-Stage Liver Disease
- PELD —
- Pediatric End-Stage Liver Disease
- UNOS —
- United Network for Organ Sharing
What’s Known on This Subject:
Low case volume has traditionally been associated with poor outcomes in complex surgical procedures, including pediatric liver transplantation.
What This Study Adds:
This retrospective analysis supports the association between low case volume and poorer outcomes in pediatric liver transplantation, and, in addition, shows that candidates listed in low-volume centers have severely limited access to transplantation.
Traditionally, research has shown an association between low case volume and poor outcomes in a multitude of surgical procedures,1–5 such as hysterectomy,6 pancreatectomy,5,7 esophagectomy,5 pelvic exenteration,5 and coronary artery bypass surgery.2,8,9 Studies have also shown this relationship in several pediatric procedures of varying complexity, such as cardiac surgery,10,11 pyloromyotomy,12 appendectomy,13,14 and inguinal hernia repairs.13,14
In the field of transplantation, a landmark manuscript published in 1999 suggested that programs performing <20 transplants per year had inferior outcomes.15 Although several studies have refuted these conclusions,16,17 other investigations into all types of solid organ transplants have upheld the relationship between low volume and poor outcomes.18–22 Specifically in the field of pediatric liver transplantation, researchers working with the Scientific Registry of Transplant Recipients database have shown inferior outcomes in transplantation associated with low-volume centers.14 In this study, we used multivariate analysis to investigate the relationship between wait-list and transplant outcomes and pediatric transplant liver center volume. We sought to look beyond these outcomes by analyzing the effect of center volume on transplant access.
Allocation of pediatric livers for transplantation is designed to serve the sickest patients first. These patients with acute liver failure are designated Status 1A, whereas Status 1B is reserved for very sick chronically ill pediatric patients. Aside from these special status categories, the priority for transplantation is based on the Pediatric End-Stage Liver Disease (PELD) score for children aged ≤11 years and on the Model for End-Stage Liver Disease (MELD) score for children aged from 12 to 18 years. The PELD score is a model for wait-list mortality that includes bilirubin, international normalized ratio, albumin, growth failure, and age <1 year. The MELD score includes international normalized ratio, bilirubin, and creatinine. When a pediatric liver donor becomes available, it is first offered to Status 1A children in the United Network for Organ Sharing (UNOS) region and then nationally. Subsequently, it is offered to Status 1 adults, to Status 1B children, to children 0 to 11 years in order of decreasing PELD scores in the UNOS region, and to children 12 to 18 also in decreasing order of MELD scores. Transplant center characteristics do not affect allograft allocation; only candidate characteristics are considered.
We performed a retrospective analysis of the UNOS deidentified patient-level data of all candidates listed for liver transplant between March 1, 2002, and December 31, 2012. We analyzed the liver registry data collected from all transplant recipients younger than 18 years by the Organ Procurement and Transplantation Network. Donor and recipient characteristics were reported at the time of transplant. Follow-up information was collected at 6 months and then yearly after transplantation. Patients undergoing combined or multivisceral transplants (n = 767) and candidates placed on the wait-list for combined or multivisceral transplants (n = 662) were excluded. Retransplantation candidates were also excluded (n = 1064). From the date of listing, 6628 patients were followed, and 4532 candidates received a transplant during the study period. All patients were followed to either death (n = 1171) or the date of last known follow-up (n = 5457).
Data were analyzed by using a standard statistical software package, Stata 9 (Stata Corp, College Station, TX). Continuous variables were reported as mean ± SD and compared using the Student t test and the Mann-Whitney U test. Contingency table analysis was used to compare categorical variables. A P value <.05 was considered significant, and all reported P values were 2-sided. A hazard ratio (HR) >1 indicates a greater risk of mortality.
Probability of Transplant Analysis
The primary outcome measure was transplantation. All listed candidates were included in the analysis (n = 6628). Patients who died were censored. Time to transplantation was assessed as time from date of listing to date of transplantation.
Wait-List Survival Analysis
The primary outcome measure was death on the wait list. All listed candidates were included in the analysis (n = 6628). Candidates were not removed from the analysis if they were taken off the wait list. Death was established by the UNOS death date and social security death master files. Time to death was assessed as the time from date of listing to date of death while on the wait list. Candidates who underwent transplantation were censored. Early listed patients with improved liver function make up a majority of long-term wait-list survivors.
