Comparing Living Donor and Deceased Donor Liver Transplantation: A Matched National Analysis From 2007 to 2012

Transcription

1 LIVER TRANSPLANTATION 20: , 2014 ORIGINAL ARTICLE Comparing Living Donor and Deceased Donor Liver Transplantation: A Matched National Analysis From 2007 to 2012 Richard S. Hoehn, 1 Gregory C. Wilson, 1 Koffi Wima, 1 Samuel F. Hohmann, 2 Emily F. Midura, 1 E. Steve Woodle, 1 Daniel E. Abbott, 1 Ashish Singhal, 1 and Shimul A. Shah 1 1 Cincinnati Research in Outcomes and Safety in Surgery, Department of Surgery, University of Cincinnati School of Medicine, Cincinnati, OH; and 2 University Health System Consortium and Department Health Systems Management, Rush University, Chicago, IL A complete evaluation of living donor liver transplantation (LDLT) in the United States has been difficult because of the persistent low volume and the lack of adequate comparisons with deceased donor liver transplantation (DDLT). Recent reports have suggested outcomes equivalent to those for DDLT, but these studies did not adjust for differences in recipient selection. From a linkage between the University HealthSystem Consortium and Scientific Registry of Transplant Recipients databases, we identified 14,282 patients at 62 centers who underwent DDLT from 2007 to 2012 and 715 patients at 35 centers who underwent LDLT during the same period. Then, we performed 1:1 propensity score matching for 708 LDLT recipients based on age, Model for End-Stage Liver Disease (MELD) score, and pretransplant patient status. The median follow-up was 2 years. Compared with DDLTrecipients, LDLTrecipients were more likely to be white (84.5% versus 72.2%) and female (41.1% versus 31.7%), to have lower MELD scores (15 versus 19), and to be classified preoperatively as independent (65.3% versus 46.7%) and not hospitalized (91.3% versus 78.4%). The posttransplant length of stay (LOS), in-hospital mortality, costs, and survival were similar between the groups, but LDLT recipients were more likely to be readmitted within 30 days (44.9% versus 37.1%, P ). After matching, the difference in 30-day readmission rates persisted (45.1% versus 33.8%, P ), but there were no differences in the LOS, costs, patient survival, or graft survival. This national report shows that LDLT is associated with higher readmission rates in comparison with DDLT, but the results are comparable for other key patient metrics. Liver Transpl 20: , VC 2014 AASLD. Received June 3, 2014; accepted July 4, See Editorial on Page 1290 Over the past 10 years, the average annual number of liver transplants performed nationally was just more than 6000, but there are currently more than 15,000 Abbreviations: BMI, body mass index; DDLT, deceased donor liver transplantation; HCC, hepatocellular carcinoma; HV, high-volume tertile; ICU, intensive care unit; IQR, interquartile range; LDLT, living donor liver transplantation; LOS, length of stay; LV, low-volume tertile; MELD, Model for End-Stage Liver Disease; MV, medium-volume tertile; NASH, nonalcoholic steatohepatitis; OLT, orthotopic liver transplantation; Q1, quartile; Q2, quartile 2; Q3, quartile 3; Q4, quartile 4; Q5, quartile 5; SES, socioeconomic status; SRTR, Scientific Registry of Transplant Recipients; TIPS, transjugular intrahepatic portosystemic shunt; UHC, University HealthSystem Consortium. The institutional review board of the University of Cincinnati approved this study, and the Scientific Registry of Transplant Recipients project officer of the Health Resources and Services Administration and the Technical Advisory Committee of the Scientific Registry of Transplant Recipients approved the linkage of the 2 data sets. The data reported here have been supplied by the Minneapolis Medical Research Foundation as the contractor for the Scientific Registry of Transplant Recipients. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy of or interpretation by the Scientific Registry of Transplant Recipients or the US Government. Potential conflict of interest: Nothing to report. This study was supported by the University of Cincinnati Department of Surgery. Address reprint requests to Shimul A. Shah, M.D., M.H.C.M., Cincinnati Research in Outcomes and Safety in Surgery (CROSS), Department of Surgery, University of Cincinnati School of Medicine, 231 Albert Sabin Way, ML 0558, MSB 2006C, Cincinnati, OH Telephone: ; FAX: ; shimul.shah@uc.edu DOI /lt View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI /lt. Published on behalf of the American Association for the Study of Liver Diseases VC 2014 American Association for the Study of Liver Diseases.

