Key words : Delisting, Organ allocation, SF-LDQOL

Introduction

Liver transplantation (LT) is a lifesaving treatment for patients with end-stage liver disease.^1-3 However, there is a significant shortage of donor livers, and transplant programs have an ethical obligation to ensure that the limited supply of donor organs is allocated to those candidates who are most likely to benefit from LT.⁴ It is critical that transplant programs develop a rigorous candidate surveillance process for patients on wait lists to determine those who are nonadherent.^5,6 Unfortunately, nonadherence in candidates on wait lists is not uncommon and has been shown to be as high as 16.7% in LT candidates.⁷ Nonadherence with prescribed therapies, lack of attendance to clinical appointments, not following medical recommendations, and active substance use are recognized as contraindications to remaining on wait lists by most transplant programs.^3,8 In addition, it has been observed that patients in lower resourced socioeconomic areas have higher rates of nonadherence to medical care.^9,10

Removing a candidate from the LT wait list is stressful to the patient, to their family, and to care providers. Methods to proactively identify LT candidates at risk of being removed from the wait list due to nonadherence may help programs provide extra support to candidates to improve adherence and avoid delisting. However, there is currently no objective, standardized tool available to evaluate risk of nonadherence in LT candidates.⁸ Given the correlation between quality of life (QOL) and lower resourced socioeconomic areas to adherence behavior,^9-11 we hypothesized that LT candidate QOL metrics and socioeconomic metrics, such as the Area Deprivation Index (ADI), can help identify LT candidates at increased risk of wait list removal for nonadherence. In this single-center study, we investigated the associations between the QOL measures in the Liver Disease Health-Related Quality of Life Short Form (SF-LDQOL) and ADI to evaluate risk of wait list removal due to nonadherence for the following reasons: alcohol or illicit drug use or lack of attendance to clinical appointments.

Materials and Methods

Study population
We identified, among 720 adult patients (≥18 years old) listed for LT at the University of Washington Medical Center from September 1, 2012, to August 30, 2017, 358 candidates who completed the SF-LDQOL prior to being listed for LT. The SF-LDQOL survey was given to patients during a visit to the pretransplant clinic with the guidance to return within 1 week. Approval for this study was obtained from the University of Washington Human Subjects Division. Candidate QOL was assessed using the SF-LDQOL.^12,13 The SF-LDQOL has demonstrated reliability and validity in evaluating health-related QOL in patients with advanced chronic liver disease. Candidates requiring multiorgan transplants were excluded. Questions 3 and 4 in the survey were excluded from our analysis due to missing data.

Data collection
Data for this analysis were recorded prospectively and obtained from the University of Washington’s Transplant Database at time of listing. The wait list was followed until January 1, 2021, to determine the various reasons for wait list removal. There were 5 reasons for wait list removal: nonadherence, transplant, death or becoming too sick for transplant, cancer progression out of accordance with the Milan criteria, and clinical improvement. Wait list removal due to nonadherence was determined by a multi-disciplinary LT committee. Reasons for wait list removal due to nonadherence included candidate use of alcohol or illicit drugs while on the wait list or failure to attend clinical appointments after several attempts.

All data were recorded at the time of listing for LT. We collected candidate demographic and medical characteristics, including age, address (street, city, state), etiology of liver disease (viral hepatitis, alcoholic liver disease, nonalcoholic steatohepatitis, cholestatic liver disease, cryptogenic cirrhosis, retransplant, and other), associated hepatocellular carcinoma (HCC), sex, race/ethnicity (Asian, Black, Hispanic, White, and Other), presence of diabetes mellitus, body mass index, health insurance type (Medicare, Medicaid, or private), ABO blood type, presence of portal vein thrombosis, previous abdominal surgery, being on dialysis, Model for End-Stage Liver Disease (MELD) score as calculated by laboratory values, patient location at time of listing (outpatient, non-intensive care unit inpatient, and intensive care unit inpatient), and presence of ascites as determined by physical examination. Candidates with serum albumin <2.6 g/dL (lowest quartile) were recorded as low, and others were recorded as normal. Candidates with platelet count <57 000/μL (lowest quartile) were recorded as low, and others were recorded as normal. The functional status, including active (no assistance for activities of daily living required), low (assistance for activities of daily living required), or functional status unknown, was clinically scored by a transplant hepatologist at listing.