Posttransplant Survival Analysis
The primary outcome measure was death after transplantation. Only recipients who underwent transplantation were included in this analysis (n = 4532). Time to death was assessed as the time from date of transplantation to date of death.
The primary outcome measure was death, regardless whether it was posttransplant or wait-list death. Candidates were followed from time of listing (n = 6628) to time of death or last known follow-up, regardless of transplantation. UNOS death date and social security master death files were used to establish death. Time to death was assessed as the time from date of listing to date of death.
Kaplan-Meier analysis with log-rank test and Cox regression were used for time-to-event analysis. Survival, on the wait list or posttransplant, was the dependent variable, and the risk factors were the independent variables in the regression analysis. Risk factors that were significant in univariate analysis (P < .05) were included in the multivariate analysis. Multivariate Cox regression was performed combining 100 bootstraps. Patients lost to follow-up or alive on December 31, 2012, were censored at the date of last known follow-up.
Pediatric liver-transplant volume for each center was the average number of cases performed from 2002 to 2012. Centers were categorized as low volume when their records showed <5 cases performed per year. The slope of the curve is steepest between 0 and 4 cases per year (Fig 1). The other groups were categorized as multiples of 5. Experience of center was defined as the number of consecutive years that pediatric liver transplants were performed since 1987. To account for the impact of individual centers, we repeated all analyses with conditional Cox regression stratifying for individual transplant centers.
The recipient and donor risk factors considered in this analysis are listed in Table 1. Creatinine clearance was calculated with the updated Schwartz bedside formula: estimated glomerular filtration rate = 0.41 × height (cm)/Scr (mg/dL). Height and weight deficits were based on Centers for Disease Control and Prevention growth charts. Standard deviations were calculated according to published Z scores. Status 1 included Status1A and Status1B. Life support is a UNOS designation for ventilator support, artificial liver, or a write in entry for another mechanism.
Multiple imputation with predicted mean matching was performed for the following incomplete predictors in the Organ Procurement and Transplantation Network database: serum sodium (25.3%), cold ischemia time (9.0% missing), serum creatinine (6.9%), ascites (5.8%), encephalopathy (5.7%), recipient weight (4.1% missing), diagnosis (0.4%), PELD score (0.3%), recipient height (0.1% missing), albumin (0.1%), and Status 1 (0.1%). We found no significant difference in the missing variables between low- and higher volume centers. We also found no significant differences in our survival models if serum sodium was removed.
Candidates’ Removal From the Wait List for Conditioned Improved
These candidates were removed from the wait list for improvement in their medical condition. They did not receive a transplant at a later date or die within our study period.
To investigate clustering of low-volume centers in particular UNOS regions, we compared the proportions of low-volume listings to overall listings among UNOS geographic regions.
Data Entry Rate
Data entry completion for variables is listed in Table 1. Most variables were well populated. Multiple imputation with predicted mean values was performed for missing variables.
The study population at the time to transplant and wait-list survival analysis consisted of 6628 patients. Wait-list analysis comprised 4554 years-at-risk for liver transplant recipients. Mean follow-up was 0.7 years. The study population for the posttransplant survival analysis had 4532 patients. Posttransplant survival analysis comprised 22 549 years-at-risk for liver transplant recipients. Mean follow-up was 5.0 years. Demographic and clinical characteristics are summarized in Table 2.
Centers with >15 transplants a year had a transplant rate over the study period of 83%; those with 10 to 15 transplants per year, 94%; 5 to 9 transplants per year, 84%; and >5 transplants per year, 41%. Figure 1 shows the dot plot of transplant rates for each center’s transplant volume. Figure 2 shows the Kaplan-Meier curve of the probability of transplantation.
Wait-List Survival Analysis
The recipient and center risk factors listed in Table 1 were considered. Risk factors that were significant in multivariate analysis are presented in Table 3. The most significant risk factors were as follows: listing in a center with transplant volume <5 cases per year (HR 3.3, confidence interval [CI] 2.5–4.2) and dialysis (HR 2.6, CI 1.5–4.3). The Kaplan-Meier curve for wait-list survival is shown in Fig 3.
Posttransplant Survival Analysis
The risk factors listed in Table 1 were considered. Risk factors that were significant in multivariate analysis are presented in Table 4. The most significant risk factors were hepatoblastoma (HR 3.5, CI 2.3–5.3), life support (HR 2.7, CI 1.8–4.0), and transplantation in a center with a transplant volume <5 (HR 2.2, CI 1.4–3.4). The Kaplan-Meier curve for posttransplant survival is shown in Fig 4.