2 1348 HOEHN ET AL. LIVER TRANSPLANTATION, November 2014 people on the waiting list. 1 Living donor liver transplantation (LDLT), first performed in the United States in 1989, has been used to augment the donor pool and decrease the wait-list mortality of patients in need of liver transplantation. 2 However, many questions have been raised about the safety and efficacy of LDLT. The number of LDLT procedures performed annually in the United States increased rapidly at first but has steadily declined since During this time, there has been increased scrutiny and research regarding the outcomes related to LDLT not only as they pertain to donors but also in comparison with deceased donor liver transplantation (DDLT) in the Western world. The Adult-to-Adult Living Donor Liver Transplantation Cohort studied LDLT from 1998 to 2008 and found that although LDLT recipients had more complications and postoperative hospitalizations, these results improved with increasing center experience with LDLT. 3-6 Other studies have shown decreased graft survival with LDLT versus DDLT. 7,8 A recent review found that LDLT is safe and in select cases offers advantages over DDLT. 9 Much of the literature that has been published about LDLT is from the earlier national experience with LDLT. Our aim was to evaluate the recent LDLT experience on a national level and examine a large number of patients and a representative variety of center experiences. We hypothesized that a patient selection bias may confound reports on outcomes after LDLT, and we attempted to compare medically matched transplant recipients and to evaluate more fairly the safety and cost-effectiveness of LDLT versus DDLT. PATIENTS AND METHODS Study Population A retrospective cohort study was performed for liver transplants between January 1, 2007, and December 31, These years were chosen to represent a modern and updated experience and were available for the linkage described later. Data for this study were drawn from 2 different sources. First, clinical data were obtained from the Scientific Registry of Transplant Recipients (SRTR) Standard Analysis File, a national transplant database that provides donor and recipient data collected by the Organ Procurement and Transplantation Network. 1 These data were then linked to recipient clinical and hospital encounter data obtained from the University HealthSystem Consortium (UHC) Clinical Database/ Resource Manager. UHC is an alliance of 118 academic medical centers and 298 of their affiliated hospitals and represents approximately 95% of the nation s major not-for-profit academic medical centers. Theirs is an administrative database containing patient demographic, financial, International Classification of Diseases (Ninth Revision), and procedure data provided by the member medical centers. Hospital charges are reported for each patient encounter and converted to cost estimates with institutionspecific Medicare cost-to-charge ratios and federally reported area wage indices to normalize regional variations in labor costs. 10 The direct costs are reported as medians and reflect index transplant admissions only. A linkage of patients in these 2 data sets was performed as previously described 11 with recipient age, date of procedure, sex, and transplant center. From January 2007 to December 2012, 34,611 liver transplants from 135 centers were identified from the SRTR database. Over the same time period, 21,868 liver transplants from 67 centers were identified from the UHC database. Recipient age less than 18 years (n ) and repeat orthotopic liver transplantation (OLT) within the same hospitalization (n 5 396) were excluded from this data set. From these parameters we identified 14,282 DDLT procedures and 715 LDLT procedures performed at 62 and 35 centers, respectively. These represented 43% of the 33,470 DDLT procedures and 63% of the 1141 LDLT procedures performed nationally during the same period. Donor and recipient demographic characteristics of this population were similar to the national transplant cohort. The