- Email: [email protected]
Early Allograft Dysfunction Is Associated With Excess Resource Utilization After Liver Transplantation
Early Allograft Dysfunction Is Associated With Excess Resource Utilization After Liver Transplantation K.P. Croome, R. Hernandez-Alejandro, and N. Chandok ABSTRACT Background. There are limited data on length of stay (LOS) following liver transplantation (LT), yet this is an important health services metric that directly correlates with early post-LT health care costs. The primary objective of this study was to examine the relationship between early allograft dysfunction (EAD) and LOS after LT. The secondary objective was to identify additional recipient, donor, and operative factors associated with LOS. Methods. Adult patients undergoing primary LT over a 32-month period were prospectively examined at a single center. Subjects fulfilling standard criteria for EAD were compared with those not meeting the definition. Variables associated with increased LOS on ordinal logistic regression were identified. Results. Subjects with EAD had longer mean hospital LOS than those without (42.5 ⫾ 38.9 days vs 27.4 ⫾ 31 days; P ⫽ .003). Subjects with EAD also had longer mean intensive care LOS (8.61 ⫾ 10.28 days vs 5.45 ⫾ 11.6 days; P ⫽ .048). Additional factors significantly associated with LOS included Model for End-Stage Liver Disease (MELD) score, recipient location before LT, and postoperative surgical complications. Conclusions. EAD is associated with longer hospitalization after LT. MELD score, preoperative recipient location, and postoperative complications were significantly associated with LOS. From a cost-containment perspective, these findings have implications on resource allocation. iver transplantation (LT) is the only effective treatment for well selected patients with decompensated cirrhosis, acute liver failure, primary metabolic disorders, or early stage hepatocellular carcinoma (HCC), yet it is among the most costly of surgeries.1–3 Given the current era of cost containment in health care and the increased emphasis on accountability of resource consumption within health systems, it is important to identify possible areas of wasteful expenditure where implementation of new policies or strategies would be beneficial. Length of stay (LOS) is considered to be a reliable surrogate for postoperative LT resource utilization, and it is certainly among the most important parameters to study when estimating inpatient health care expenditure.4,5 There are limited published data on factors affecting the LOS following LT in the modern era. Early allograft dysfunction (EAD) is an important endpoint when assessing clinical outcomes of LT, but the extent to which EAD affects postoperative resource consumption has been inadequately studied. Although there are several
L
© 2013 by Elsevier Inc. All rights reserved. 360 Park Avenue South, New York, NY 10010-1710 Transplantation Proceedings, 45, 259 –264 (2013)
definitions of EAD, the most widely accepted criteria include ⱖ1 of the following laboratory parameters by day 7: serum bilirubin ⱖ171 mmol/L, international normalized ratio ⱖ1.6, or aspartate or alanine transaminase ⱖ2,000 U/L.6 The clinical significance of EAD, which occurs in nearly a quarter of adult recipients, is that its occurrence is a validated predictor of poor graft and patient outcome in recipients receiving deceased-donor liver allografts.6 Unfortunately, liver recipients with EAD experience rates of
From the Multi-Organ Transplant Program, London Health Sciences Centre (K.P.C., R.H.-A., N.C.), Department of Surgery (K.P.C., R.H.-A.), and Department of Medicine (N.C.), Western University, London, Ontario, Canada. Address reprint requests to Natasha Chandok, MD, MPH, Multi-Organ Transplant Program, Western University, 339 Windermere Road, London, Ontario, N6A 5A5 Canada. E-mail: [email protected] 0041-1345/–see front matter http://dx.doi.org/10.1016/j.transproceed.2012.07.147 259
260
death and graft failure of ⬃20% and ⬃26%, respectively, within 6 months after transplantation.6 It is a reasonable hypothesis that EAD might affect LOS following LT, but there have been no recent publications in the current era of hepatic transplantation to assess the extent of this phenomenon. A pre– Model for End-Stage Liver Disease (MELD) era publication using the National Institute of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database had shown that durations of intensive care stay and total hospitalization after LT were significantly longer in subjects with EAD than those without (4 days vs 3 days [P ⫽ .0001] and 24 days vs 15 days [P ⫽ .001], respectively).7 Given that present liver allocation is widely based on the MELD system, it may be a reasonable hypothesis that the MELD score may exaggerate the effects of EAD on prolongation of hospitalization. Earlier authors have shown that MELD score is associated with increased postoperative LOS because MELD score is a surrogate for overall degree of sickness.8,9 However, the relationship among MELD score, EAD, and LOS in the current era of transplantation needs further exploration. The primary objective of the present study was to determine if early allograft dysfunction (EAD) influenced LOS after LT. The secondary objective was to identify additional recipient, donor, and operative factors associated with LOS. METHODS Selection and Description of Participants As part of a quality assurance initiative, consecutive adult LT recipients at London Health Sciences Center, Canada, were prospectively followed from January 2007 to September 2010. Inclusion criteria consisted of patient age ⱖ18 years and hospitalization for or resulting in LT. Patients were categorized into 3 groups based on their length of hospitalization: Group 1 (ideal length of stay) was defined as ⱕ14 days from the date of LT (based on the historical median length of stay of 14 days at our institution and on Ontario provincial targets); group 2 (long length of stay) was defined as 15–29 days from the date of LT; and group 3 (prolonged length of stay) was defined as ⱖ30 days from the date of LT. The definition of prolonged length of stay was modeled from an earlier publication.9 Approval for the study was obtained from the Institutional Review Board at Western University.
Data Collection Patient demographics and clinical characteristics that were collected included: MELD score within 24 hours of LT (calculated from the formula available at www.mayoclinic.org/meld/mayomodel6. html), underlying etiology of liver disease, presence of preexistent diabetes, donor risk index (DRI), postoperative medical debility (requiring ⬎14 days of physical therapy from time of transplantation), surgical complication (abscess requiring radiologic or surgical intervention, intra-abdominal bleeding requiring surgical intervention, hepatic artery thrombosis, portal vein thrombosis, or biliary complication [leak or stricture]), 90-day reoperation rates, and 90-day readmission rates. DRI was calculated using the validated scoring system described by Feng et al.10 EAD was defined using the definition by Olthoff et al described earlier.6
CROOME, HERNANDEZ-ALEJANDRO, AND CHANDOK EAD was only calculated in patients receiving donation after brain death (DBD) liver allografts (n ⫽ 172).