The Kaplan-Meier curve for survival of all listed candidates from time of listing is shown in Fig 5.
Cadaveric Technical Variant Grafts
Centers performing <5 transplants per year used cadaveric technical variant grafts 18% of the time. This percentage was significantly smaller (P < .001) than the percentages of the other volume groups: 5 to 9 transplants per year, 29%; 10 to 15 transplants per year, 29%; and >15 transplants per year, 29%.
The survival of recipients under 6 kg treated in centers performing >15 transplants per year was significantly better than the survival in other centers. Centers with >15 transplants a year had a 1-year survival of 94.1%, 5-year survival of 92.0%, and 10-year survival of 92.0%. Centers performing 10 to 15 transplants per year had a 1-year survival of 84.8%, 5-year survival of 82.5%, and 10-year survival of 82.5%. For those centers performing 5 to 9 transplants per year, 1-year survival was 87.9%, 5-year survival 83.6%, and 10-year survival 82.0%. Centers performing <5 transplants per year had a 1-year survival of 89.6%, 5-year survival of 82.2%, and 10-year survival of 82.2%. The P value was <.001 for each group by log-rank test with reference to >15 transplants per year.
The survival of recipients on life support treated in centers performing >15 transplants per year was significantly better than the survival in other centers. Centers with >15 transplants a year had a 1-year survival of 90.2%, 5-year survival of 84.7%, and 10-year survival of 83.4%. Centers that performed 10 to 15 transplants per year had a 1-year survival of 80.7%, 5-year survival of 79.7%, and 10-year survival of 78.0%. Those that performed 5 to 9 transplants per year reached a 1-year survival of 74.3%, 5-year survival of 71.6%, and 10-year survival of 68.6%. Centers performing <5 transplants per year had a 1-year survival of 75.8%, 5-year survival of 66.6%, and 10-year survival of 63.3%. The P value was <.001 for each group by log-rank test with reference to >15 transplants per year.
Candidate Status at the End of Follow-up
The candidate status at the end of follow-up on December 31, 2012, including alive posttransplant, dead posttransplant, alive on the wait list, and dead on the wait list is reported in Table 2.
Candidates Removal From the Wait List for Conditioned Improved
There is a disproportionate number of candidates removed for condition improved from the low-volume center wait list (Table 2). The 3 most common diagnoses for removed candidates for improved liver function were acute liver failure (25%), idiopathic etiology (24%), and biliary atresia (14%). If these patients are removed from the analysis, the adjusted transplant rate for low-volume centers (<5 transplants per year) is 53.6% compared with 87.4% (>15 transplants per year), 96.0% (10–15 transplants per year), and 90.9% (5–10 transplants per year). The adjusted HR for wait-list mortality on the low-volume wait list is HR 4.76 (CI 3.63–6.24) compared with the unadjusted HR of 3.27 (CI 2.53–4.23).
We report on variations of low-volume listings by UNOS region in Table 5. Low-volume center listings in regions 1, 4, 7, and 11 were overrepresented. Only region 1 is considered disadvantageous in terms of donor allograft supply and demand23 (Table 5).
Impact of Individual Centers
We found no significant differences in the outcomes or regression analyses when we accounted for the impact of individual centers.
Investigators have established that low case volumes adversely affect survival outcomes in a variety of procedures, from those as simple as inguinal hernia repairs to those as complex as pediatric liver transplantation.1–9 Although studies have refuted this assertion,5 the majority seems to accept the relationship. Overall, this understanding has also been reached in the field of transplantation. A number of researchers have shown better outcomes in high-volume centers for renal, cardiac, and liver transplantations,18–22 but others have refuted these assertions.16,17 Parallel studies in pediatric renal, cardiac, and liver transplantation have also demonstrated this low-volume–poor outcome relationship.14,20,24 As a result of the general acceptance of this conclusion, insurance companies routinely require a minimum yearly case volume to designate a program as a center of excellence in adult and pediatric transplantation.25
Our analysis is unique in the body of literature on this topic because it looks beyond survival outcomes into access to transplantation. We do see worse wait-list and posttransplant survival in low-volume pediatric liver transplant centers, but the most profound differences are in the likelihood of transplantation. Low-volume centers (<5 transplants per year), where 39% of the children are listed, have a smaller transplant rate (41%) than the other centers (85%). This staggering difference in the likelihood of transplantation is exacerbated by poorer wait-list and posttransplant survival in low-volume centers. The 5-year wait-list survival in low-volume centers (63%) was also lower than the corresponding survival (92%) in higher volume centers (those with >15 transplants per year). The 5-year posttransplant survival was 83% in low-volume centers and 92% in high-volume centers. Listing and transplantation in low-volume centers were among the strongest risk factors in our multivariate analysis for wait-list and posttransplant mortality (HR 3.3 and 2.2, respectively).