data were verified to be similar to the SRTR complete data set with regard to donor characteristics as well as recipient cause of liver disease, severity of disease, and survival data. By linking these 2 data sets, we were able to assess transplant-specific outcomes, such as patient and graft survival, as well as hospital-level outcomes, including 30-day readmission, discharge disposition, length of stay (LOS), and cost. The following donor characteristics were collected: age (years), sex, race (white, black, Hispanic, or other), body mass index (BMI), cold ischemia time (hours), and warm ischemia time (minutes). Next, the following recipient characteristics were collected: age (years), sex, race (white, black, Hispanic, or other), BMI, socioeconomic status (SES), insurance type (private, government, or other), cause of liver disease, medical history, functional status (independent, dependent, severely ill), admission status (elective or emergent), physical capacity, severity of illness, pretransplant location [intensive care unit (ICU), ward, not hospitalized], and Model for End-Stage Liver Disease (MELD) score. MELD scores were calculated for each recipient based on the UNOS modification to the formula described as medical MELD scores. 12 For the statistical analysis, patients were stratified according to MELD groups similar to those previously used. 13 SES A summary measure of SES from U.S. Census American Community Survey 5-year estimates from 2011 for each U.S. zip code was constructed. The individual variables chosen were based on previous methods and included 3 measures of wealth and income (log of median household income; log of median value of housing unit; and percentage of households with interest, dividend, or rental income), 2 measures of

3 LIVER TRANSPLANTATION, Vol. 20, No. 11, 2014 HOEHN ET AL TABLE 1. Comparison of the Liver Allograft Donors for the Linked Cohort of 14,997 Patients Who Underwent DDLT and LDLT From 2007 to 2012 Living Donors Deceased Donors Donor Characteristics (n or Median) % or IQR (n or Median) % or IQR P Value Overall number of patients % 14, % Sex <0.001 Male % % Female % % Age of donor (years) <0.001 < % % % % % % % % % % Race <0.001 White % % Black % % Hispanic % % Other % % BMI (kg/m 2 ) Cold ischemia time (hours) <0.001 Warm ischemia time (minutes) <0.001 Patients with missing data in a field were excluded from the percentage calculations and statistical analysis for that field. education (proportion of adult residents completing high school and proportion of adult residents completing college), and 1 measure of employment (percentage of residents with management, business, science, and arts occupations). 14,15 To construct the summary measure of SES, z scores for each recipient were determined by subtracting the overall mean (across all recipients) and dividing by the standard deviation. This was completed for each of the 6 variables. 14,15 The summation of the 6 z-score variables then yielded the summary SES score. All recipients were sorted based on their SES and were stratified into quintiles for means of analysis. Matched Cohort Next, we created a 1:1 matched cohort of similar LDLT and DDLT patients. With propensity scores, we were able to match 708 LDLT recipients to 708 DDLT recipients based on age, MELD score, and pretransplantation patient status in an effort to compare outcomes between 2 groups of patients undergoing different procedures with similar preoperative illness. 