In a 12-Year Study, Sustainability of Quality of Life Benefits After Liver Transplantation Varies With Pretransplantation Diagnosis

GASTROENTEROLOGY 2010;139:1619 –1629

In a 12-Year Study, Sustainability of Quality of Life Benefits After Liver Transplantation Varies With Pretransplantation Diagnosis KRISTINE RUPPERT,* SHIHCHEN KUO,* ANDREA DIMARTINI,‡ and VIJAYAN BALAN§

BACKGROUND & AIMS: As life-extending benefits of liver transplantation (LTX) are realized, the focus of improving outcomes after LTX transitioned from merely extending quantity of life to improving quality of life (QOL). Numerous cross-sectional studies have shown that QOL improves within 1 year after LTX; however, the long-term prospective pattern of QOL is not known. We assessed the sustainability of early QOL benefits after LTX. METHODS: Patients who underwent LTX (n ⫽ 381) were followed for 12 years. We collected clinical information, survival data, and data on 5 QOL domains: physical distress (PHY), psychological distress (PSY), social/role function (SRF), personal function (PF), and general health perception (GHP). Mixed model analysis was used to determine whether initial gains in QOL were sustained long term. Outcomes were analyzed according to the primary liver diagnosis. RESULTS: After 12 years, liver recipients had marked declines in PHY (P ⬍ .001), SRF (P ⫽ .006), and GHP (P ⬍ .001) scores, whereas improvements in PSY (P ⫽ .09) and PF (P ⫽ .09) were sustained. Women had worse outcomes in PHY and PF than men. Patients with autoimmune disease had decreased QOL PHY, SRF, PF, and GHP domains. Patients with alcoholic liver disease and hepatitis C initially had lower QOL in all domains, with the greatest decreases in PHY (P ⬍ .001) and PF (P ⫽ .03). CONCLUSIONS: Although QOL improves within 1 year after patients receive liver transplants, not all groups of patients achieve or sustain the same level of QOL for 12 years. QOL decreases with time in most areas. Efforts should be made to improve QOL and function in the initial recovery period in liver transplant recipients. Keywords: Liver Transplantation; Quality of Life; Chronic Liver Disease.

F

or more than 2 decades as the life-prolonging benefits of liver transplantation (LTX) have been achieved, the focus of improving outcomes after LTX has transitioned from merely extending quantity of life to enhancing quality of life (QOL). QOL is particularly important in health care, where it does not simply mean the absence of disease or infirmity.1 As a predictor, QOL can even outweigh other classic predictors of survival.2

QOL encompasses physical, mental, and social well-being and is best assessed with the use of specially designed and tested instruments, measuring patient’s perceptions and functioning along these 3 parameters. Ten years ago, 2 exhaustive, quantitative reviews of the existing QOL literature, including all solid organ types (218 independent studies),3 and a meta-analysis of all LTX QOL studies (49 independent studies)4 concluded that QOL is reliably improved by transplantation. Although these studies did not find transplant recipients’ QOL equaled nonpatient normative samples, the most consistent finding was improvements from before to after LTX in QOL across a variety of QOL domains. In LTX, in particular, all QOL studies found overall QOL improvements from before to after LTX,5–9 ⬎80% of studies found improvements in physical functioning and ⱖ60% of studies found improvements in mental and social functioning.3,4 When considering the severity of most LTX candidate’s physical health before LTX, it is perhaps not surprising that comparatively their immediate and short-term (first several years after LTX) QOL improves. However, whether QOL gains in the short term are sustained long term had not been investigated and still has not. On the basis of these findings, the investigators of these meta-analyses recommended that future studies investigate (1) the stability of QOL improvements over time and in addition to transplant-specific factors that may account for within-sample QOL variability, (2) personal and environmental factors that may affect QOL positively or negatively. Since that time, several studies of longer term (ie, ⬎10 years after LTX) QOL have shown most liver transplant recipients are satisfied with their health status and overall QOL,10,11 but some have limitations in their mobility, Abbreviations used in this paper: AIH, autoimmune hepatitis; ALD, alcoholic liver disease; CDC, Centers for Disease Control and Prevention; HBV, hepatitis B virus; HCV, hepatitis C virus; HDV, hepatitis D virus; INR, The international normalized ratio; LTD, Liver Transplantation Database; LTX, liver transplantation; MELD, Model for End-Stage Liver Disease; NIDDK, National Institutes of Diabetes and Digestive and Kidney Diseases; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; PT, prothrombin time; QOL, quality of life. © 2010 by the AGA Institute 0016-5085/$36.00 doi:10.1053/j.gastro.2010.06.043

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

*Department of Epidemiology, University of Pittsburgh Graduate School of Public Health, Pittsburgh, Pennsylvania; ‡Department of Psychiatry, Thomas E. Starzl Transplant Institute, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania; and §Division of Transplantation Medicine, Mayo Clinic, Phoenix, Arizona

1620

RUPPERT ET AL

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

usual activities, physical functioning,11,12 cognitive functioning,13 and pain.12 Even after 10 years, recipients did not rate their health-related QOL as similar to healthy controls.11,13,14 However, these studies are limited by small sample sizes, cross-sectional designs, and the lack of examination of specific correlates of QOL. Identifying factors that over time enhance or diminish QOL may provide the critical information to extend QOL benefits to all liver transplant recipients. We now have the opportunity to examine these long-term QOL outcomes in a more systematic way. Our intent was to address this knowledge gap by analyzing data from a large nationally representative LTX sample as well as use mixed effects models, with random effects for patient. No QOL study for LTX has been found that used this method. With this longitudinal study design, it (1) allows us to characterize the within-individual change in QOL over time and (2) permits the discovery of individual characteristics (eg, primary diagnosis for LTX) that can explain the interindividual differences in change in QOL over time. These issues are not possible to assess with the use of a cross-sectional design. In 1990, the National Institutes of Diabetes and Digestive and Kidney Diseases (NIDDK) established a database of liver transplant recipients from 3 large transplantation programs.15 We have used this dataset to address our specific questions on the stability of and factors correlated with QOL changes over time.

Materials and Methods Setting and Participants A total of 916 patients received liver transplants between April 15, 1990, and June 30, 1994, at 1 of 3 medical centers: Mayo Clinic, Rochester, Minnesota; University of Nebraska, Omaha, Nebraska; or University of California at San Francisco. Clinical follow-up data were collected prospectively through June 30, 1995. Vital status and QOL were collected through November 2002. All data were obtained from the NIDDK-Liver Transplant Database (LTD). The original study was coordinated at the University of Pittsburgh’s Graduate School of Public Health, Pittsburgh, Pennsylvania. The institutional review boards of each of the centers approved the study protocol. Of the 916 people who received a transplant, 668 were single organ nonfulminant adults (aged ⱖ 16 years) who survived the first year. To investigate the aims of this study and examine long-term QOL, only patients who completed a QOL questionnaire both at year 1 after LTX and at least 1 other at year 4 or later were selected in the analysis. The final sample size was 381 recipients with 1966 QOL forms (Supplementary Figure 1). We examined QOL from year 1 after LTX longitudinally through 12 years after LTX. Although the primary objective of this study was to examine those recipients of a transplant for

GASTROENTEROLOGY Vol. 139, No. 5

chronic liver diseases, a limited analysis of patients who received a transplant for fulminant liver failure was also performed to compare them with those who received a transplant for chronic liver diseases.