Our analysis does shed light on the reasons behind the poor access to transplantation in low-volume centers. In low-volume centers, patients may be listed for transplantation too early in their disease progression. About 22.9% of candidates listed for transplant in low-volume centers were removed from the wait list due to improvement of the liver function, whereas only 0.6% to 1.8% of the candidates were removed from the wait list in the other centers. These early listing practices may lead to some patients receiving unnecessary liver transplants. The transplant rate for candidates listed with acute liver failure in low-volume centers was 28.6% compared with 71.4% (>15 transplants per year), 82.1% (10–15 transplants per year), and 66.7% (<5 transplants per year). This will clearly affect the number of candidates who die on the list as well as the number of patients who are removed for recovering liver function. There were no significant differences with the percentage of patients transplanted with acute liver failure (Table 2). Early listing is not the sole explanation; we consider that much greater wait-list mortality may also contribute. At the end of follow-up, 21.6% of waitlisted patients in low-volume centers died, compared with 3.2% to 4.6% in other centers. The management of end-stage liver disease in children is challenging and requires extensive medical, critical care, and institutional expertise. Our analysis also suggests that low-volume centers are less likely to take technical variant cadaveric allografts and less likely to take other extended criteria donors. Once listed, candidates from low-volume centers are also less likely to transfer to high-volume centers (only 6.2% of candidates). Other factors beyond medical issues also contributed to the differences, including policies and strategies to maintain active programs and surgeon availability. Regional variations in the distribution of low-volume centers did not explain the differences in the likelihood of transplantation or outcomes.
The significant differences in survival outcomes and in the likelihood of transplantation suggest that pediatric liver transplantation should be consolidated in higher volume centers. If substantiated by other studies, this change in policy would not be trivial because 39% of patients are listed in centers performing <5 transplants per year. We would strongly suggest a policy change to exempt centers serving geographically isolated populations. We did not find significant differences in the likelihood of transplantation or survival outcomes among centers in which ≥5 transplants were performed per year. However, when we considered our riskiest pediatric recipients (<6 kg and on life support), we found the best outcomes in our highest volume centers (>15 transplants per year).
Since the passage of the National Transplantation Act of 1984, data entry has been mandatory for all US transplant centers. Nevertheless, all patient registries often suffer from variability in data entry. The findings from this study were based on large cohorts of patients and are unlikely to be significantly affected by small amounts of missing data. We attempted to account for missing data with multiple imputation analysis. Another significant limitation was that center-specific factors could not be appropriately accounted for.
Of pediatric transplant candidates, 38.5% are listed in low-volume transplant centers and have a reduced likelihood of transplantation and poorer outcomes. If further studies substantiated these findings, we would advocate consolidating pediatric liver transplantation in higher volume centers.
We thank Ana María Rodríguez, PhD, a member of the Baylor College of Medicine Michael E. DeBakey Department of Surgery Research Core, for her editorial assistance during preparation of the manuscript.
- Accepted April 3, 2015.
- Address Correspondence to Abbas Rana, MD, Michael E. DeBakey Department of Surgery, Division of Abdominal Transplantation and Division of Hepatobiliary Surgery, Baylor College of Medicine, One Baylor Plaza, MS:BCM390, Houston, TX 77030; E-mail:
Drs Rana and Goss participated in conceptualization of the study and data analysis, drafted the initial manuscript, and reviewed and revised the final manuscript; Drs Pallister, Halazun, O’Mahony, and Goss and Ms Nalty reviewed and revised the final manuscript; and all authors approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: This study was funded by the Cade R. Alpard Foundation.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Hannan EL,
- Racz M,
- Kavey RE,
- Quaegebeur JM,
- Williams R
- Safford SD,
- Pietrobon R,
- Safford KM,
- Martins H,
- Skinner MA,
- Rice HE
- Sloan FA,
- Shayne MW,
- Doyle MD
- Copyright © 2015 by the American Academy of Pediatrics