16 Propensity scores were created via logistic regression. The predicted probabilities from this model served as propensity scores, which were then used in an SAS macro to form matched pairs between LDLT patients and DDLT patients. The balance in the baseline characteristics between the 2 groups was assessed by testing for within-pair differences in baseline covariates. McNemar s test was employed to assess within-pair differences in the matched samples for binary outcomes, whereas the Wilcoxon signedrank test was used for continuous outcomes. Primary outcome variables were the same as those listed above. Statistical Analysis We performed univariate analysis on both the overall cohort and the matched cohort. An a level of 0.05 was used for all significance tests. The Kaplan-Meier method was used to estimate the effect of LDLT and DDLT on allograft and patient survival. The data were analyzed with the statistical packages SAS 9.3 and JMP Pro 11 (SAS Institute, Cary, NC). RESULTS Donor characteristics are outlined in Table 1. LDLT donors were equally male and female, whereas DDLT donors were more often male (59.6% versus 40.4%). LDLT donors were younger, with 82.9% under the age of 50 years versus only 64.1% of DDLT donors. LDLT donors were also more likely to be white and had a lower BMI. Table 2 shows recipient characteristics from the overall cohort. Compared with DDLT, LDLT recipients had a higher percentage of patients who were female, white, and under the age of 50 years. They also had a lower BMI, were more likely to be in the 2 highest SES quintiles (55.8% versus 39.2%), and were more likely to have private insurance. With regard to their health and pretransplant status, LDLT recipients were less likely to be severely ill, severely limited, in the ICU, on hemodialysis, or on a

4 1350 HOEHN ET AL. LIVER TRANSPLANTATION, November 2014 TABLE 2. Comparison of the 14,997 Patients in the Linked Database Who Underwent DDLT and LDLT From 2007 to 2012 Living Donors Deceased Donors Recipient Characteristics (n or Median) % or IQR (n or Median) % or IQR P Value Overall number of patients % 14, % Sex <0.001 Male % % Female % % Age (years) < % % % % % % % % % % % % Average <0.001 Race <0.001 White % 10, % Black % % Hispanic % % Other % % SES <0.001 Q1 (lowest) % % Q % % Q % % Q % % Q5 (highest) % % Insurance <0.001 Private % % Government % % Other 4 0.6% % BMI (kg/m 2 ) <0.001 Cause of liver disease <0.001 Alcohol % % HCC % % NASH % % Viral hepatitis % % Other % % Recipient medical history Diabetes % % 0.05 Angina % % 0.02 Hemodialysis 3 0.4% % <0.001 Bacterial peritonitis % % 0.57 Portal vein thrombosis % % 0.43 TIPS % % 0.53 Approved for MELD exception % % <0.001 Functional status <0.001 Independent % % Dependent % % Severely ill % % Unknown % % Admission status <0.001 Elective % % Emergent % 11, % Other 3 0.4% % Physical capacity <0.001 Hospitalized/severely limited % % Limited % % No limitations % % MELD < % % % % % % % %

5 LIVER TRANSPLANTATION, Vol. 20, No. 11, 2014 HOEHN ET AL Living Donors TABLE 2. Continued Deceased Donors Recipient Characteristics (n or Median) % or IQR (n or Median) % or IQR P Value % % Median MELD <0.001 Severity of illness <0.001 Minor % % Moderate % % Major % % Extreme % % Before OLT <0.001 ICU % % Hospital ward % % Not hospitalized % 11, % Recipient on life support 6 0.8% % <0.001 Recipient on ventilator 6 0.8% % <0.001 Discharge home % 11, % <0.001 Readmission (30 days) % % <0.001 LOS (OLT to discharge) ICU LOS <0.001 Mortality % % 0.63 Direct cost (OLT to discharge) $108,675 $40,730 $103,092 $63, Patients with missing data in a field were excluded from the percentage calculations and statistical analysis for that field. ventilator. They also had a significantly lower average MELD score. LDLT recipients had a shorter postoperative ICU LOS and were more likely to be discharged home. However, LDLT recipients were also more likely to be readmitted within 30 days. LDLT and DDLT recipient graft survivals were similar. With a median follow-up of 2 years, DDLT had lower patient survival according to the Kaplan-Meier analysis (P ; Fig. 1). Table 3 describes our matched cohort of LDLT and DDLT recipients. Patients were matched from propensity scores based on age, MELD score, and pretransplant location, none of which was statistically different between the 2 groups after matching. Between the groups, there were no differences in ICU LOS, discharge home, cost of hospitalization, or mortality. However, the LDLT cohort had younger donors, and the recipients were of higher SES, had longer LOS, and had significantly higher 30-day