Statistical Techniques Continuous variables were reported as mean with SD, and categoric variables were reported as frequency and percentage. Differences between groups were analyzed using the unpaired t test for continuous variables and by the 2 test or continuity correction method for categoric variables. Univariate ordinal logistic regression for the odds of increased length of stay (ⱕ14 days vs 15–29 days vs ⱖ30 days) was performed. A multivariate ordinal logistic regression was performed using a backwards stepwise elimination. All statistical tests were 2 sided, and differences were considered significant when P was ⬍.05. Statistical analyses were performed using SAS version 9.1.2 (SAS, Cary, NC).
RESULTS
A total of 208 subjects fulfilled the inclusion criteria. The mean number of days in hospital was 32.0 ⫾ 34.4, and the median number of days in hospital was 17. The mean and median number of days in the intensive care unit were 6.4 ⫾ 12.7 and 3, respectively. Of the 208 subjects, 116 patients (55.8%) had ideal length of stay (group 1), 46 patients (22.1%) had long length of stay (group 2), and the remaining 46 patients (22.1%) had prolonged length of stay (group 3). The clinicodemographic characteristics of each group are summarized in Table 1. In the entire cohort, there were 31 subjects with evidence of HCC (diagnosis confirmed on explant), the majority of whom had ideal length of stay. Location of the recipient 24 hours before LT were home in 110 subjects (54.2%), hospital ward in 59 (29.1%), intensive care unit but not mechanically ventilated in 19 (9.4%), and in intensive care with mechanical ventilation in 15 (7.4%). Mean donor age was 44.8 ⫾ 18.4 years, and mean DRI was 1.57 ⫾ 0.42. In total, 25 patients (12.0%) had postoperative surgical complications, and 18 of these patients required surgical intervention for management. EAD was seen in 26 (22.4%), 8 (17.4%), and 16 (34.8%) patients in groups 1, 2, and 3, respectively. Patients fulfilling the criteria for EAD had significantly longer mean hospital LOS than patients not meeting the definition (42.5 ⫾ 38.9 days [median 25.5] vs 27.4 ⫾ 31 days [median 14], respectively; P ⫽ .003; Fig 1A). Patients meeting the definition of EAD also had significantly longer LOS in the intensive care unit than patients not meeting the definition of EAD (8.61 ⫾ 10.28 days [median 4] vs 5.45 ⫾ 11.6 days [median 3], respectively; P ⫽ .048; Fig 1B). Debility, defined in this analysis as the need for ⱖ14 days of physiotherapy, was present in 19 subjects in the postoperative period. Univariate ordinal logistic regression for the odds of increased length of stay (ⱕ14 days vs 15–29 days vs ⱖ30 days) was performed (Table 2). Patient characteristics positively associated with LOS included male sex (P ⫽ .012) and MELD score (P ⬍ .001); the presence of HCC was negatively associated with LOS (P ⬍ .001). Patient location before transplantation also was associated with LOS (P ⬍
RESOURCE UTILIZATION AFTER LIVER TRANSPLANTATION
261
Table 1. Patient Demographics and Clinical Characteristics in Group 1 (Normal Stay), Group 2 (Long Stay), and Group 3 (Prolonged Stay) Post Liver Transplant Length of Stay Normal Stay (ⱕ14 Days) (n ⫽ 116)
Age at transplantation, y Sex, Female Diabetes Diagnosis HCV/HCV ⫹ Alcohol Alcohol NAFLD Autoimmune Other/combinations Diagnosis of hepatocellular carcinoma MELD score Albumin INR Bilirubin Creatinine Location before transplantation Home Hospital ward bed ICU Donor type, NDD Donor age, y Cold ischemia time, min Surgical complication Hepatic artery thrombosis Portal vein thrombosis Bile leak Post-op bleeding Wound infection Primary nonfunction Debility Reoperation EAD, yes DRI DRI ⬍1.7
Long Stay (15–29 Days) (n ⫽ 46)
Prolonged Stay (ⱖ30 Days) (n ⫽ 46)
All (n ⫽ 208)
52.4 (9.4) 17 (14.7%) 19 (16.4%)
54.5 (11.2) 17 (37.0%) 11 (23.9%)
53.3 (9.3) 13 (28.3%) 7 (15.2%)
53.1 (9.8) 47 (22.6%) 37 (17.8%)
34 (29.3%) 7 (6.0%) 8 (6.9%) 14 (12.1%) 53 (45.7%) 27 (23.3%) 17.0 (8.6) 30.2 (5.9) 1.57 (0.68) 75.3 (121.8) 82.0 (37.9)
11 (23.9%) 8 (17.4%) 6 (13.0%) 7 (15.2%) 14 (30.4%) 3 (6.5%) 23.6 (9.6) 31.8 (7.0) 1.88 (0.62) 177.9 (188.2) 121.2 (89.4)
8 (17.4%) 7 (15.2%) 6 (13.0%) 12 (26.1%) 13 (28.3%) 1 (2.2%) 25.9 (9.1) 30.8 (8.1) 1.97 (0.74) 199.4 (221.7) 124.4 (96.1)
53 (25.5%) 22 (10.6%) 20 (9.6%) 33 (15.9%) 80 (38.5%) 31 (14.9%) 20.4 (9.7) 30.7 (6.7) 1.73 (0.70) 125.5 (172.8) 100.0 (70.4)
83 (72.8%) 21 (18.4%) 10 (8.8%) 95 (81.9%) 45.2 (17.3) 396.7 (157.9) 25 (21.6%) 1 (0.9%) 15 (12.9%) 3 (2.6%) 5 (4.3%) 4 (3.5%) 0 (0.0%) 3 (2.6%) 18 (15.7%) 26 (22.4%) 1.56 (0.42) 76 (67.3%)
14 (31.1%) 19 (42.2%) 12 (26.7%) 40 (87.0%) 43.9 (18.7) 441.7 (165.7) 11 (23.9%) 2 (4.4%) 2 (4.4%) 0 (0.0%) 5 (10.9%) 5 (10.9%) 0 (0.0%) 6 (13.0%) 10 (21.7%) 8 (17.4%) 1.62 (0.43) 23 (52.3%)
13 (29.6%) 19 (43.2%) 12 (27.3%) 37 (80.4%) 44.6 (21.0) 381.4 (165.0) 23 (50.0%) 1 (2.2%) 9 (19.6%) 8 (17.4%) 7 (15.2%) 6 (13.0%) 1 (2.2%) 10 (21.7%) 18 (40.0%) 16 (34.8%) 1.56 (0.42) 28 (66.7%)
110 (54.2%) 59 (29.1%) 34 (16.8%) 172 (82.7%) 44.8 (18.4) 403.2 (161.9) 59 (28.4%) 4 (1.9%) 26 (12.5%) 11 (5.3%) 17 (8.2%) 15 (7.2%) 1 (0.5%) 19 (9.1%) 46 (22.3%) 50 (24.0%) 1.57 (0.42) 127 (63.8%)
Values are presented as mean (SD) or n (%). Abbreviations: HCV, hepatis C virus; NAFLD, nonalcoholic fatty liver disease; MELD, Model for End-Stage Liver Disease; INR, international normalized ratio; ICU, intensive care unit; NDD, ; EAD, early allograft dysfunction; DRI, donor risk index.