Liver Disease Health-Related Quality of Life Short Form responses
Liver transplant candidates completed the SF-LDQOL from September 1, 2012, to August 30, 2017. Only SF-LDQOL responses completed before listing were included in the analysis. Several questions in the SF-LDQOL shared similar themes. To reduce variables in the competing risk analysis, we combined thematically similar questions into single variables and the majority response for that variable was recorded. We reduced the 45 separate questions in the SF-LDQOL to 21 responses. Table 1 contains our labeling scheme for survey questions.

Area Deprivation Index
The 2019 ADI file (block group files version 3.0) was obtained through the University of Wisconsin-Madison School of Medicine and Public Health.14,15 This file provided ADI scores for 220 333 census block groups in the United States. A census block group is the smallest geographic unit that is considered a “neighborhood.” According to the University of Washington institutional review board regulations for electronic transfer of addresses, patient addresses were geocoded to obtain a 12-digit Federal Information Processing Standards (FIPS) code. This 12-digit FIPS code was then linked to the ADI file containing FIPS codes and ADI scores to assign each patient a neighborhood ADI value according to the census block group in which that patient resided. An ADI percentile ranking of 1 represents the lowest level of deprivation among US census block groups, and an ADI of 100 represents the highest level of deprivation. We analyzed ADI as a continuous variable and separately as a categorical variable. Liver transplant candidates in this study were stratified into 2 ADI groups: low deprivation (ADI of 1 to 60) and high deprivation (ADI of >60). This categorization was done with kernel-smoothing to find the change in the linear pattern in association with delisting for nonadherence and multiple binning for different ranges to determine the best association to nonadherence.

Statistical analyses
Competing risk analysis was selected as the time-to-event analysis due to 5 different competing risks or endpoints for being removed from the wait list. Univariable and multivariable analyses were performed to estimate risks of wait list removal due to nonadherence associated with SF-LDQOL responses, ADI scores, and candidate clinical variables at time of listing.¹⁶ To ensure accuracy of the estimated regression coefficients, we used the number of events per variables as proposed by Austin and colleagues.¹⁷ Their proposal stated that, if all covariates are binary with a moderate prevalence, then the number of events per variable can be as low as 10. Because only 23 candidates were removed from the wait list because of nonadherence, we could only have 2 variables in the competing risk analysis as per Austin and colleagues.¹⁷ We elected to evaluate all 2-variable combinations for the remaining clinical variables, including the ADI score as one group and all remaining SF-LDQOL questions as another group. Only the significant variables were recorded in the multivariable analysis. Candidates who remained actively listed at the end of follow-up were censored. Before we conducted the multivariable competing risk analysis for wait list removal due to nonad-herence, several highly correlated variables were removed. Diagnosis of HCC, patient location at time of listing, presence of ascites, and low serum albumin level were all highly correlated with MELD score, so only MELD score was kept in the analysis. The SF-LDQOL questions pertaining to “change in health within last year” highly correlated with the questions pertaining to “overall health,” so only “overall health” was kept. Other survey questions excluded from the competing risk analysis due to high correlation were “food tastes bad,” “activity limited,” “burdened by liver disease,” “drowsy during the day,” “dark future,” and “others feel embarrassed of me.”

Wait list removal risks are presented as subdis-tribution hazard ratios (SHR) with 95% confidence intervals. Continuous demographic variables are presented as median and interquartile ranges (IQR). Categorical demographic variables are presented as counts and percentages. Few clinical variables had missing values, and these were given median values if continuous and the majority values if categorical. All variables were evaluated for significant correlation, and those demonstrating collinearity were removed from further analysis. P < .05 indicated statistical significance. Statistical analyses were performed using R version 4.0.0 and the cmprsk 2.2-10 package.¹⁶

Results

Study population and baseline characteristics
For the 720 total LT candidates listed at the University of Washington Medical Center from September 1, 2012, to August 30, 2017, the ages ranged from 18 to 72 years and the median age was 58.2 years (IQR, 51.5-62.7 y). The median ADI was 29 (IQR, 18-45) (Figure 1).