readmissions. Kaplan-Meier analysis of patient and graft survival showed no differences between the groups (Fig. 2A,B). Next, we analyzed center LDLT volume and its effect on outcomes. Figure 3 shows the distribution of total LDLT volume across the studied time period ( ). The total volume per center ranged from 1 to 99, and the annual volume ranged from 1 to 24. The highest total volume quartile (34-99 LDLT procedures) had the lowest 30-day readmission rate, in-hospital mortality, total LOS, and ICU LOS but also had the lowest rate of discharge to home (Table 4). Centers in the lower 2 volume quartiles (1-3 LDLT procedures and 4-9 LDLT procedures) had significantly higher mortality rates during the study period in comparison Figure 1. Kaplan-Meier analysis of patient survival after DDLT and LDLT in the overall cohort. DDLT recipients had worse overall patient survival (P ). with centers that performed more than 10 LDLT procedures during that time. DISCUSSION This study provides an update of previous reports on the short-term outcomes following LDLT in a national sample. Compared with DDLT recipients, LDLT recipients more often were younger and had a lower BMI, a

6 1352 HOEHN ET AL. LIVER TRANSPLANTATION, November 2014 TABLE 3. Comparison of the 1:1 Propensity Score Matched Cohort of 1416 Liver Transplant Patients Who Underwent LDLT and DDLT From 2007 to 2012 Living Donors Deceased Donors Variable (n or Median) % or IQR (n or Median) % or IQR P Value Number of patients % % Recipient age (years) % % % % % % % % % % % % MELD % % % % % % % % % 4 0.6% Outpatient status 0.05 ICU 8 1.1% % Hospital ward % % Not hospitalized % % Cause of liver disease <0.001 Viral % % HCC % % Alcohol % % NASH % % Other % % Center volume 0.08 LV % % MV % % HV % % SES <0.001 Q1 (lowest) % % Q % % Q % % Q % % Q5 (highest) % % Donor age > 60 years 6 0.9% % <0.001 Readmission (30 day) % % <0.001 Routine discharge home % % 0.11 Mortality % % 0.21 LOS (OLT to discharge) ICU LOS Direct cost (OLT to discharge) $108,325 $40,573 $103,513 $63, Patients with missing data in a field were excluded from the percentage calculations and statistical analysis for that field. lower MELD score, a higher SES, and private insurance. Before transplantation, LDLT recipients were less often on a ventilator, on hemodialysis, in the ICU, or severely ill. Postoperatively, LDLT recipients had a shorter ICU LOS and better overall survival on Kaplan-Meier analysis but were more likely to be readmitted within 30 days. Even after propensity score matching, LDLT recipients had higher 30-day readmission rates but similar patient and graft survival rates. Much of the LDLT literature to date has been from single-institution experiences or from small groups of specialized LDLT centers. These reports have suggested that LDLT recipients are healthier than the DDLT population. 4,8 Many report higher complications and readmissions 4,5,17 with decreased graft survival after LDLT. 7,8 This study examines outcomes of approximately two-thirds of the LDLT procedures performed in the United States between 2007 and 2012 from a diverse group of 35 LDLT centers. This is the first analysis to compare outcomes of LDLT with a matched DDLT cohort based on pretransplant medical conditions. Because of the high degree of patient selection in LDLT, a case-controlled analysis is critical

7 LIVER TRANSPLANTATION, Vol. 20, No. 11, 2014 HOEHN ET AL Figure 3. Total number of LDLT procedures performed per center during the study period ( ). Each bar represents 1 center. Our analysis included 35 centers performing between 1 and 99 LDLT procedures during the study period. Figure 2. (A) Kaplan-Meier analysis of patient survival after DDLT and LDLT in the propensity score matched cohort. Patient survival was not statistically different between the 2 groups (P ). (B) Kaplan-Meier analysis of graft survival after DDLT and LDLT in the propensity score matched cohort. Graft survival was not statistically different between the 2 groups (P ). for evaluating LDLT and comparing it with DDLT. We found