.001). Postoperative factors associated with LOS included surgical complications (P ⫽ .001) and presence of EAD (P ⫽ .025). There was no association between any of the individual donor variables or DRI with length of hospitalization in ordinal logistic regression. A multivariate ordinal logistic regression, using the aforementioned variables, was performed using a backward stepwise elimination (Table 2). Patient characteristics remaining significant included the presence of HCC and MELD score (odds ratio [OR] 0.26 [95% confidence interval (CI), 0.08 – 0.85; P ⫽ .025] and OR 1.04 [95% CI, 1.01–1.08; P ⫽ .024], respectively). Patient location before LT other than home was significantly associated with LOS, with OR 2.70 (95% CI, 1.31–5.59) for ward bed and OR 2.49 (95% CI, 0.99 – 6.27) for intensive care with or without mechanical ventilation. The only postoperative factor re-
maining significant was surgical complications with OR 2.86 (P ⫽ .001). DISCUSSION
Although it is a life-saving intervention, LT consumes considerable resources. Because many costs are difficult to quantify or they vary broadly across health systems, LOS is routinely used in health services research as a surrogate of resource consumption. Earlier studies on cost of LT have shown that the most contributory element was LOS.8,11 In our prospective analysis, 55.8% of patients had ideal LOS (ⱕ14 days). Historically, our center and the Ministry of Health in the Province of Ontario, Canada, have used this definition for expected LOS in adult LT recipients. In our cohort, 22% had long stay (15–29 days), and another 22% had prolonged stay (⬎30 days). This might suggest
262
Fig 1. (A) Mean total length of hospital stay in those not meeting the definition of early allograft dysfunction compared with those meeting the definition. (B) Mean intensive care length of hospital stay in those not meeting the definition of early allograft dysfunction compared with those meeting the definition.
that the current expectations of LOS (and by extension the cost of postoperative care) do not reflect the present reality and that sicker and/or older patients with more comorbidities than in times past are receiving transplants. The present analysis clearly showed that MELD score was associated with LOS. This finding corroborates a previous multicenter study in the USA that determined that marginal liver grafts and/or transplantation of recipients with high MELD scores resulted independently in increased length of hospitalization after LT.8,12 The association of MELD with LOS has been shown also in a population from the UK.9 This latter study showed that patients with a MELD score ⬎24 had a significantly longer postoperative intensive care stay and total hospitalization, greater need for renal replacement therapy, and higher cost of care within the intensive care environment.9 Recipient location before transplantation was also significantly associated with LOS. This variable, like the MELD score, is likely a reliable indicator of a recipient’s severity of illness before transplantation. Our group has previously published our experience on the impact of age on resource consumption after LT,
CROOME, HERNANDEZ-ALEJANDRO, AND CHANDOK
and our analysis found no association between recipient age ⱖ60 years and LOS after LT, although there was a trend toward increased stay in the intensive care among older recipients, which requires a larger multicenter study for further exploration.13 A diagnosis of HCC was associated with shorter LOS on univariate regression. This likely reflects that patients listed with HCC generally have a lower MELD score and therefore less hepatic decompensation than other patients on the wait list. It was, however, somewhat surprising that this trend remained significant after adjustment for the MELD score on multivariate regression. This may reflect a type 1 error. Another hypothesis for this phenomenon is that patients with HCC awaiting LT may receive more pretransplantation care because of their need for protocolized monitoring at a transplantation center to intermittently assess the stage of malignancy, and this rigorous follow-up might translate into improved wellness before undergoing LT. Our analysis did not find any significant association between DRI and LOS. This contrasts with earlier United Network for Organ Sharing– based studies that have shown that subjects with hepatic grafts corresponding to higher DRI were found to have a significant increase in hospital LOS within MELD groups.8 The lack of significance of DRI in our study may reflect center-specific practices in donor graft selection and the fact that not all variables influencing graft outcome are captured by the DRI.14 The present study showed an association between postoperative surgical complications and LOS. This corresponds to earlier studies looking at resource allocation in liver transplantation.11,15,16 A significant association was also seen between EAD and LOS. Patients fulfilling the criteria for EAD had significantly longer hospital LOS and intensive care LOS than patients not meeting the definition. There was also a larger proportion of patients fulfilling EAD criteria in the prolonged stay group. As stated earlier, EAD was previously shown to be a valid predictor of graft and patient survival within 6 months of transplantation.6,17 The present study showed a strong correlation between EAD and LOS, suggesting that it is a predictor of post-LT resource consumption. On multivariate ordinal logistic regression, EAD likely did not maintain significance because of colinearity with both MELD and DRI, which we have previously shown to be associated with EAD.16 There are strengths and limitations in the present analysis. Strengths include the prospective design and large sample size. Furthermore, this study helps to fulfill a void in our current understanding of resource utilization following LT, and the specific focus on EAD and location of the recipient before LT have broad implications on both policy and practice. Although the monocentric nature of this study represents a limitation in terms of generalizability, it also eliminates center-specific variations in practice. Moreover, the Canadian system is a singlepayer health system, making socioeconomic factors less likely to influence time to discharge. Additional weaknesses in this study include the fact that nonmedical
RESOURCE UTILIZATION AFTER LIVER TRANSPLANTATION
263
Table 2. Univariate and Multivariate Ordinal Logistic Regression: Variables Associated with Length of Stay Univariable OR (95% CI)
Age at transplantation Sex, female Diabetes Diagnosis HCV/HCV ⫹ Alcohol Alcohol NAFLD Autoimmune Other/other combinations Diagnosis of HCC MELD score Albumin INR Bilirubin Creatinine Location before transplantation Home Hospital ward bed ICU Donor type, NDD Donor age Cold ischemia time, min Surgical complication Debility Reoperation EAD, yes DRI (continuous) DRI ⬍1.7
1.01 (0.99–1.04) 2.20 (1.19–4.06) 1.09 (0.55–2.16) 1.06 (0.52–2.16) 3.31 (1.35–8.12) 2.68 (1.06–6.80) 2.89 (1.32–6.29) Reference 0.15 (0.05–0.44) 1.09 (1.06–1.12) 1.02 (0.98–1.06) 2.12 (1.42–3.16) 1.004 (1.002–1.005) 1.007 (1.003–1.011) Reference 5.01 (2.63–9.55) 6.09 (2.84–13.07) 1.03 (0.52–2.07) 0.998 (0.984–1.012) 1.000 (0.998–1.002) 2.58 (1.45–4.58) 5.51 (2.19–13.88) 2.69 (1.45–5.01) 1.88 (1.08–3.26) 1.14 (0.60–2.17) 0.82 (0.47–1.43)
Multivariable P Value
OR (95% CI)
.399 .012 .802 .006
⬍.001 ⬍.001 .375 ⬍.001 ⬍.001 ⬍.001 ⬍.001
P Value
NS NS NS NS
0.26 (0.08–0.85) 1.04 (1.01–1.08) Not considered
.025 .024 NS NS NS .022
Reference 2.70 (1.31–5.59) 2.49 (0.99–6.27) .752 .902 .001 ⬍.001 .002 .025 .698 .475
2.86 (1.52–5.41) 3.12 (1.13–8.60)
NS NS NS .001 .028 NS NS NS NS
Abbreviations as in Table 1.
variables (e.g., social factors) that could influence LOS were not accounted for, and early postoperative resource consumption might not have a bearing on long-term resource consumption in LT recipients. In conclusion, the present study demonstrates that measures of severity of recipient illness before hepatic transplant are strong predictors of postoperative LOS, whereas donor factors based on DRI appear to be less important. The present study also shows that EAD can be used as a predictor of LOS in the current era of transplantation. From a cost-containment point of view, the results of this analysis might support policies that favor transplantation of recipients from home versus the hospital ward or intensive care. Clearly, multicenter studies assessing recipient, donor, and operative factors that influence postoperative resource consumption following LT are needed to corroborate our findings and to help advance policies to better utilize the limited resources available for patients in need of LT. Targeted studies to assess ways to reduce the incidence of EAD and improve clinical outcomes following EAD are sorely needed. It is vital that LT programs perform such analyses to support evidence-based policies that best utilize resources to service the maximum quantity of patients while preserving exceptional quality in care.
REFERENCES 1. Starzl TE, Demetris AJ, van Thiel D: Liver transplantation (2). N Engl J Med 1989;321:1092. 2. Taylor MC, Greig PD, Detsky AS, et al. Factors associated with the high cost of liver transplantation in adults. Can J Surg 2002;45:425. 3. Buchanan P, Dzebisashvili N, Lentine KL, et al. Liver transplantation cost in the Model for End-Stage Liver Disease era: looking beyond the transplant admission. Liver Transpl 2009;15: 1270. 4. Showstack J, Katz PP, Lake JR, et al. NIDDK Liver Transplantation Database Group: Resource utilization in liver transplantation: effects of patient characteristics and clinical practice. JAMA 1999;281:1381. 5. Washburn WK, Pollock BH, Nichols L, et al. Impact of recipient MELD score on resource utilization. Am J Transplant 2006;6:2449. 6. Olthoff KM, Kulik L, Samstein B, et al. Validation of a current definition of early allograft dysfunction in liver transplant recipients and analysis of risk factors. Liver Transpl 2010;16:943. 7. Deschenes M, Belle SH, Krom RA, et al. Early allograft dysfunction after liver transplantation: a definition and predictors of outcome. National Institute of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database. Transplantation 1998;66:302. 8. Washburn WK, Meo NA, Halff GA, et al. Factors influencing liver transplant length of stay at two large-volume transplant centers. Liver Transpl 2009;15:1570.
264 9. Smith JO, Shiffman ML, Behnke M, et al. Incidence of prolonged length of stay after orthotopic liver transplantation and its influence on outcomes. Liver Transpl 2009;15:273. 10. Feng S, Goodrich NP, Bragg-Gresham JL, et al. Characteristics associated with liver graft failure: the concept of a donor risk index. Am J Transplant 2006;6:783. 11. Bucuvalas JC, Zeng L, Anand R: Predictors of length of stay for pediatric liver transplant recipients. Liver Transpl 2004;10:1011. 12. Foxton MR, Al-Freah MA, Portal AJ, et al. Increased model for end-stage liver disease score at the time of liver transplant results in prolonged hospitalization and overall intensive care unit costs. Liver Transpl 2010;16:668. 13. Shankar N, Albasheer M, Marotta P, et al. Do older patients utilize excess health care resources after liver transplantation? Ann Hepatol 2011;10:477.