Table 2 presents the baseline characteristics of our study population, consisting of 358 LT candidates who completed and returned the SF-LDOQL survey before listing versus 362 LT candidates who did not complete and return the survey. Among the 358 (49.7%) candidates who completed the SF-LDQOL before listing, median time before date of listing was 52 days (IQR, 24-122 days). Among these LT candidates who completed the SF-LDQOL, 205 (57.3%) underwent LT, 79 (22.1%) died or became too sick, 23 (6.4%) were removed from the wait list due to nonadherence, 19 (5.3%) had cancer progression, and 19 (5.3%) experienced clinical improvement. Thirteen LT candidates (3.6%) remained actively listed at the end of follow-up.

Of the 23 candidates removed from wait list for nonadherence, 7 (30.4%) were because of alcohol use, 4 (17.4%) were because of illicit drug use, and 12 (52.2%) were because of lack of attendance to clinical appointments after several attempts. For the 7 who were removed for alcohol use, only 3 had a prior history of alcohol use disorder. Only 1 of the 4 removed for illicit drug use had a prior known history of illicit drug use disorder.

Candidates who completed the SF-LDQOL spent more months on the wait list (median of 10.2 mo; IQR, 4.2-22.7 mo) than candidates who did not (median of 4.2 mo; IQR, 0.7-12.2 mo). Candidates who completed the SF-LDQOL also had lower MELD scores and were more likely to have active functional status. In additions, candidates who completed the survey were less likely to be on dialysis, to be admitted as inpatients, or to have hepatic encephalopathy than candidates who did not complete the survey. Of the candidates who completed the survey and who were removed from the wait list for nonadherence, approximately 40% were removed within the first 12 months of listing and approximately 60% were removed by 24 months (Figure 2).

The most commonly endorsed QOL measures in the SF-LDQOL were “bodily pain” (74.9%), “change in health within last year” (58.1%), “swelling in feet/legs” (55.6%), and “activity limited” (55.6%) (Table 1). Over one-third (36.0%) endorsed “poor memory,” while 15.1% endorsed “poor future outlook.” This met the standard of the variables being binary with a moderate prevalence. Questions 3 and 4 were left blank by most candidates, and there was no pattern to the many candidates who did not answer questions 3 and 4.

Associations in Liver Disease Health-Related Quality of Life Short Form survey responses, the Area Deprivation Index scores, and clinical variables for wait list removal due to nonadherence
In the univariable competing risk analysis, female sex (SHR: 0.26; 95% CI, 0.09-0.76; P = .01) and previous abdominal surgery (SHR: 0.29; 95% CI, 0.12-0.75; P = .01) were associated with a lower risk of wait list removal due to nonadherence and were the only significant variables (Table 3). The SF-LDQOL responses indicating “poor memory” (SHR: 3.44; 95% CI, 1.45-8.12; P = .005) and “poor future outlook” (SHR: 3.97; 95% CI, 1.44-10.96; P = .008) were associated with a higher risk of wait list removal due to nonadherence and were the only significant variables. The ADI as a continuous variable or a categorical variable of 61 to 100 was not significantly associated with risk of wait list removal due to nonadherence in the univariable analysis.

In the multivariable competing risk analysis for the clinical variables, female sex (SHR: 0.31; 95% CI, 0.10-0.93; P = .04) and previous abdominal surgery (SHR: 0.36; 95% CI, 0.14-0.92; P = .03) continued to be the only variables associated with lower risk of wait list removal. In the multivariable analysis for the SF-LDQOL survey, responses indicating “poor memory” (SHR: 3.53; 95% CI, 1.49-8.36; P = .004) and “poor future outlook” (SHR: 2.94; 95% CI, 1.07-8.07; P = .03) continued to be the only variables associated with increased risk of wait list removal due to nonadherence (Table 4). Of note, when using these 4 significant variables in all 2-variable combinations, the variables of “poor memory” and “poor future outlook” were always significant when controlling for female sex or previous abdominal surgery.

Discussion

In this retrospective cohort study of LT candidates, we found that SF-LDQOL responses indicating “poor memory” and “poor future outlook” were inde-pendently associated with an increased risk of wait list removal due to nonadherence as a result of alcohol or illicit drug use or lack of attendance to clinical appointments. The use of these questions from the SF-LDQOL, as part of a comprehensive transplant evaluation, may help identify candidates who are at an increased risk of delisting for nonadherence. Proactive identification of these LT candidates would enable transplant programs to provide additional support for these patients to hopefully prevent wait list removal.

There are multiple biopsychosocial reasons for why LT candidates at risk of being removed from the wait list for nonadherence may report “poor memory.” Further research is needed to investigate the mechanisms through which patients who report “poor memory” result in wait list removal. A well-known complication of advanced liver disease is hepatic encephalopathy, which can impair memory.¹⁸ However, only self-reported “poor memory” and not hepatic encephalopathy was associated with wait list removal due to nonadherence in our multivariable competing risk analysis, suggesting that factors beyond hepatic encephalopathy are contributing to the risk of delisting due to nonadherence. In our study, candidates were not removed from the wait list due to encephalopathy. Other potential causes of “poor memory” could include depression,¹⁹ substance use disorder,²⁰ and social isolation.²¹ There is a high relapse rate in patients who have previously struggled with substance use disorder,²² and resumption of active substance use is an indication for removal from the transplant list. However, our study patients removed from the wait list for active alcohol or substance use were less likely to have had a known prior history of alcohol or substance use disorder.

“Poor future outlook” responses may indicate low self-efficacy, poor perceived current health, or depressive symptoms, which have all been tied to nonadherence.^23,24 Liver transplant candidates who perceive their present health to be poor may be more pessimistic about their future and see futility in adhering to medical recommendations despite the negative effect that this would have on their health and listing status. Again, further research is needed to investigate the mechanisms through which patients who report “poor future outlook” ultimately lead to LT wait list removal.

An SF-LDQOL response indicating “limited companionship” was not found to be significantly associated with an increased risk of wait list removal due to nonadherence, as had been previously reported.^7,23,24 Patients in this study had undergone a thorough social work evaluation and had been deemed to have adequate care support in order to be actively listed for LT. Likewise, ADI score was not associated with wait list removal for nonadherence in our analysis. We had expected to find that ADI scores indicating high deprivation areas to be associated with an increased risk of wait list removal, given that nonadherence has previously been observed to be associated with lower socioeconomic status.^9,10 Because these were all LT candidates who were initially listed, presumably these candidates had already shown sufficient access to insurance or financial support to be selected for listing despite their ADI score.

Female sex and previous abdominal surgery were associated with a lower risk of wait list removal due to nonadherence, consistent with prior studies.^25,26 We and others suspected that patients who have previously undergone abdominal surgery may be accustomed to the associated preoperative routines and may therefore be more likely to adhere to these routines while on the transplant wait list.

Limitations
We acknowledge study limitations. First, we had a 49.7% completion and return rate of the survey. Those who completed the survey had lower MELD scores and better functional status, so more critically ill patients were likely underrepresented in our study as they were less likely to have completed a survey. In addition, patients with higher MELD scores tend to spend less time on the transplant list as they are either selected to undergo transplant sooner or experience death or clinical deterioration sooner; thus they have less time to develop nonadherence leading to wait list removal as defined in our study. The competing risk analyses adjusted for this effect. Second, the single-center study design may limit its generalizability to other transplant centers. However, delisting due to alcohol or illicit drug use or lack of attendance at medical appointments is generally considered standard procedure for transplant programs.^3,8 Finally, although our recorded patient demographic variables did not include current psychiatric data at the time of listing, future studies investigating the prevalence of psychiatric comorbidities, such as depression, in LT candidates may be informative in a study of wait list removal for nonadherence. However, because we noted that most patients removed from the wait list for alcohol or illicit drug use did not have a prior history of substance use disorder, prior comorbidities may not be completely helpful.

Conclusions
Our study demonstrated that, in LT candidates, self-reporting “poor memory” and “poor future outlook” in the SF-LDQOL was associated with a higher risk of eventual wait list removal due to nonadherence for alcohol or illicit drug use or lack of attendance to clinical appointments. In future work, we aim to use a modified version of the SF-LDQOL that focuses on these 2 measures to screen for LT candidates who may be at risk of delisting for nonadherence. By proactively identifying these at-risk LT candidates, we are better able to connect these patients with additional resources and improve their chances of avoiding removal from the wait list.

References:

1. Jadlowiec CC, Taner T. Liver transplantation: Current status and challenges. World J Gastroenterol. 2016;22(18):4438-4445. doi:10.3748/wjg.v22.i18.4438
   CrossRef - PubMed
   
2. Merion RM, Schaubel DE, Dykstra DM, Freeman RB, Port FK, Wolfe RA. The survival benefit of liver transplantation. Am J Transplant. 2005;5(2):307-313. doi:10.1111/j.1600-6143.2004.00703.x
   CrossRef - PubMed
   
3. Varma V, Mehta N, Kumaran V, Nundy S. Indications and contraindications for liver transplantation. Int J Hepatol. 2011;2011:121862. doi:10.4061/2011/121862
   CrossRef - PubMed
   
4. Meirelles Junior RF, Salvalaggio P, Rezende MB, et al. Liver transplantation: history, outcomes and perspectives. Einstein (Sao Paulo). 2015;13(1):149-152. doi:10.1590/S1679-45082015RW3164
   CrossRef - PubMed
   
5. Telles-Correia D, Barbosa A, Mega I, Monteiro E. Adherence correlates in liver transplant candidates. Transplant Proc. 2009;41(5):1731-1734. doi:10.1016/j.transproceed.2009.02.067
   CrossRef - PubMed
   
6. Majumder P, Sarkar S. A review of the prevalence of illicit substance use in solid-organ transplant candidates and the effects of illicit substance use on solid-organ transplant treatment outcomes. Cureus. 2020;12(7):e8986. doi:10.7759/cureus.8986
   CrossRef - PubMed
   
7. Dobbels F, Vanhaecke J, Desmyttere A, Dupont L, Nevens F, De Geest S. Prevalence and correlates of self-reported pretransplant nonadherence with medication in heart, liver, and lung transplant candidates. Transplantation. 2005;79(11):1588-1595. doi:10.1097/01.tp.0000158430.06507.87
   CrossRef - PubMed
   
8. Sandal S, Chen T, Cantarovich M. Evaluation of transplant candidates with a history of nonadherence: an opinion piece. Can J Kidney Health Dis. 2021;8:2054358121990137. doi:10.1177/2054358121990137
   CrossRef - PubMed
   
9. Berquist RK, Berquist WE, Esquivel CO, Cox KL, Wayman KI, Litt IF. Adolescent non-adherence: prevalence and consequences in liver transplant recipients. Pediatr Transplant. 2006;10(3):304-310. doi:10.1111/j.1399-3046.2005.00451.x
   CrossRef - PubMed
   
10. Wayda B, Clemons A, Givens RC, et al. Socioeconomic disparities in adherence and outcomes after heart transplant: a UNOS (United Network for Organ Sharing) registry analysis. Circ Heart Fail. 2018;11(3):e004173. doi:10.1161/circheartfailure.117.004173
    CrossRef - PubMed
    
11. Zhao SM, Dong FF, Qiu HZ, Li D. Quality of life, adherence behavior, and social support among renal transplant recipients in China: a descriptive correlational study. Transplant Proc. 2018;50(10):3329-3337. doi:10.1016/j.transproceed.2018.05.026
    CrossRef - PubMed
    
12. Gralnek IM, Hays RD, Kilbourne A, et al. Development and evaluation of the Liver Disease Quality of Life instrument in persons with advanced, chronic liver disease--the LDQOL 1.0. Am J Gastroenterol. 2000;95(12):3552-3565. doi:10.1111/j.1572-0241.2000.03375.x
    CrossRef - PubMed
    
13. Kanwal F, Spiegel BM, Hays RD, et al. Prospective validation of the short form liver disease quality of life instrument. Aliment Pharmacol Ther. 2008;28(9):1088-1101. doi:10.1111/j.1365-2036.2008.03817.x
    CrossRef - PubMed
    
14. University of Wisconsin School of Medicine and Public Health. The Neighborhood Atlas. About the Area Deprivation Index. Accessed August 2021. https://www.neighborhoodatlas.medicine.wisc.edu/
    CrossRef - PubMed
    
15. Kind AJH, Buckingham WR. Making neighborhood-disadvantage metrics accessible - the neighborhood atlas. N Engl J Med. 2018;378(26):2456-458. doi:10.1056/NEJMp1802313
    CrossRef - PubMed
    
16. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446):496-509. Doi:10.2307/2670170
    CrossRef - PubMed
    
17. Austin PC, Allignol A, Fine JP. The number of primary events per variable affects estimation of the subdistribution hazard competing risk model. J Clin Epidemiol. 2017;83:75-84. doi:10.1016/j.jclinepi.2016.11.017
    CrossRef - PubMed
    
18. Bahceci F, Yildirim B, Karincaoglu M, Dogan I, Sipahi B. Memory impairment in patients with cirrhosis. J Natl Med Assoc. 2005;97(2):213-216.
    CrossRef - PubMed
    
19. Dillon DG, Pizzagalli DA. Mechanisms of memory disruption in depression. Trends Neurosci. 2018;41(3):137-149. doi:10.1016/j.tins.2017.12.006
    CrossRef - PubMed
    
20. Verdejo-Garcia A, Garcia-Fernandez G, Dom G. Cognition and addiction. Dialogues Clin Neurosci. 2019;21(3):281-290. doi:10.31887/DCNS.2019.21.3/gdom
    CrossRef - PubMed
    
21. Read S, Comas-Herrera A, Grundy E. Social isolation and memory decline in later-life. J Gerontol B Psychol Sci Soc Sci. 2020;75(2):367-376. doi:10.1093/geronb/gbz152
    CrossRef - PubMed
    
22. Nunes EV, Gordon M, Friedmann PD, et al. Relapse to opioid use disorder after inpatient treatment: Protective effect of injection naltrexone. J Subst Abuse Treat. 2018;85:49-55. doi:10.1016/j.jsat.2017.04.016
    CrossRef - PubMed
    
23. Denhaerynck K, Dobbels F, Cleemput I, et al. Prevalence, consequences, and determinants of nonadherence in adult renal transplant patients: a literature review. Transpl Int. 2005;18(10):1121-1133. doi:10.1111/j.1432-2277.2005.00176.x
    CrossRef - PubMed
    
24. Nevins TE, Nickerson PW, Dew MA. Understanding medication nonadherence after kidney transplant. J Am Soc Nephrol. 2017;28(8):2290-2301. doi:10.1681/ASN.2017020216
    CrossRef - PubMed
    
25. Kobayashi S, Tsutsui J, Okabe S, Hideki I, Akaho R, Nishimura K. Medication nonadherence after kidney transplantation: an internet-based survey in Japan. Psychol Health Med. 2020;25(1):91-101. doi:10.1080/13548506.2019.1622745
    CrossRef - PubMed
    
26. Hu W, Hu S, Zhu Y, Chen H, Chen Y. Assessing oral medication adherence and identifying predictors of low adherence in Chinese inflammatory bowel disease patients. Patient Prefer Adherence. 2020;14:1083-1092. doi:10.2147/PPA.S250935
    CrossRef - PubMed