no differences in patient or graft survival in a large population of medically similar patients. Several authors have discussed the learning curve associated with LDLT. 2,3,5,17 These reports were mostly from specialized centers and showed survival benefits once a center had performed 20 LDLT cases. Our analysis included centers that performed between 1 and 99 procedures from 2007 to 2012 (Fig. 3). This represents a broad range of expertise and thus a more nationally representative sample of LDLT than previously reported. We were surprised to find many institutions performing 3 or fewer LDLT procedures in a given year. We believe that this gives the current analysis a representative snapshot of the experience and volume performed nationally in comparison with other studies that have focused on high-volume institutions. Our analysis confirms the learning curve associated with LDLT and suggests a substantial decrease in patient mortality with increasing center LDLT volume. This linkage of the UHC and SRTR databases allowed us to examine a large population of patients and to focus on not only transplant-specific outcomes but also perioperative hospital-based metrics. With this information, we were able to link medically equivalent LDLT and DDLT patients and examine differences in their perioperative resource utilization. We found no differences in LOS or total cost of the transplantation admission. However, we did find a higher 30-day readmission rate among LDLT recipients, which is consistent with existing literature. 4 Our study showed a 30-day readmission rate of 45% for LDLT and 32.5% for DDLT. Although many studies have looked at specific complications after LDLT, there has been no large study to identify a 30- day readmission rate. With a growing focus on readmission as a quality metric 18 it is important to have an understanding of current readmission rates for specific procedures, and we found LDLT readmission rates to be significantly higher than those for DDLT, despite equal patient and graft survival. LDLT is a technically difficult procedure with higher complication rates compared with DDLT. 5,17 Furthermore, some have suggested that readmission may actually improve patient survival. 19 Perhaps the increased readmission rate with LDLT is necessary and beneficial to the recipients. There are several limitations to this study with a large, retrospective, administrative database. Although our linked database represents 43% of DDLT procedures and 63% of LDLT procedures performed during the study period, it is possible that it is

8 1354 HOEHN ET AL. LIVER TRANSPLANTATION, November 2014 TABLE 4. LDLT Center Volume Analysis of Hospital Level Factors P Value %or IQR Quartile 4 (n or Median) %or IQR Quartile 3 (n or Median) %or IQR Quartile 2 (n or Median) %or IQR Quartile 1 Variable (n or Median) Readmission % % % % 0.01 Discharge home % % % % 0.78 Mortality % % 5 4.5% % <0.001 LOS (OLT to discharge) ICU LOS <0.001 Direct cost (OLT $101,410 $96,127 $128,770 $49,894 $95,391 $66,105 $107,515 $34, to discharge) NOTE: Q1, 1 to 3 LDLT procedures during the study period ( ); Q2, 4 to 9 LDLT procedures; Q3, 11 to 28 LDLT procedures; and Q4, 34 to 99 LDLT procedures. weighted toward areas of high SES in urban zip codes. This would allow SES Q5 patients to choose centers with high volume and quality regardless of location and would not account for many of the differences in access that SES Q1 candidates might face if in a rural area with only 1 transplant center in the state. Also, data from UHC may not represent true national trends because the participating centers likely represent a coalition skewed toward large, academic centers. We were unable accurately to follow complications or determine cause of readmission to identify why the differences exist between LDLT and DDLT. The use of residence zip codes as the surrogate marker for SES compared with individual SES creates some generalized assumptions regarding the individual patient. This method has been validated and has a high degree of correlation with actual SES components. 