Measurement Instruments The NIDDK-LTD is a longitudinal dataset with follow-up data ⱕ12 years after LTX. The NIDDK-LTD QOL questionnaire includes both physical and mental domains. Specific areas include physical and psychological distress, social and role function, personal function, and general health perception. Within these domains are measures such as perceived levels of distress caused by symptoms of disease, general well-being, and the ability to perform routine physical and social functions. The LTD QOL questionnaire for adults, used in similar patient populations,5,7,9,14 was developed from standardized instruments (Appendix 1). Specifics for each domain are detailed in a previous LTD study.16 The QOL questionnaire was completed at each anniversary of the initial transplantation either during the follow-up visit or by mail. To increase the response rate, multiple mailings were used. If more than 1 questionnaire was returned within a year, the last one was kept for analysis.

Physical Distress This domain contains 21 items that describe physical symptoms. The responder graded the severity of the symptoms in the last month from 0 to 4 (0 ⫽ not at all and 4 ⫽ extremely; maximum score is 84, indicating all symptoms with the greatest severity).

Psychological Distress This domain consisted of 5 items related to anxiety and depression. Physical symptoms were rated 0 (not at all) to 4 (extremely) with a maximum distress score of 20.

Social and Role Function This 8-item domain measures the state of health on emotional, professional, and social life with higher scores indicating worse social and role functioning (range, 0 –20).

Personal Function This domain measures the ability to work as well as the level of any physical handicap. The scoring for this 9-item questionnaire can range from 0 to 4 with the higher scores indicating greater personal function.

General Health Perception This domain measures general health perception. This score is made up of 5 items with higher scores indicating better general health perception (range, 0 –10).

Statistical Analyses Descriptive statistics (mean, median, and range) of the demographic variables, Model for End-Stage Liver

Disease (MELD) scores at the point of LTX, and follow-up years were estimated initially. Variables used for determining the MELD score (prothrombin time, bilirubin, and creatinine) were collected in all patients. The international normalized ratio (INR) was calculated using the formula: INR ⫽ prothrombin time (PT; patient)/PT (normal) ⫻ International Sensitivity Index. Various mixed-effects regression models with the random intercept and slope17–19 were used to test (1) whether the person’s QOL status in each domain changed, (2) whether QOL differed by primary diagnosis for transplantation, and (3) whether individual items in each domain varied over the 12-year follow-up period. This modeling technique was used for 3 reasons. First, such models make inferences with information on all available data from the entire cohort collected at all follow-up time points. Second, these models handle missing data more efficiently than other methods of analysis or data imputation.20 Third, mixed models are powerful for analyzing correlated data. A key feature of mixed models is that, by introducing random effects in addition to fixed effects, they allow one to account for multiple source of variations, eg, within- and between-subject variations in the longitudinal study. Factors selected a priori included time (years after LTX; continuous linear covariate) and recurrence of disease (yes/no), and baseline (time of LTX) covariates: age at the initial transplantation (centered at the mean value), gender, primary diagnosis for transplantation [5 groups: (1) viral hepatitis, ie, hepatitis B virus (HBV), HBV/hepatitis D virus (HDV), or hepatitis C virus (HCV), (2) autoimmune (autoimmune hepatitis/primary biliary cirrhosis/ primary sclerosing cholangitis or AIH/PBC/PSC), (3) alcoholic liver disease (ALD), (4) cryptogenic/other, and (5) ALD⫹HCV (ALD⫹C), and MELD score (centered at the mean value)]. To determine whether HCV (not combined with HBV, HBV/HDV) alone was different from the other diagnoses groups, post hoc analysis was conducted. Routine model building processes were used to select the final models. Interaction terms that involved gender and diagnosis with time were explored. Secondary analysis included examining the effects of personal and environmental factors, such as living, employment, and tobacco and alcohol use. Two-tailed P values less than .05 were considered statistically significant for main and interaction effects. SAS Version 9.2 (SAS Institute Inc, Cary, NC) was used for the analyses. Further details of the analysis are in Appendix 2.

Results Of the 381 recipients, 53% were male and 85% were white. The mean age was 40 years, MELD score was 16.9, and follow-up time was 9.7 years. The primary causes for transplantation were PSC (24%) and PBC (17%); 63/381 (16.5%) of the population died before year

LONG-TERM QOL AFTER LIVER TRANSPLANTATION

1621

12 (25% because of liver failure, 18% because of malignancy, and 57% because of other causes). Recurrence of disease was found in only 66 (17%) recipients. As expected, 89% were in those persons who received a transplant for HCV or ALD⫹C. Table 1 details demographics of the 381 study cohort recipients (column B) and those (column A) excluded in this analysis. The study population had baseline characteristics similar to the excluded group, except more were white, fewer were Hispanics, and more had PBC or PSC with fewer HCV or ALD⫹C. In addition, the study population expectedly had the longer follow-up period. Recurrence of disease and MELD score were not significant in any of the models; thus, they were dropped from further analysis. Analyses that used diagnosis group (HBV, HBV/HDV, or HCV) or HCV alone all yielded similar results; therefore, the graphs and tables reflect only the combined group.

Physical Distress After adjusting for age and gender, the overall adjusted mean score (higher score ⫽ more physical distress) for all recipients increased 0.44 points per year (P ⬍ .001). Females in all diagnostic groups reported more physical distress at year 1 than males (3.15 points higher; P ⫽ .003). Within-group comparisons showed that the ALD⫹C group reported worse QOL at year 1 (mean, 19.9) and also had the greatest rate of decline (1.28 points per year; P ⬍ .001). Likewise, autoimmune and cryptogenic/other groups had a significant decline (0.46- and 0.49-point increase per year; P ⬍ .001 and P ⫽ .002). Among each diagnosis group, females reported more physical distress at year 1 than males (4.36 points higher; P ⬍ .001), but there was no difference in their rates of decline. Between-group comparisons showed that the ALD⫹C group had a greater rate of decline (P range, .02–.004) compared with the other 4 groups (Table 2). The fulminant group had a nonsignificant rate of decline (0.37 points per year; P ⫽ .30), which was similar to the ALD⫹C group (P ⫽ .07; see Table 2, all nonfulminant patients and Figures 1 and 2) (output data for the fulminant patients are presented in Appendix 3). When patients with HCV were analyzed as their own diagnosis group (not combined with those with HBV or HBV/HDV), the results were similar with those from analysis of combined viral hepatitis: The HCV group had no significant QOL decline (0.12-point increase per year; P ⫽ .529), and the ALD⫹C group had a significantly greater rate of QOL decline (P ⫽ .003) than the HCV group. The percentage of people reporting at least a moderate level of distress in specific symptoms at 1 year after LTX were (fatigue 45%, muscle weakness 44%, excessive appetite 32%, headaches 31%, backaches 30%, and bruising/ fragile skin 30%). When the levels of physical distress were compared over time, mixed-effects logistic models showed that almost all (16/21) symptoms were found to worsen over time. Symptoms not found to worsen over

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

November 2010

1622

RUPPERT ET AL

GASTROENTEROLOGY Vol. 139, No. 5

Table 1. Baseline (Time of Transplantation) Demographics of Patients Included and Not Included in the Study Patients who were alive at year 1 or beyond after transplantation (N ⫽ 668)

Variable

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

Recipient gender, n (%) Male Female Recipient race, n (%) White Black Hispanic Other Diagnosis before LTX diagnosis, n (%) HBV or HBV⫹HDV HCV ALD ALD⫹C AIH Cryptogenic PBC PSC Other Recipient age at first transplantation (y) Mean ⫾ SD Median Range MELD score Mean ⫾ SD Median Range Follow-up period (y)e Mean ⫾ SD Median Range

Column A (patients who were not included in the studyb) (n ⫽ 287)c

Column B (patients who had at least 1 QOL form both at year 1 and year 4 or beyond post transplantation; study cohort) (n ⫽ 381)d

171 (59.6) 116 (40.4)

203 (53.3) 178 (46.7)

216 (75.3) 11 (3.8) 36 (12.5) 24 (8.4)

323 (84.8) 14 (3.7) 27 (7.1) 17 (4.5)

12 (4.2) 56 (19.5) 41 (14.3) 30 (10.5) 18 (6.3) 32 (11.1) 28 (9.8) 28 (9.8) 42 (14.6)

14 (3.7) 54 (14.2) 38 (10.0) 20 (5.2) 26 (6.8) 39 (10.2) 63 (16.5) 91 (23.9) 36 (9.4)

48.8 ⫾ 11.2 49.4 16.4–71.7

48.9 ⫾ 11.1 49.7 16.1–71.9

17.2 ⫾ 6.0 16.0 8.0–40.0

16.9 ⫾ 5.9 16.0 7.0–40.0

5.9 ⫾ 3.3 6.3 0.5–12.0

9.7 ⫾ 1.5 9.9 4.2–12.5

Pa .10

.01

⬍.001

.87

.55

⬍.001

AIH, autoimmune hepatitis; ALD, alcoholic liver disease; ALD⫹C, alcoholic liver disease⫹hepatitis C virus; HBV, hepatitis B virus; HCV, hepatitis C virus; HDV, hepatitis D virus; LTX, liver transplantation; MELD, model for end-stage liver disease; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; QOL, quality of life; SD, standard deviation. aFor continuous variables, P values were derived from 2-sample t test for comparing mean values between groups. For categorical variables, P values were derived from ␹2 test for independence. bPatients were not included in the study if they (1) had no QOL form at year 1 after transplantation, (2) died before year 4 after transplantation, or (3) had no QOL form at year 4 or beyond after transplantation. cTen patients had missing information on MELD score. dSeven patients had missing information on MELD score. ePatients were followed to the latest date for which vital status was known.

time were nausea/vomiting, abdominal swelling, trembling, poor vision, and jaundice.

Psychological Distress Examining the anxiety and depression levels in these recipients, it was found that the overall adjusted mean score (higher score ⫽ more distress) was 4.8 at year 1 and increased 0.04 points per year (P ⫽ .09). The ALD⫹C group reported the worst QOL at year 1 (mean, 6.66). However, comparing the progression of anxiety and depression over time within and between the 5 diagnosis groups, there were no significant differences (Table 2). When including the fulminants in the analysis, the ALD⫹C group still reported the worst QOL at year 1

after LTX, followed closely by the fulminant group (mean, 5.73; Figure 1C; Appendix 3). Again, when patients with just HCV were analyzed as their own diagnosis group, the results were similar with those from analysis of combined viral hepatitis. The HCV group had no significant QOL decline (0.04-point increase per year; P ⫽ .09) and no difference in rate of QOL decline compared with the other groups. Twenty-eight percent of this population reported at least a moderate amount of sleeplessness and mood swings at 1 year after LTX. Other areas of moderate distress included nervousness (26%), feelings of depression (25%), and difficulty concentrating (23%). Only 7% had feelings of happiness; however, over time only feel-

November 2010

Table 2. Age- and Gender-Adjusted Model and the Final Model for Longitudinal Analysis of 5 QOL Domains in Patients From the NIDDK-LTD Physical distress (high score ⫽ bad QOL) Factor

P

␤ (95% CI)

Social and role function (high score ⫽ good QOL)

Personal function (high score ⫽ good QOL)

P

␤ (95% CI)

P

15.9 (15.4, 16.3) ⫺0.07 (⫺0.12, ⫺0.02) 0.002 (⫺0.03, 0.03) 0.39 (⫺0.20, 0.98)

⬍.001 .006 .90 .19

␤ (95% CI)

General health perception (high score ⫽ good QOL) P

␤ (95% CI)

P

1.95 (1.77, 2.13) ⫺0.02 (⫺0.03, 0.002) ⫺0.02 (⫺0.03, ⫺0.01) ⫺0.13 (⫺0.38, 0.12)

⬍.001 .09 ⬍.001 .30

8.20 (7.96, 8.44) ⫺0.21 (⫺0.24, ⫺0.18) ⫺0.01 (⫺0.03, 0.01) ⫺0.06 (⫺0.40, 0.28)

⬍.001 ⬍.001 .20 .72

⬍.001 .03 ⬍.001 .02

7.15 (6.41, 7.89) ⫺0.21 (⫺0.24, ⫺0.18) ⫺0.01 (⫺0.03, 0.003) —

⬍.001 ⬍.001 .11 —

14.9 (13.4, 16.3) 0.44 (0.30, 0.58) 0.07 (⫺0.03, 0.17) 3.15 (1.05, 5.25)

⬍.001 ⬍.001 .15 .003

4.82 (4.27, 5.37) 0.04 (⫺0.01, 0.09) ⫺0.03 (⫺0.07, 0.01) 0.10 (⫺0.67, 0.87)

⬍.001 .09 .10 .80

19.9 (15.2, 24.6) 1.28 (0.62, 1.94) 0.08 (⫺0.01, 0.18) 4.36 (2.14, 6.57)

⬍.001 ⬍.001 .09 ⬍.001

6.66 (4.95, 8.37) 0.04 (⫺0.01, 0.09) ⫺0.03 (⫺0.06, 0.01) —

⬍.001 .09 .16 —

13.6 (12.1, 15.2) 0.10 (⫺0.14, 0.34) ⫺0.005 (⫺0.03, 0.02) —

⬍.001 .42 .74 —

1.41 (0.82, 1.99) ⫺0.09 (⫺0.17, ⫺0.01) ⫺0.02 (⫺0.03, ⫺0.01) ⫺0.31 (⫺0.57, ⫺0.05)

⫺3.55 (⫺8.91, 1.81) ⫺7.02 (⫺12.11, ⫺1.92) ⫺5.42 (⫺11.28, 0.43) ⫺5.56 (⫺10.90, ⫺0.22) Reference

.19 .007 .07 .04 —

⫺1.23 (⫺3.17, 0.71) ⫺2.17 (⫺3.97, ⫺0.37) ⫺2.07 (⫺4.19, 0.06) ⫺1.71 (⫺3.63, 0.20) Reference

.21 .02 .06 .08 —

1.74 (0.02, 3.47) 2.65 (1.05, 4.26) 2.41 (0.54, 4.29) 3.00 (1.27, 4.74) Reference

.05 .001 .01 ⬍.001 —

0.24 (⫺0.43, 0.91) 0.90 (0.26, 1.53) 0.51 (⫺0.22, 1.24) 0.56 (⫺0.11, 1.22) Reference

.48 .006 .17 .10 —

⫺1.10 (⫺1.84, ⫺0.36) ⫺0.82 (⫺1.51, ⫺0.13) ⫺1.00 (⫺1.81, ⫺0.19) ⫺0.79 (⫺1.52, ⫺0.06) Reference

.004 .02 .02 .03 —

— — — — —

— — — — —

⫺0.06 (⫺0.33, 0.21) ⫺0.17 (⫺0.42, 0.07) ⫺0.29 (⫺0.58, 0.01) ⫺0.24 (⫺0.50, 0.03) Reference

.67 .17 .05 .08 —

0.13 (0.04, 0.22) 0.06 (⫺0.02, 0.15) 0.05 (⫺0.05, 0.15) 0.08 (⫺0.01, 0.17) Reference

.005 .14 .29 .07 —

— — — — —

— — — — —

0.04 (⫺0.01, 0.09) 0.04 (⫺0.01, 0.09) 0.04 (⫺0.01, 0.09) 0.04 (⫺0.01, 0.09) 0.04 (⫺0.01, 0.09)

.09 .09 .09 .09 .09

0.04 (⫺0.08, 0.16) ⫺0.08 (⫺0.14, ⫺0.01) ⫺0.19 (⫺0.36, ⫺0.02) ⫺0.14 (⫺0.26, ⫺0.02) 0.10 (⫺0.14, 0.34)

.53 .03 .03 .02 .42

0.04 (⫺0.003, 0.08) ⫺0.03 (⫺0.05, ⫺0.002) ⫺0.04 (⫺0.09, 0.02) ⫺0.01 (⫺0.05, 0.03) ⫺0.09 (⫺0.17, ⫺0.01)

.07 .04 .22 .67 .03

⫺0.21 (⫺0.24, ⫺0.18) ⫺0.21 (⫺0.24, ⫺0.18) ⫺0.21 (⫺0.24, ⫺0.18) ⫺0.21 (⫺0.24, ⫺0.18) ⫺0.21 (⫺0.24, ⫺0.18)

⬍.001 ⬍.001 ⬍.001 ⬍.001 ⬍.001

—

—

As indicated above

.11 .03 .04 .67 .23 .34 .17 .65 .05 .08

As indicated above

—

—

0.18 (⫺0.15, 0.51) 0.46 (0.26, 0.65) 0.28 (⫺0.19, 0.75) 0.49 (0.17, 0.80) 1.28 (0.62, 1.94)

.29 ⬍.001 .24 .002 ⬍.001

As indicated above

.16 .73 .19 .004 .50 .88 .02 .48 .02 .03

.009 .04 .11 .005 .76 .44 .14 .43 .29 .07

0.96 (0.11, 1.80) 1.18 (0.40, 1.96) 0.98 (0.06, 1.90) 0.99 (0.16, 1.83) Reference

.03 .003 .04 .02 —

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

1623

AIH, autoimmune hepatitis; ALD, alcoholic liver disease; ALD⫹C, alcoholic liver disease⫹hepatitis C virus; CI, confidence interval; HBV, hepatitis B virus; HCV, hepatitis C virus; HDV, hepatitis D virus; MELD, Model for End-Stage Liver Disease; NIDDK-LTD, National Institutes of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; QOL, quality of life. Only those with P value ⬍.10 are included. aMale patients were the reference group. bDiagnosis group 1 (DXG 1) included patients with hepatitis virus (HBV/HBV⫹HDV/HCV); diagnosis group 2 (DXG 2) included patients with autoimmune (AIH/PBC/PSC); diagnosis group 3 (DXG 3) included patients with ALD; and diagnosis group 4 (DXG 4) included patients with cryptogenic/other. Diagnosis group 5 (DXG 5), including patients with ALD⫹C, was the reference group (fulminant data can be found in Appendix 3). cRegardless of statistical significance of time, age, and primary diagnosis for transplantation, they were included in the final models, and 5 covariates (ie, gender, recurrence of disease, MELD score, and the interaction terms of gender-by-time and diagnosis-by-time) were also adjusted in the final models if they were of statistical significance.

LONG-TERM QOL AFTER LIVER TRANSPLANTATION

Age- and gender-adjusted model Intercept Time (per year) Age (per year) Gender (female)a Final modelc Intercept Time (per year) Age (per year) Gender (female)a Primary diagnosis for transplantationb HBV/HBV⫹HDV/HCV (DXG 1) AIH/PBC/PSC (DXG 2) ALD (DXG 3) Cryptogenic/other (DXG 4) ALD⫹C (DXG 5) Time ⫻ primary diagnosis for transplantationb Time ⫻ DXG 1 Time ⫻ DXG 2 Time ⫻ DXG 3 Time ⫻ DXG 4 Time ⫻ DXG 5 Testing for the within-DXG annual change in quality of life over time from the final model Primary diagnosis for transplantationb DXG 1 DXG 2 DXG 3 DXG 4 DXG 5 Comparison of the between-DXG annual change in quality of life over time from the final model Primary diagnosis for transplantationb DXG 1 vs DXG 2 DXG 1 vs DXG 3 DXG 1 vs DXG 4 DXG 1 vs DXG 5 DXG 2 vs DXG 3 DXG 2 vs DXG 4 DXG 2 vs DXG 5 DXG 3 vs DXG 4 DXG 3 vs DXG 5 DXG 4 vs DXG 5

␤ (95% CI)

Psychological distress (high score ⫽ bad QOL)

1624

RUPPERT ET AL

GASTROENTEROLOGY Vol. 139, No. 5

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

Figure 1. Age- and gender-adjusted longitudinal results of QOL domains by diagnosis before LTX: physical distress (A and B), psychological distress (C), and social and role function (D).

ings of unhappiness significantly increased (P ⫽ .001). No significant changes were found in the other areas of psychological status.

Social and Role Function When examining how the recipients perceived their social and professional life, the overall adjusted mean score (higher score ⫽ better QOL) was 15.9 at year 1 and decreased 0.07 points per year (P ⫽ .006). When QOL was compared within each diagnosis group, the ALD⫹C and hepatitis groups showed slight improvements in QOL, whereas the others (autoimmune, ALD, and cryptogenic/other) showed declines (0.08-, 0.19-, and 0.14-point decrease per year; P ⫽ .03, P ⫽ .03, and P ⫽ .02, respectively). Again, the patients in the ALD⫹C group started out with the worst QOL at year 1 (mean,

13.6). Between-group comparisons showed differences in the rate of decline between the hepatitis and ALD group (0.04 vs ⫺0.19 per year; P ⫽ .03) as well as the hepatitis and cryptogenic/other group (0.04 vs ⫺0.14 per year; P ⫽ .04). No gender differences were noted (Table 2). When including the fulminants in the analysis, they started out with the worst QOL at year 1 after LTX (mean, 13.5; Figure 1D; Appendix 3), followed closely by the ALD⫹C group. The HCV group showed insignificant improvements in QOL (0.05-point increase per year; P ⫽ .492). Betweengroup comparisons showed differences in the rate of QOL decline between the HCV and ALD group (0.05 vs ⫺0.19 per year; P ⫽ .03) as well as the HCV and cryptogenic/other group (0.05 vs ⫺0.14 per year; P ⫽ .04). Again, these results were similar when combined with the other viral hepatitis diseases.

Figure 2. Age- and gender-adjusted longitudinal results of QOL domains by diagnosis before LTX: personal function (A and B) and general health perception (C).

At year 1 after LTX, 94%–96% of the population was satisfied with their family life and marriage, 84% were satisfied with their work or school, 28% reported a problem with interests and hobbies, and 26% had problems with their sex life. Logistic mixed models showed that the only areas to worsen over time were sex life (P ⫽ .001) and home life (P ⫽ .04).

Personal Function Measuring the ability to work and the level of physical handicap (higher score ⫽ better QOL) in these recipients, it was found that the overall adjusted mean score was 1.95 at year 1 and decreased 0.02 points per year (P ⫽ .09). In this domain, increasing age was a significant risk factor (P ⬍ .001). When QOL was compared within the 5 diagnosis groups, the ALD⫹C group again reported worse QOL at year 1 (mean, 1.41) and also had a decline (0.09-point decrease per year; P ⫽ .03) in QOL over time. Likewise, the autoimmune group did worse (0.03-point decrease per year; P ⫽ .04). Within each diagnosis group, females reported worse personal function at year 1 than males (0.31 points lower; P ⫽ .02), but there was no difference in declining QOL over time. Between-group comparisons showed that the hepatitis group’s rate of improvement was significantly different from the other 3 groups’ (autoimmune, ALD, and ALD⫹C) rate of decline (0.03-, 0.04-, and 0.09-point decrease; P ⫽ .009, P ⫽ .04, and P ⫽ .005, respectively; Table 2). When including the fulminants in the analysis, their rate of improvement was different from the ALD⫹C group (P ⫽ .02; Figure 2A and B; Appendix 3). Similar to the viral hepatitis group, persons with HCV showed slight improvements in QOL (0.05-point increase per year; P ⫽ .04). Between-group comparisons showed that the HCV group’s rate of improvement was significantly different from the other 3 groups’ (autoimmune, ALD, and ALD⫹C) rate of decline (0.03-, 0.04-, and 0.09-point decrease; P ⫽ .006, P ⫽ .03, and P ⫽ .003, respectively). When examining the individual items in this domain at year 1 after LTX, 77% reported limitations in vigorous activity, 57% were limited in the kind and amount of work they performed, and 36% were prevented from attending work/school. None of the above areas changed during the follow-up period. Thirty-five percent and 33% of the people reported difficulty walking or climbing stairs and bending/stooping, respectively. Longitudinal analysis did show that these 2 areas of personal function worsened (P ⬍ .001) over time.

General Health Perception Examining how overall health was perceived (higher score ⫽ better QOL) in the recipients, it was found that the overall adjusted mean score was 8.2 at year 1 and significantly decreased by 0.21 points per year (P ⬍ .001). When the QOL was compared among 5

LONG-TERM QOL AFTER LIVER TRANSPLANTATION

1625

diagnosis groups, the ALD⫹C group started out with the worst QOL at year 1 (mean, 7.15). All diagnosis groups deteriorated over time (0.21-point decrease per year; P ⬍ .001). No gender differences were found (Table 2). When the fulminants were included in the analysis, the ALD⫹C group still reported the worst QOL at year 1 after LTX, followed closely by the fulminant group (mean, 7.64; Figure 2C; Appendix 3). The HCV group had significant QOL decline (0.21point decrease per year; P ⬍ .001) and no difference in the rate of QOL decline compared with the other groups. The HCV group was again similar to the viral hepatitis group. When asked “over the last month how much bodily pain have you had” (none, mild, moderate, severe), 25% reported moderate to severe pain at year 1. When asked “how satisfied are you at the present time with your health,” only 10% reported being dissatisfied or completely dissatisfied at year 1. When asked “how would you rate your overall health at the present time,” 26% reported fair or poor at year 1. Over time, there was a decrease in the overall health and satisfaction with health and an increase in bodily pain (P ⬍ .001).

Other Findings When examining the 5 domains together, it was found that there were significant (P ⬍ .001) correlations between physical distress and the other domains. As physical distress became more severe, QOL in the other domains worsened. The same was found between psychological distress and social role and personal function and overall perception of health. In an attempt to more closely examine why the ALD⫹C group had lower QOL overall, living and employment status, and tobacco and alcohol use were analyzed. None of these factors was found to support any conclusions that they affected long-term QOL. Individual items in each domain were examined to see which specific items negatively affected the ALD⫹C group more than the other group. Table 3 shows the individual items by domain that the ALD⫹C group identified as being worse. Of interest were the 9 recipients that indicated feeling depressed. Of those 9 recipients, it was found that they were also the ones who more likely reported the following: muscle aches; sleeplessness/insomnia; mood swings; and problems with social situations, home, and work. When we compared our results with the nationwide health-related QOL trend in the general population reported by the Centers for Disease Control and Prevention (CDC) from 1993 to 2006,21 we found that a greater percentage of liver transplant recipients in our study reported their QOL as fair or poor (Figure 3). However, compared with people with other chronic diseases, the liver transplant recipients had similar or even better QOL.22

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

November 2010

1626

RUPPERT ET AL

GASTROENTEROLOGY Vol. 139, No. 5

Table 3. Percentages of Patients Getting Distressed, Problem, or Dissatisfaction at Year 1 After LTX by Diagnoses Before LTX (Only Those With P ⬍ .10 Are Included) Distressed, problem, or dissatisfaction at year 1 after LTX, n/N (%)a

Item

Patients with the ALD⫹C diagnosis before LTX

Patients with other diagnoses before LTXb

Pc

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

Physical distress Nausea/vomiting 4/19 (21.1) 21/352 (6.0) .03 Muscle aches 9/19 (47.4) 99/351 (28.2) .07 Backaches 9/19 (47.4) 103/352 (29.3) .09 Dark urine 3/19 (15.8) 15/352 (4.3) .06 Psychological distress Sleeplessness/ 9/19 (47.4) 97/351 (27.6) .06 insomnia Mood swings 9/19 (47.4) 95/350 (27.1) .06 Feeling depressed 9/18 (50.0) 82/352 (23.3) .02 Happiness 4/19 (21.1) 23/356 (6.5) .04 Social and role function Problem with social life 7/20 (35.0) 59/357 (16.5) .06 Problem with home life 5/18 (27.8) 44/351 (12.5) .07 Problem with vacations 8/19 (42.1) 83/358 (23.2) .09 Dissatisfaction with 7/20 (35.0) 51/332 (15.4) .03 work or school Dissatisfaction with 5/19 (26.3) 16/356 (4.5) .002 family life Personal function Problem with work 13/19 (68.4) 91/299 (30.4) ⬍.001 General health perception Dissatisfaction with 6/20 (30.0) 33/357 (9.2) .01 health ALD⫹C, alcoholic liver disease ⫹ hepatitis C; HBV, hepatitis B virus; HCV, hepatitis C virus; HDV, hepatitis D virus; LTX, liver transplantation. an/N indicates the number of patients getting distressed/total number of patients with records at year 1 after LTX. bOther diagnoses before LTX included hepatitis virus (HBV/ HBV⫹HDV/HCV), autoimmune (autoimmune hepatitis/primary biliary cirrhosis/primary sclerosing cholangitis), alcoholic liver disease, and cryptogenic/other. cP values were derived from the ␹2 test or Fisher’s exact test.

Discussion It has long been known that the QOL of liver transplant recipients, both global and domain-specific areas, improve from before to after LTX.3 An earlier analysis of the NIDDK-LTD QOL data also showed similar improvements from before to after LTX.5 However, the QOL of liver transplant recipients is more likely to be equal to other medically ill populations but not equal to healthy people.3,23 A recent systematic review of QOL continues to suggests that, although general QOL strikingly improves after LTX, recipients continue to have significant deficits, implying that even after a decade of technologic advances QOL may never approximate healthy persons.

Now, extending the time frame of the analyses, we show that many of these QOL improvements can also be largely sustained over a much longer period; however, we identified some notable findings. First, QOL in some areas gradually and consistently diminishes over time for all diagnostic groups. These declines are mostly seen in the areas of increasing reports of physical symptoms and worsening perceptions of general health. Although it may be tempting to blame these changes on normal aging, studies of QOL in cohorts of nonmedical community populations show QOL across domains of physical and psychological health as well as emotional and social role functioning do not deteriorate substantially with 10 years of aging.24 Increasing physical symptoms over time may be age related, but also they may be due to the chronically immunosuppressed state and the physicaland health-related consequences of long-term immunosuppression. That nearly 50% of recipients endorsed moderate levels of fatigue and muscle weakness by 1 year after LTX suggests that deconditioning may also be an important aspect of early recovery. Second, we found that gender was associated with perceptions of QOL after LTX; that is, females reported worse scores in physical distress and personal function than males. A recent systematic review also confirmed this observation and indicated that explanations for the phenomenon of gender-associated differences in QOL after LTX are speculative, but most of the studies have pointed to social and psychological issues.23 Third, we considered whether the use of alcohol or tobacco could account for reported declines in physical health over time. However, the use of alcohol or tobacco (only 5% of recipients) was not associated with worsening health, or was the diagnosis of ALD before LTX associated with poorer QOL outcomes. Fourth, although psychological symptoms (including symptoms of anxiety and depression) worsen albeit minimally over time, even by year one, ⱖ25% of recipients endorsed moderate feelings of nervousness, depression, sleeplessness, and mood swings. These results are not surprising, given that the rates of anxiety and depressive disorders can increase over time after LTX,25 although our QOL data were not designed to diagnose specific disorders. Fifth, when the fulminant group was compared with the others, the fulminant group’s QOL most closely resembles the ALD⫹C group, except in the domain of personal function. Because only 33% (13/40) of the fulminant group cohort met the inclusion criteria, these results may not be generalizable to the entire fulminant population. Nevertheless, these data provide preliminary results on QOL outcomes for this group. Finally, recipients who had both ALD⫹C, but not either diagnosis alone, had consistently worse QOL in all domains at the 1-year time point, and they significantly worsened in both physical functioning and physical

LONG-TERM QOL AFTER LIVER TRANSPLANTATION

1627

Figure 3. Comparison between NIDDK-LTD and the Centers for Disease Control surveys. *Data are from the 1993–2002 Behavioral Risk Factor Surveillance System (BRFSS). All respondents to the BRFSS are noninstitutionalized residents, 18 years old or older. Available at http://apps. nccd.cdc.gov/HRQOL. Accessed September 22, 2009. †Zahran et al.22

symptoms over time. These differences remained after controlling for other medical factors (age, MELD scores, etc) before LTX. In addition, these patients were not more likely than others to be tobacco or alcohol users after LTX. Certain limitations of this study should be considered. Regarding generalizability, we appreciate that there could be a selection bias, which may be due to 2 reasons. First, we carefully selected target study population with LTX to thoroughly and longitudinally examine their long-term QOL. Only 381/668 patients from the entire cohort were included. Nevertheless, the patients included did represent a true sample of the whole cohort because the distribution of patients according to diagnosis and other demographics was similar. There is the appearance that some patients did not have follow-up data, and the timing of the follow-up after transplantation was sporadic and obtained at variable intervals in different patients. Although it is true that many did not meet the stringent criteria for inclusion, in any long-term QOL study there will be some bias in the cohort selection because only those that are alive or compliant can complete the questionnaires. Second, these data precede the MELD allocation system; thus, this population may not be representative of our current liver transplant recipients in the United States. However, according to data from the United Network for Organ Sharing, the distribution of recipient race and diagnoses in our study population was similar to the racial and diagnoses breakdown for all LTXs in the United States between the time frame of our patient enrollment (1990 –1994), which was 79% white, 8% Afri-

can Americans, 9% Hispanics, and 4% others.26 Although there appears to be an overrepresentation of recipients with PBC and PSC, it is reflective of the patient population receiving transplants at that time in the United States. Furthermore, apart from the overrepresentation of cholestatic liver diseases, the distribution of other diagnoses included is similar to those receiving transplants currently. Our goal was not only to see how each diagnosis compared with each other, but also primarily to examine how each diagnosis category performed within itself. In that regard, even though the diagnostic distribution is slightly different from what is seen currently, our results provide relevant information for specific diagnostic groups. Regarding applicability, the immunosuppression regimens from the time frame of our patient enrollment (1990 –1994) may be significantly different from those currently in use. However, to examine long-term outcomes by necessity requires the examination of a cohort that received transplants many years earlier. Thus, although this may be considered a limitation, it would be always a factor in such longitudinal studies. The more effective and well-tolerated immunosuppression regimens presently being used may provide better perceptions of QOL for liver transplant recipients. In addition, on the basis of the best available nationwide surveys from the CDC, the liver transplant recipients may have similar or even better QOL than persons with coronary heart disease, diabetes, stroke, liver condition, or congestive heart failure (Figure 3).22 Overall, one of the most important findings is that with few exceptions the QOL achieved at the end of year

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

November 2010

1628

RUPPERT ET AL

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

1 typically represents the highest point of QOL and gradual losses are expected thereafter. This suggests that the first postoperative year represents a critical time for QOL recovery. It may perhaps also represent a critical time for intervention. Although most transplantation teams are focusing on the immediate surgical recovery, adjustment of immunosuppression, and restoration of basic physical functioning, greater emphasis may be required for total physical and emotional rehabilitation. Perhaps more aggressive measures should be considered or even a longitudinal program of rehabilitation to both restore and maintain a more complete and optimal physical and emotional recovery. Patients and their caregivers/ families should be advised about the need to continue working on maintaining health and functioning over the longer term. It is unknown whether assistance through some type of formal services would enhance their ultimate level of recovery and restoration of functioning during this critical first year after LTX. However, rehabilitation medicine has shown success when comprehensive, holistic rehabilitation is applied to certain populations.27 As we learn more about the details of long-term QOL after LTX as identified through this study, some consideration of how to maintain and even improve functioning broadly defined requires our clinical attention. In summary, the life expectancy of patients with endstage liver disease has been increased in large part because of advances in LTX and successful medical and surgical postoperative management. The next major area of enhancement in care after LTX is to improve the long-term QOL perhaps by preserving the QOL benefits derived early on. As we found after the peak improvement at 1 year after LTX, QOL slowly declines over time in most domains. In addition, liver transplant recipients show worse QOL after LTX than the general population, yet they have similar or even better QOL after LTX than populations with other chronic diseases. Finally, not all diagnosis groups sustain the same level of QOL, and there are gender differences as well. Hence, we suggest that more attention in future studies and clinical care should be paid to the QOL outcomes in (1) specific domains, including physical distress, social and role function, personal function, or general health perception; (2) liver transplant recipients with ALD⫹C; and (3) female liver transplant recipients. Recognizing the goal of early and comprehensive restoration of maximal QOL and continuing efforts to maintain this peak will allow liver transplant recipients and their providers to create appropriate interventions, possibly forestall further worsening in their QOL, and provide more satisfaction overall with QOL after LTX.

Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of

GASTROENTEROLOGY Vol. 139, No. 5

Gastroenterology at www.gastrojournal.org, and at doi: 10.1053/j.gastro.2010.06.043. References 1. World Health Organization (WHO). Constitution of the World Health Organization Basic Documents. Geneva, Switzerland: WHO, 1948. 2. Corn BW, Moughan J, Knisely JP, et al. Prospective evaluation of quality of life and neurocognitive effects in patients with multiple brain metastases receiving whole-brain radiotherapy with or without thalidomide on Radiation Therapy Oncology Group (RTOG) trial 0118. Int J Radiat Oncol Biol Phys 2008;71:71–78. 3. Dew MA, Switzer GE, Goycoolea JM, et al. Does transplantation produce quality of life benefits? A quantitative analysis of the literature. Transplantation 1997;64:1261–1273. 4. Bravata DM, Olkin I, Barnato AE, et al. Health-related quality of life after liver transplantation: a meta-analysis. Liver Transpl Surg 1999;5:318 –331. 5. Belle SH, Porayko MK, Hoofnagle JH, et al. Changes in quality of life after liver transplantation among adults. National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Liver Transplantation Database (LTD). Liver Transpl Surg 1997;3:93–104. 6. Burra P, De Bona M, Canova D, et al. Longitudinal prospective study on quality of life and psychological distress before and one year after liver transplantation. Acta Gastroenterol Belg 2005;68: 19 –25. 7. Gross CR, Malinchoc M, Kim WR, et al. Quality of life before and after liver transplantation for cholestatic liver disease. Hepatology 1999;29:356 –364. 8. Holzner B, Kemmler G, Kopp M, et al. Preoperative expectations and postoperative quality of life in liver transplant survivors. Arch Phys Med Rehabil 2001;82:73–79. 9. Karam V, Castaing D, Danet C, et al. Longitudinal prospective evaluation of quality of life in adult patients before and one year after liver transplantation. Liver Transpl 2003;9:703–711. 10. Kizilisik AT, Grewal HP, Shokouh-Amiri MH, et al. Impact of longterm immunosuppressive therapy on psychosocial and physical well being in liver transplant recipients. Prog Transplant 2003; 13:278 –283. 11. Kousoulas L, Neipp M, Barg-Hock H, et al. Health-related quality of life in adult transplant recipients more than 15 years after orthotopic liver transplantation. Transpl Int 2008;21:1052– 1058. 12. de Kroon L, Drent G, van den Berg AP, et al. Current health status of patients who have survived for more than 15 years after liver transplantation. Neth J Med 2007;65:252–258. 13. Lewis MB, Howdle PD. Cognitive dysfunction and health-related quality of life in long-term liver transplant survivors. Liver Transpl 2003;9:1145–1148. 14. Karam VH, Gasquet I, Delvart V, et al. Quality of life in adult survivors beyond 10 years after liver, kidney, and heart transplantation. Transplantation 2003;76:1699 –1704. 15. Wei YL, Detre KM, Everhart JE. The NIDDK liver transplantation database. Liver Transpl Surg 1997;3:10 –22. 16. Belle SH, Porayko MK. Improvement in quality of life after transplantation for recipients in the NIDDK Liver Transplantation Database. Transplant Proc 1995;27:1230 –1232. 17. SAS education. Longitudinal data analysis with discrete and continuous responses course notes. Cary, NC: SAS Institute Inc, 2008. 18. Brown H, Prescott R. Applied mixed models in medicine. 2nd ed. West Sussex: John Wiley & Sons Ltd, 2006. 19. Littell RC, Milliken GA, Stroup WW, et al. SAS for mixed models. 2nd ed. Cary, NC: SAS Institute Inc, 2006.

20. Heyting A, Tolboom JT, Essers JG. Statistical handling of dropouts in longitudinal clinical trials. Stat Med 1992;11:2043– 2061. 21. Centers for Disease Control and Prevention. National Center for Chronic Disease Prevention and Health Promotion. Health-related quality of life. Available at: http://apps.nccd.cdc.gov/ HRQOL. Accessed September 22, 2009. 22. Zahran HS, Kobau R, Moriarty DG, et al; Centers for Disease Control and Prevention. Health-related quality of life surveillance-United States, 1993–2002. MMWR Surveill Summ 2005;54:1–35. 23. Tome S, Wells JT, Said A, et al. Quality of life after liver transplantation. A systematic review. J Hepatol 2008;48:567–577. 24. Walters SJ, Munro JF, Brazier JE. Using the SF-36 with older adults: a cross-sectional community-based survey. Age Ageing 2001;30:337–343. 25. Dew MA, Kormos RL, DiMartini AF, et al. Prevalence and risk of depression and anxiety-related disorders during the first three years after heart transplantation. Psychosomatics 2001;42: 300 –313. 26. United Network of Organ Sharing (UNOS) website. Available at: http://www.OPTN.org. Accessed June 4, 2009.

LONG-TERM QOL AFTER LIVER TRANSPLANTATION

1629

27. Klonoff PS, Lamb DG, Henderson SW, et al. Outcome assessment after milieu-oriented rehabilitation: new considerations. Arch Phys Med Rehabil 1998;79:684 – 690.

Received November 25, 2009. Accepted June 4, 2010. Reprint requests Address requests for reprints to: Kristine Ruppert, DrPH, University of Pittsburgh Diabetes Institute, 4601 Baum Boulevard, Pittsburgh, Pennsylvania 15213. e-mail: [email protected]; fax: (412) 624-3775. Conflicts of interest The authors disclose no conflicts. Funding This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases N01-DK-0-2251 grant DK 55883; this work was prepared for the National Institute of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database.

CLINICAL–LIVER, PANCREAS, AND BILIARY TRACT

November 2010

1629.e1

RUPPERT ET AL

GASTROENTEROLOGY Vol. 139, No. 5

Appendix 1

November 2010

Appendix 2. Analysis Descriptive statistics (mean, median, and range) of the demographic variables, MELD scores at the point of LTX, and follow-up years were estimated initially. Mixed-effects regression analysis with the random intercept and slope17–19 was used to test whether the QOL status in each domain changed over 12 follow-up years, and if each domain over time was different by primary diagnosis for transplantation. Longitudinal linear mixed-effects modeling for repeated measures was used for 3 reasons. First, such model makes inferences by using information on all available data from the entire cohort collected at all follow-up time points. Second, the prespecified last-observation-carried-forward methods could have resulted in bias in either direction.20 Third, mixed models are powerful for analyzing correlated data. A key feature of mixed models is that, by introducing random effects in addition to fixed effects, they allow one to account for multiple source of variations, eg, within- and between-subject variations in the longitudinal study. Factors selected a priori included time (years after LTX; continuous linear covariate) and recurrence of disease (yes/no), and baseline (time of LTX) covariates: age at the initial transplantation (centered at the mean value), gen-

LONG-TERM QOL AFTER LIVER TRANSPLANTATION

1629.e2

der, primary diagnosis for transplantation [5 groups: (1) viral hepatitis (HBV, HBV/HDV, or HCV), (2) autoimmune (AIH/PBC/PSC), (3) ALD, (4) cryptogenic/other, and (5) ALD⫹C and MELD score (centered at the mean value)]. Routine model-building processes were used for models selection. Interaction terms involving gender and diagnosis with time were explored to select the final models. In addition, we separated patients with HCV (not combined with those with HBV or HBV/HDV) from the viral hepatitis group and reran the models, including 6 diagnosis groups to explore QOL status change in the HCV group. Secondary analysis included examining the effects of personal and environmental factors, such as living, employment, and smoking status as well as alcohol use. Parameter estimates were obtained by using restricted maximum likelihood and an unstructured covariance matrix to account for the variances of the intercepts and slopes, along with the covariance between the intercepts and slopes across patients. Appropriate model diagnostics and sensitivity analysis (comparing model with and without outliers) was also performed. Two-tailed P values ⬍ .05 were considered statistically significant for main and interaction effects. SAS Version 9.2 (SAS Institute, Inc, Cary, NC) was used for this analyses.

1629.e3

Appendix 3. Final Model for Longitudinal Analysis of 5 QOL Domains in 13 Fulminants and 381 Nonfulminants Physical distress (high score ⫽ bad QOL)

␤ (95% CI)

P

␤ (95% CI)

Social and role function (high score ⫽ good QOL) P

␤ (95% CI)

Personal function (high score ⫽ good QOL)

General health perception (high score ⫽ good QOL)

P

〉 (95% CI)

P

␤ (95% CI)

P

1.41 (0.83, 1.99) ⫺0.09 (⫺0.17, ⫺0.01) ⫺0.02 (⫺0.03, ⫺0.01) ⫺0.33 (⫺0.58, ⫺0.08)

⬍.001 .03 ⬍.001 .01

7.15 (6.40, 7.90) ⫺0.21 (⫺0.23, ⫺0.18) ⫺0.01 (⫺0.03, 0.004) —

⬍.001 ⬍.001 .16 —

modela

19.9 (15.1, 24.6) 1.28 (0.62, 1.94) 0.08 (⫺0.02, 0.18) 4.54 (2.32, 6.76)

⬍.001 ⬍.001 .10 ⬍.001

6.66 (4.94, 8.38) 0.04 (⫺0.01, 0.09) ⫺0.03 (⫺0.06, 0.01) —

⬍.001 .12 .11 —

13.7 (12.1, 15.3) 0.09 (⫺0.14, 0.33) ⫺0.003 (⫺0.03, 0.03) —

⬍.001 .44 .85 —

⫺3.57 (⫺9.00, 1.86) ⫺7.10 (⫺12.26, ⫺1.94) ⫺5.44 (⫺11.37, 0.49) ⫺5.61 (⫺11.02, ⫺0.20) Reference ⫺5.55 (⫺13.21, 2.12)

.20 .007 .07 .04 — .16

⫺1.23 (⫺3.18, 0.72) ⫺2.16 (⫺3.96, ⫺0.36) ⫺2.05 (⫺4.18, 0.09) ⫺1.69 (⫺3.62, 0.23) Reference ⫺0.93 (⫺3.62, 1.77)

.21 .02 .06 .08 — .50

1.72 (⫺0.07, 3.51) 2.63 (0.95, 4.30) 2.38 (0.44, 4.33) 2.98 (1.18, 4.78) Reference ⫺0.21 (⫺2.62, 2.19)

.06 .002 .02 .001 — .86

0.24 (⫺0.42, 0.91) 0.91 (0.28, 1.54) 0.51 (⫺0.21, 1.24) 0.57 (⫺0.10, 1.23) Reference 0.36 (⫺0.57, 1.29)

.47 .005 .17 .09 — .45

0.96 (0.11, 1.81) 1.18 (0.39, 1.96) 0.97 (0.04, 1.90) 0.98 (0.14, 1.82) Reference 0.49 (⫺0.69, 1.67)

⫺1.10 (⫺1.84, ⫺0.36) ⫺0.82 (⫺1.51, ⫺0.13) ⫺1.00 (⫺1.81, ⫺0.18) ⫺0.79 (⫺1.52, ⫺0.06) Reference ⫺0.90 (⫺1.87, 0.06)

.004 .02 .02 .03 — .07

— — — — — —

— — — — — —

⫺0.05 (⫺0.32, 0.21) ⫺0.17 (⫺0.41, 0.08) ⫺0.28 (⫺0.57, 0.01) ⫺0.23 (⫺0.50, 0.03) Reference ⫺0.10 (⫺0.44, 0.23)

.69 .18 .06 .08 — .55

0.13 (0.04, 0.22) 0.06 (⫺0.02, 0.15) 0.05 (⫺0.05, 0.15) 0.08 (⫺0.01, 0.17) Reference 0.14 (0.02, 0.26)

.005 .14 .29 .07 — .02

— — — — — —

— — — — — —

0.04 (⫺0.01, 0.09) 0.04 (⫺0.01, 0.09) 0.04 (⫺0.01, 0.09) 0.04 (⫺0.01, 0.09) 0.04 (⫺0.01, 0.09) 0.04 (⫺0.01, 0.09)

.12 .12 .12 .12 .12 .12

0.04 (⫺0.08, 0.16) ⫺0.08 (⫺0.14, ⫺0.01) ⫺0.19 (⫺0.36, ⫺0.02) ⫺0.14 (⫺0.26, ⫺0.02) 0.09 (⫺0.14, 0.33) ⫺0.01 (⫺0.25, 0.23)

.55 .03 .03 .02 .44 .93

0.04 (⫺0.003, 0.08) ⫺0.03 (⫺0.05, ⫺0.002) ⫺0.04 (⫺0.09, 0.02) ⫺0.01 (⫺0.05, 0.03) ⫺0.09 (⫺0.17, ⫺0.01) 0.05 (⫺0.03, 0.14)

⫺0.21 (⫺0.23, ⫺0.18) ⫺0.21 (⫺0.23, ⫺0.18) ⫺0.21 (⫺0.23, ⫺0.18) ⫺0.21 (⫺0.23, ⫺0.18) ⫺0.21 (⫺0.23, ⫺0.18) ⫺0.21 (⫺0.23, ⫺0.18)

⬍.001 ⬍.001 ⬍.001 ⬍.001 ⬍.001 ⬍.001

As indicated above

.11 .03 .04 .69 .73 .22 .34 .18 .60 .64 .06 .23 .08 .33 .55

—

—

0.18 (⫺0.15, 0.51) 0.46 (0.26, 0.65) 0.28 (⫺0.19, 0.75) 0.48 (0.17, 0.79) 1.28 (0.62, 1.94) 0.37 (⫺0.33, 1.08)

.29 ⬍.001 .24 .002 ⬍.001 .30

As indicated above

.16 .72 .19 .004 .62 .51 .88 .02 .83 .48 .02 .83 .03 .78 .07

—

—

As indicated above

.07 .04 .22 .67 .03 .23

.009 .04 .10 .005 .75 .76 .45 .14 .09 .44 .29 .09 .07 .20 .02

.03 .004 .04 .02 — .42

CI, confidence interval. of statistical significance of time, age, and primary diagnosis for transplantation, they were included in the final models, and 5 covariates (ie, gender, recurrence of disease, MELD score, and the interaction terms of gender-by-time and diagnosis-by-time) were also adjusted in the final models if they were of statistical significance. bMale patients were the reference group. cDiagnosis group 1 (DXG 1) included patients with hepatitis virus (HBV/HBV⫹HDV/HCV); diagnosis group 2 (DXG 2) included patients with autoimmune (autoimmune hepatitis/primary biliary cirrhosis/primary sclerosing cholangitis; AIH/PBC/PSC); diagnosis group 3 (DXG 3) included patients with alcoholic liver disease (ALD); and diagnosis group 4 (DXG 4) included patients with cryptogenic/other. Diagnosis group 5 (DXG 5), including patients with alcoholic liver disease plus hepatitis C (ALD⫹C), was the reference group. Diagnosis group 6 (DXG 6) included patients with fulminant liver disease. aRegardless

GASTROENTEROLOGY Vol. 139, No. 5

Final Intercept Time (per year) Age (per year) Gender (female)b Primary diagnosis for transplantationc HBV/HBV⫹HDV/HCV (DXG 1) AIH/PBC/PSC (DXG 2) ALD (DXG 3) Cryptogenic/other (DXG 4) ALD⫹C (DXG 5) Fulminants (DXG 6) Time ⫻ primary diagnosis for transplantationc Time ⫻ DXG 1 Time ⫻ DXG 2 Time ⫻ DXG 3 Time ⫻ DXG 4 Time ⫻ DXG 5 Time ⫻ DXG 6 Testing for the within-DXG annual change in QOL over time from the final model Primary diagnosis for transplantationc DXG 1 DXG 2 DXG 3 DXG 4 DXG 5 DXG 6 Comparison of the between-DXG annual change in quality of life over time from the final model Primary diagnosis for transplantationc DXG 1 vs DXG 2 DXG 1 vs DXG 3 DXG 1 vs DXG 4 DXG 1 vs DXG 5 DXG 1 vs DXG 6 DXG 2 vs DXG 3 DXG 2 vs DXG 4 DXG 2 vs DXG 5 DXG 2 vs DXG 6 DXG 3 vs DXG 4 DXG 3 vs DXG 5 DXG 3 vs DXG 6 DXG 4 vs DXG 5 DXG 4 vs DXG 6 DXG 5 vs DXG 6

RUPPERT ET AL

Factor

Psychological distress (high score ⫽ bad QOL)

November 2010

LONG-TERM QOL AFTER LIVER TRANSPLANTATION

1629.e4

Supplementary Figure 1. Consolidated Standards of Reporting Trials (CONSORT) flow diagram of selecting NIDDK-LTD study population.