CROOME, HERNANDEZ-ALEJANDRO, AND CHANDOK 14. Ammori JB, Pelletier SJ, Lynch R, et al. Incremental costs of post-liver transplantation complications. J Am Coll Surg 2008;206:89. 15. Brown RS, Jr., Ascher NL, Lake JR, et al. The impact of surgical complications after liver transplantation on resource utilization. Arch Surg 1997;132:1098. 16. Croome KP, Marotta P, Wall WJ, Dale C, Levstik MA, Chandok N, Hernandez-Alejandro R. Should a Lower Quality Organ go to the Least Sick Patient?:MELD and Donor Risk Index as Predictors of Early Allograft. Transplant Proc. 2012 Jun;44(5): 1303– 6. 17. Croome KP, Quan D, Wall W, Hernandez-Alejandro R. Evaluation of the Updated Definition of Early Allograft Dysfunction in Donation after Brain Death and Donation after Cardiac Death Liver Allografts. Hepatobiliary Pancreat Dis Int. 2012 Aug;11(4):372– 6.
L
© 2013 by Elsevier Inc. All rights reserved. 360 Park Avenue South, New York, NY 10010-1710 Transplantation Proceedings, 45, 259 –264 (2013)
definitions of EAD, the most widely accepted criteria include ⱖ1 of the following laboratory parameters by day 7: serum bilirubin ⱖ171 mmol/L, international normalized ratio ⱖ1.6, or aspartate or alanine transaminase ⱖ2,000 U/L.6 The clinical significance of EAD, which occurs in nearly a quarter of adult recipients, is that its occurrence is a validated predictor of poor graft and patient outcome in recipients receiving deceased-donor liver allografts.6 Unfortunately, liver recipients with EAD experience rates of
From the Multi-Organ Transplant Program, London Health Sciences Centre (K.P.C., R.H.-A., N.C.), Department of Surgery (K.P.C., R.H.-A.), and Department of Medicine (N.C.), Western University, London, Ontario, Canada. Address reprint requests to Natasha Chandok, MD, MPH, Multi-Organ Transplant Program, Western University, 339 Windermere Road, London, Ontario, N6A 5A5 Canada. E-mail: [email protected] 0041-1345/–see front matter http://dx.doi.org/10.1016/j.transproceed.2012.07.147 259
260
death and graft failure of ⬃20% and ⬃26%, respectively, within 6 months after transplantation.6 It is a reasonable hypothesis that EAD might affect LOS following LT, but there have been no recent publications in the current era of hepatic transplantation to assess the extent of this phenomenon. A pre– Model for End-Stage Liver Disease (MELD) era publication using the National Institute of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database had shown that durations of intensive care stay and total hospitalization after LT were significantly longer in subjects with EAD than those without (4 days vs 3 days [P ⫽ .0001] and 24 days vs 15 days [P ⫽ .001], respectively).7 Given that present liver allocation is widely based on the MELD system, it may be a reasonable hypothesis that the MELD score may exaggerate the effects of EAD on prolongation of hospitalization. Earlier authors have shown that MELD score is associated with increased postoperative LOS because MELD score is a surrogate for overall degree of sickness.8,9 However, the relationship among MELD score, EAD, and LOS in the current era of transplantation needs further exploration. The primary objective of the present study was to determine if early allograft dysfunction (EAD) influenced LOS after LT. The secondary objective was to identify additional recipient, donor, and operative factors associated with LOS. METHODS Selection and Description of Participants As part of a quality assurance initiative, consecutive adult LT recipients at London Health Sciences Center, Canada, were prospectively followed from January 2007 to September 2010. Inclusion criteria consisted of patient age ⱖ18 years and hospitalization for or resulting in LT. Patients were categorized into 3 groups based on their length of hospitalization: Group 1 (ideal length of stay) was defined as ⱕ14 days from the date of LT (based on the historical median length of stay of 14 days at our institution and on Ontario provincial targets); group 2 (long length of stay) was defined as 15–29 days from the date of LT; and group 3 (prolonged length of stay) was defined as ⱖ30 days from the date of LT. The definition of prolonged length of stay was modeled from an earlier publication.9 Approval for the study was obtained from the Institutional Review Board at Western University.
Data Collection Patient demographics and clinical characteristics that were collected included: MELD score within 24 hours of LT (calculated from the formula available at www.mayoclinic.org/meld/mayomodel6. html), underlying etiology of liver disease, presence of preexistent diabetes, donor risk index (DRI), postoperative medical debility (requiring ⬎14 days of physical therapy from time of transplantation), surgical complication (abscess requiring radiologic or surgical intervention, intra-abdominal bleeding requiring surgical intervention, hepatic artery thrombosis, portal vein thrombosis, or biliary complication [leak or stricture]), 90-day reoperation rates, and 90-day readmission rates. DRI was calculated using the validated scoring system described by Feng et al.10 EAD was defined using the definition by Olthoff et al described earlier.6
CROOME, HERNANDEZ-ALEJANDRO, AND CHANDOK EAD was only calculated in patients receiving donation after brain death (DBD) liver allografts (n ⫽ 172).
Statistical Techniques Continuous variables were reported as mean with SD, and categoric variables were reported as frequency and percentage. Differences between groups were analyzed using the unpaired t test for continuous variables and by the 2 test or continuity correction method for categoric variables. Univariate ordinal logistic regression for the odds of increased length of stay (ⱕ14 days vs 15–29 days vs ⱖ30 days) was performed. A multivariate ordinal logistic regression was performed using a backwards stepwise elimination. All statistical tests were 2 sided, and differences were considered significant when P was ⬍.05. Statistical analyses were performed using SAS version 9.1.2 (SAS, Cary, NC).
RESULTS
A total of 208 subjects fulfilled the inclusion criteria. The mean number of days in hospital was 32.0 ⫾ 34.4, and the median number of days in hospital was 17. The mean and median number of days in the intensive care unit were 6.4 ⫾ 12.7 and 3, respectively. Of the 208 subjects, 116 patients (55.8%) had ideal length of stay (group 1), 46 patients (22.1%) had long length of stay (group 2), and the remaining 46 patients (22.1%) had prolonged length of stay (group 3). The clinicodemographic characteristics of each group are summarized in Table 1. In the entire cohort, there were 31 subjects with evidence of HCC (diagnosis confirmed on explant), the majority of whom had ideal length of stay. Location of the recipient 24 hours before LT were home in 110 subjects (54.2%), hospital ward in 59 (29.1%), intensive care unit but not mechanically ventilated in 19 (9.4%), and in intensive care with mechanical ventilation in 15 (7.4%). Mean donor age was 44.8 ⫾ 18.4 years, and mean DRI was 1.57 ⫾ 0.42. In total, 25 patients (12.0%) had postoperative surgical complications, and 18 of these patients required surgical intervention for management. EAD was seen in 26 (22.4%), 8 (17.4%), and 16 (34.8%) patients in groups 1, 2, and 3, respectively. Patients fulfilling the criteria for EAD had significantly longer mean hospital LOS than patients not meeting the definition (42.5 ⫾ 38.9 days [median 25.5] vs 27.4 ⫾ 31 days [median 14], respectively; P ⫽ .003; Fig 1A). Patients meeting the definition of EAD also had significantly longer LOS in the intensive care unit than patients not meeting the definition of EAD (8.61 ⫾ 10.28 days [median 4] vs 5.45 ⫾ 11.6 days [median 3], respectively; P ⫽ .048; Fig 1B). Debility, defined in this analysis as the need for ⱖ14 days of physiotherapy, was present in 19 subjects in the postoperative period. Univariate ordinal logistic regression for the odds of increased length of stay (ⱕ14 days vs 15–29 days vs ⱖ30 days) was performed (Table 2). Patient characteristics positively associated with LOS included male sex (P ⫽ .012) and MELD score (P ⬍ .001); the presence of HCC was negatively associated with LOS (P ⬍ .001). Patient location before transplantation also was associated with LOS (P ⬍
RESOURCE UTILIZATION AFTER LIVER TRANSPLANTATION
261
Table 1. Patient Demographics and Clinical Characteristics in Group 1 (Normal Stay), Group 2 (Long Stay), and Group 3 (Prolonged Stay) Post Liver Transplant Length of Stay Normal Stay (ⱕ14 Days) (n ⫽ 116)
Age at transplantation, y Sex, Female Diabetes Diagnosis HCV/HCV ⫹ Alcohol Alcohol NAFLD Autoimmune Other/combinations Diagnosis of hepatocellular carcinoma MELD score Albumin INR Bilirubin Creatinine Location before transplantation Home Hospital ward bed ICU Donor type, NDD Donor age, y Cold ischemia time, min Surgical complication Hepatic artery thrombosis Portal vein thrombosis Bile leak Post-op bleeding Wound infection Primary nonfunction Debility Reoperation EAD, yes DRI DRI ⬍1.7
Long Stay (15–29 Days) (n ⫽ 46)
Prolonged Stay (ⱖ30 Days) (n ⫽ 46)
All (n ⫽ 208)
52.4 (9.4) 17 (14.7%) 19 (16.4%)
54.5 (11.2) 17 (37.0%) 11 (23.9%)
53.3 (9.3) 13 (28.3%) 7 (15.2%)
53.1 (9.8) 47 (22.6%) 37 (17.8%)
34 (29.3%) 7 (6.0%) 8 (6.9%) 14 (12.1%) 53 (45.7%) 27 (23.3%) 17.0 (8.6) 30.2 (5.9) 1.57 (0.68) 75.3 (121.8) 82.0 (37.9)
11 (23.9%) 8 (17.4%) 6 (13.0%) 7 (15.2%) 14 (30.4%) 3 (6.5%) 23.6 (9.6) 31.8 (7.0) 1.88 (0.62) 177.9 (188.2) 121.2 (89.4)
8 (17.4%) 7 (15.2%) 6 (13.0%) 12 (26.1%) 13 (28.3%) 1 (2.2%) 25.9 (9.1) 30.8 (8.1) 1.97 (0.74) 199.4 (221.7) 124.4 (96.1)
53 (25.5%) 22 (10.6%) 20 (9.6%) 33 (15.9%) 80 (38.5%) 31 (14.9%) 20.4 (9.7) 30.7 (6.7) 1.73 (0.70) 125.5 (172.8) 100.0 (70.4)
83 (72.8%) 21 (18.4%) 10 (8.8%) 95 (81.9%) 45.2 (17.3) 396.7 (157.9) 25 (21.6%) 1 (0.9%) 15 (12.9%) 3 (2.6%) 5 (4.3%) 4 (3.5%) 0 (0.0%) 3 (2.6%) 18 (15.7%) 26 (22.4%) 1.56 (0.42) 76 (67.3%)
14 (31.1%) 19 (42.2%) 12 (26.7%) 40 (87.0%) 43.9 (18.7) 441.7 (165.7) 11 (23.9%) 2 (4.4%) 2 (4.4%) 0 (0.0%) 5 (10.9%) 5 (10.9%) 0 (0.0%) 6 (13.0%) 10 (21.7%) 8 (17.4%) 1.62 (0.43) 23 (52.3%)
13 (29.6%) 19 (43.2%) 12 (27.3%) 37 (80.4%) 44.6 (21.0) 381.4 (165.0) 23 (50.0%) 1 (2.2%) 9 (19.6%) 8 (17.4%) 7 (15.2%) 6 (13.0%) 1 (2.2%) 10 (21.7%) 18 (40.0%) 16 (34.8%) 1.56 (0.42) 28 (66.7%)
110 (54.2%) 59 (29.1%) 34 (16.8%) 172 (82.7%) 44.8 (18.4) 403.2 (161.9) 59 (28.4%) 4 (1.9%) 26 (12.5%) 11 (5.3%) 17 (8.2%) 15 (7.2%) 1 (0.5%) 19 (9.1%) 46 (22.3%) 50 (24.0%) 1.57 (0.42) 127 (63.8%)
Values are presented as mean (SD) or n (%). Abbreviations: HCV, hepatis C virus; NAFLD, nonalcoholic fatty liver disease; MELD, Model for End-Stage Liver Disease; INR, international normalized ratio; ICU, intensive care unit; NDD, ; EAD, early allograft dysfunction; DRI, donor risk index.
.001). Postoperative factors associated with LOS included surgical complications (P ⫽ .001) and presence of EAD (P ⫽ .025). There was no association between any of the individual donor variables or DRI with length of hospitalization in ordinal logistic regression. A multivariate ordinal logistic regression, using the aforementioned variables, was performed using a backward stepwise elimination (Table 2). Patient characteristics remaining significant included the presence of HCC and MELD score (odds ratio [OR] 0.26 [95% confidence interval (CI), 0.08 – 0.85; P ⫽ .025] and OR 1.04 [95% CI, 1.01–1.08; P ⫽ .024], respectively). Patient location before LT other than home was significantly associated with LOS, with OR 2.70 (95% CI, 1.31–5.59) for ward bed and OR 2.49 (95% CI, 0.99 – 6.27) for intensive care with or without mechanical ventilation. The only postoperative factor re-
maining significant was surgical complications with OR 2.86 (P ⫽ .001). DISCUSSION
Although it is a life-saving intervention, LT consumes considerable resources. Because many costs are difficult to quantify or they vary broadly across health systems, LOS is routinely used in health services research as a surrogate of resource consumption. Earlier studies on cost of LT have shown that the most contributory element was LOS.8,11 In our prospective analysis, 55.8% of patients had ideal LOS (ⱕ14 days). Historically, our center and the Ministry of Health in the Province of Ontario, Canada, have used this definition for expected LOS in adult LT recipients. In our cohort, 22% had long stay (15–29 days), and another 22% had prolonged stay (⬎30 days). This might suggest
262
Fig 1. (A) Mean total length of hospital stay in those not meeting the definition of early allograft dysfunction compared with those meeting the definition. (B) Mean intensive care length of hospital stay in those not meeting the definition of early allograft dysfunction compared with those meeting the definition.
that the current expectations of LOS (and by extension the cost of postoperative care) do not reflect the present reality and that sicker and/or older patients with more comorbidities than in times past are receiving transplants. The present analysis clearly showed that MELD score was associated with LOS. This finding corroborates a previous multicenter study in the USA that determined that marginal liver grafts and/or transplantation of recipients with high MELD scores resulted independently in increased length of hospitalization after LT.8,12 The association of MELD with LOS has been shown also in a population from the UK.9 This latter study showed that patients with a MELD score ⬎24 had a significantly longer postoperative intensive care stay and total hospitalization, greater need for renal replacement therapy, and higher cost of care within the intensive care environment.9 Recipient location before transplantation was also significantly associated with LOS. This variable, like the MELD score, is likely a reliable indicator of a recipient’s severity of illness before transplantation. Our group has previously published our experience on the impact of age on resource consumption after LT,
CROOME, HERNANDEZ-ALEJANDRO, AND CHANDOK
and our analysis found no association between recipient age ⱖ60 years and LOS after LT, although there was a trend toward increased stay in the intensive care among older recipients, which requires a larger multicenter study for further exploration.13 A diagnosis of HCC was associated with shorter LOS on univariate regression. This likely reflects that patients listed with HCC generally have a lower MELD score and therefore less hepatic decompensation than other patients on the wait list. It was, however, somewhat surprising that this trend remained significant after adjustment for the MELD score on multivariate regression. This may reflect a type 1 error. Another hypothesis for this phenomenon is that patients with HCC awaiting LT may receive more pretransplantation care because of their need for protocolized monitoring at a transplantation center to intermittently assess the stage of malignancy, and this rigorous follow-up might translate into improved wellness before undergoing LT. Our analysis did not find any significant association between DRI and LOS. This contrasts with earlier United Network for Organ Sharing– based studies that have shown that subjects with hepatic grafts corresponding to higher DRI were found to have a significant increase in hospital LOS within MELD groups.8 The lack of significance of DRI in our study may reflect center-specific practices in donor graft selection and the fact that not all variables influencing graft outcome are captured by the DRI.14 The present study showed an association between postoperative surgical complications and LOS. This corresponds to earlier studies looking at resource allocation in liver transplantation.11,15,16 A significant association was also seen between EAD and LOS. Patients fulfilling the criteria for EAD had significantly longer hospital LOS and intensive care LOS than patients not meeting the definition. There was also a larger proportion of patients fulfilling EAD criteria in the prolonged stay group. As stated earlier, EAD was previously shown to be a valid predictor of graft and patient survival within 6 months of transplantation.6,17 The present study showed a strong correlation between EAD and LOS, suggesting that it is a predictor of post-LT resource consumption. On multivariate ordinal logistic regression, EAD likely did not maintain significance because of colinearity with both MELD and DRI, which we have previously shown to be associated with EAD.16 There are strengths and limitations in the present analysis. Strengths include the prospective design and large sample size. Furthermore, this study helps to fulfill a void in our current understanding of resource utilization following LT, and the specific focus on EAD and location of the recipient before LT have broad implications on both policy and practice. Although the monocentric nature of this study represents a limitation in terms of generalizability, it also eliminates center-specific variations in practice. Moreover, the Canadian system is a singlepayer health system, making socioeconomic factors less likely to influence time to discharge. Additional weaknesses in this study include the fact that nonmedical
RESOURCE UTILIZATION AFTER LIVER TRANSPLANTATION
263
Table 2. Univariate and Multivariate Ordinal Logistic Regression: Variables Associated with Length of Stay Univariable OR (95% CI)
Age at transplantation Sex, female Diabetes Diagnosis HCV/HCV ⫹ Alcohol Alcohol NAFLD Autoimmune Other/other combinations Diagnosis of HCC MELD score Albumin INR Bilirubin Creatinine Location before transplantation Home Hospital ward bed ICU Donor type, NDD Donor age Cold ischemia time, min Surgical complication Debility Reoperation EAD, yes DRI (continuous) DRI ⬍1.7
1.01 (0.99–1.04) 2.20 (1.19–4.06) 1.09 (0.55–2.16) 1.06 (0.52–2.16) 3.31 (1.35–8.12) 2.68 (1.06–6.80) 2.89 (1.32–6.29) Reference 0.15 (0.05–0.44) 1.09 (1.06–1.12) 1.02 (0.98–1.06) 2.12 (1.42–3.16) 1.004 (1.002–1.005) 1.007 (1.003–1.011) Reference 5.01 (2.63–9.55) 6.09 (2.84–13.07) 1.03 (0.52–2.07) 0.998 (0.984–1.012) 1.000 (0.998–1.002) 2.58 (1.45–4.58) 5.51 (2.19–13.88) 2.69 (1.45–5.01) 1.88 (1.08–3.26) 1.14 (0.60–2.17) 0.82 (0.47–1.43)
Multivariable P Value
OR (95% CI)
.399 .012 .802 .006
⬍.001 ⬍.001 .375 ⬍.001 ⬍.001 ⬍.001 ⬍.001
P Value
NS NS NS NS
0.26 (0.08–0.85) 1.04 (1.01–1.08) Not considered
.025 .024 NS NS NS .022
Reference 2.70 (1.31–5.59) 2.49 (0.99–6.27) .752 .902 .001 ⬍.001 .002 .025 .698 .475
2.86 (1.52–5.41) 3.12 (1.13–8.60)
NS NS NS .001 .028 NS NS NS NS
Abbreviations as in Table 1.
variables (e.g., social factors) that could influence LOS were not accounted for, and early postoperative resource consumption might not have a bearing on long-term resource consumption in LT recipients. In conclusion, the present study demonstrates that measures of severity of recipient illness before hepatic transplant are strong predictors of postoperative LOS, whereas donor factors based on DRI appear to be less important. The present study also shows that EAD can be used as a predictor of LOS in the current era of transplantation. From a cost-containment point of view, the results of this analysis might support policies that favor transplantation of recipients from home versus the hospital ward or intensive care. Clearly, multicenter studies assessing recipient, donor, and operative factors that influence postoperative resource consumption following LT are needed to corroborate our findings and to help advance policies to better utilize the limited resources available for patients in need of LT. Targeted studies to assess ways to reduce the incidence of EAD and improve clinical outcomes following EAD are sorely needed. It is vital that LT programs perform such analyses to support evidence-based policies that best utilize resources to service the maximum quantity of patients while preserving exceptional quality in care.
REFERENCES 1. Starzl TE, Demetris AJ, van Thiel D: Liver transplantation (2). N Engl J Med 1989;321:1092. 2. Taylor MC, Greig PD, Detsky AS, et al. Factors associated with the high cost of liver transplantation in adults. Can J Surg 2002;45:425. 3. Buchanan P, Dzebisashvili N, Lentine KL, et al. Liver transplantation cost in the Model for End-Stage Liver Disease era: looking beyond the transplant admission. Liver Transpl 2009;15: 1270. 4. Showstack J, Katz PP, Lake JR, et al. NIDDK Liver Transplantation Database Group: Resource utilization in liver transplantation: effects of patient characteristics and clinical practice. JAMA 1999;281:1381. 5. Washburn WK, Pollock BH, Nichols L, et al. Impact of recipient MELD score on resource utilization. Am J Transplant 2006;6:2449. 6. Olthoff KM, Kulik L, Samstein B, et al. Validation of a current definition of early allograft dysfunction in liver transplant recipients and analysis of risk factors. Liver Transpl 2010;16:943. 7. Deschenes M, Belle SH, Krom RA, et al. Early allograft dysfunction after liver transplantation: a definition and predictors of outcome. National Institute of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database. Transplantation 1998;66:302. 8. Washburn WK, Meo NA, Halff GA, et al. Factors influencing liver transplant length of stay at two large-volume transplant centers. Liver Transpl 2009;15:1570.
264 9. Smith JO, Shiffman ML, Behnke M, et al. Incidence of prolonged length of stay after orthotopic liver transplantation and its influence on outcomes. Liver Transpl 2009;15:273. 10. Feng S, Goodrich NP, Bragg-Gresham JL, et al. Characteristics associated with liver graft failure: the concept of a donor risk index. Am J Transplant 2006;6:783. 11. Bucuvalas JC, Zeng L, Anand R: Predictors of length of stay for pediatric liver transplant recipients. Liver Transpl 2004;10:1011. 12. Foxton MR, Al-Freah MA, Portal AJ, et al. Increased model for end-stage liver disease score at the time of liver transplant results in prolonged hospitalization and overall intensive care unit costs. Liver Transpl 2010;16:668. 13. Shankar N, Albasheer M, Marotta P, et al. Do older patients utilize excess health care resources after liver transplantation? Ann Hepatol 2011;10:477.
CROOME, HERNANDEZ-ALEJANDRO, AND CHANDOK 14. Ammori JB, Pelletier SJ, Lynch R, et al. Incremental costs of post-liver transplantation complications. J Am Coll Surg 2008;206:89. 15. Brown RS, Jr., Ascher NL, Lake JR, et al. The impact of surgical complications after liver transplantation on resource utilization. Arch Surg 1997;132:1098. 16. Croome KP, Marotta P, Wall WJ, Dale C, Levstik MA, Chandok N, Hernandez-Alejandro R. Should a Lower Quality Organ go to the Least Sick Patient?:MELD and Donor Risk Index as Predictors of Early Allograft. Transplant Proc. 2012 Jun;44(5): 1303– 6. 17. Croome KP, Quan D, Wall W, Hernandez-Alejandro R. Evaluation of the Updated Definition of Early Allograft Dysfunction in Donation after Brain Death and Donation after Cardiac Death Liver Allografts. Hepatobiliary Pancreat Dis Int. 2012 Aug;11(4):372– 6.