14,15 In conclusion, LDLT is a viable option for expanding the liver donor pool with outcomes comparable to those with DDLT. Not surprisingly, LDLT is performed on younger, healthier patients with lower MELD scores, higher SES, and private insurance in comparison with DDLT. Patient selection will likely continue to be important in achieving optimal outcomes. Additionally, there appears to be a large volume-outcome relationship with LDLT, and centers performing these complex procedures should closely monitor their volumes and outcomes to ensure acceptable results. These data support the continued use of LDLT for select patients in need of liver transplantation and provide insight into the expected readmission rate associated with this procedure. REFERENCES 1. Organ Procurement and Transplantation Network. Transplants by Donor Type. Richmond, VA: United Network for Organ Sharing. Accessed Feb. 12, Berg CL, Gillespie BW, Merion RM, Brown RS, Abecassis MM, Trotter JF, et al; for A2ALL Study Group. Improvement in survival associated with adult-to-adult living donor liver transplantation. Gastroenterology 2007;133: Olthoff KM, Merion RM, Ghobrial RM, Abecassis MM, Fair JH, Fisher RA, et al; for A2ALL Study Group. Outcomes of 385 adult-to-adult living donor liver transplant recipients: a report from the A2ALL consortium. Ann Surg 2005;242: Merion RM, Shearon TH, Berg CL, Everhart JE, Abecassis MM, Shaked A, et al; for A2ALL Study Group. Hospitalization rates before and after adult-to-adult living donor or deceased donor liver transplantation. Ann Surg 2010;251: Freise CE, Gillespie BW, Koffron AJ, Lok AS, Pruett TL, Emond JC, et al; for A2ALL Study Group. Recipient morbidity after living and deceased donor liver transplantation: findings from the A2ALL retrospective cohort study. Am J Transplant 2008;8: Salvalaggio PR, Modanlou KA, Edwards EB, Harper AM, Abecassis MM. Hepatic artery thrombosis after adult living donor liver transplantation: the effect of center volume. Transplantation 2007;84:

9 LIVER TRANSPLANTATION, Vol. 20, No. 11, 2014 HOEHN ET AL Thuluvath PJ, Yoo HY. Graft and patient survival after adult living donor liver transplantation compared to a matched cohort who received a deceased donor transplantation. Liver Transpl 2004;10: Abt PL, Mange KC, Olthoff KM, Markmann JF, Reddy KR, Shaked A. Allograft survival following adult-to-adult living donor liver transplantation. Am J Transplant 2004;4: Quintini C, Hashimoto K, Uso TD, Miller C. Is there an advantage of living over deceased donation in liver transplantation? Transpl Int 2013;26: Macomber CW, Shaw JJ, Santry H, Saidi RF, Jabbour N, Tseng JF, et al. Centre volume and resource consumption in liver transplantation. HPB (Oxford) 2012;14: Salvalaggio PR, Dzebisashvili N, MacLeod KE, Lentine KL, Gheorghian A, Schnitzler MA, et al. The interaction among donor characteristics, severity of liver disease, and the cost of liver transplantation. Liver Transpl 2011; 17: Kamath PS, Wiesner RH, Malinchoc M, Kremers W, Therneau TM, Kosberg CL, et al. A model to predict survival in patients with end-stage liver disease. Hepatology 2001;33: Ozhathil DK, Li YF, Smith JK, Tseng JF, Saidi RF, Bozorgzadeh A, Shah SA. Impact of center volume on outcomes of increased-risk liver transplants. Liver Transpl 2011;17: Diez Roux AV, Merkin SS, Arnett D, Chambless L, Massing M, Nieto FJ, et al. Neighborhood of residence and incidence of coronary heart disease. N Engl J Med 2001;345: Birkmeyer NJ, Gu N, Baser O, Morris AM, Birkmeyer JD. Socioeconomic status and surgical mortality in the elderly. Med Care 2008;46: Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997;127: Wan P, Yu X, Xia Q. Operative outcomes of adult living donor versus deceased donor liver transplantation: a systematic review and meta-analysis. Liver Transpl 2014; 20: Axon RN, Williams MV. Hospital readmission as an accountability measure. JAMA 2011;305: Gorodeski EZ, Starling RC, Blackstone EH. Are all readmissions bad readmissions? N Engl J Med 2010;363: