A randomized, controlled study of treatment for alcohol dependence in patients awaiting liver transplantation

Authors


  • See Editorial on Page 492

  • Financial Support: Supported by National Institute on Alcohol Abuse and Alcoholism grant 1R01AA12299-01.

Abstract

Alcohol is the second most common cause of cirrhosis necessitating liver transplantation in the United States, yet rates of posttransplant drinking approach 50% and no controlled clinical trials of alcoholism treatment exist in this population. Eligible patients were randomly assigned to receive Motivational Enhancement Therapy (MET), or referral to local treatment sources (“treatment as usual” [TAU]). Addictive behavior, mood states, and general health were compared. Candor concerning alcohol use was encouraged by keeping drinking questionnaires in confidence, except in medical emergencies. Ninety-one subjects were studied; 46 received MET, 45 received TAU, 29 proceeded to transplantation (MET, n = 13; TAU, n = 16). A total of 69 subjects completed 24 weeks of observation, and 25 subjects were assessed at 96 weeks. No difference in study attendance was observed, but significantly more MET subjects attended 1 or more treatment sessions. Twenty-three subjects (25% of sample) drank after randomization but before transplant. Excluding an extreme outlier, MET drinkers had significantly fewer drinks per drinking days than TAU drinkers. Neither treatment plan resulted in significant variances in measures of psychosocial health. In conclusion, although MET afforded no significant benefit over TAU for mood or general health outcomes, this study provides some degree of support for MET to limit the quantity and frequency of pretransplant alcohol consumption among liver transplant candidates with alcohol dependence. However, because of the limited number of study subjects, these data must be interpreted cautiously. Further research to validate our findings or to identify better methods to identify and intervene with patients at risk of pretransplant and posttransplant drinking should continue. Liver Transpl 17:539–547, 2011. © 2011 AASLD.

Alcohol-related cirrhosis is the second most common indication for liver transplantation in the United States (based on Organ Procurement and Transplantation Network data as of September 24, 2010). Up to 50% of patients with alcoholism who receive a liver transplant return to alcohol use in the first 5 postoperative years.1, 2 The reported rates of alcohol use probably represent underreporting, particularly in studies using retrospective design.1, 2 Significantly, despite difficulties in getting accurate data on alcohol relapse, drinking in an abusive fashion after transplantation can lead to acute injury to the allograft and is an important cause of death or graft loss.3, 4 In addition, addictive drinking carries all the general health and psychosocial tolls observed in the nontransplant population. For these reasons, it is widely agreed that patients with alcoholism under consideration for liver transplantation should undergo evaluation by an addictive disorders specialist and should abstain from alcohol.5

The first approach to ensuring abstinence after transplantation has been to try to select only those patients likely to maintain abstinence.6 A recent meta-analysis by Dew et al. identified 3 pretransplant factors associated with alcoholic relapse after transplantation: lack of social support, a family history of alcoholism, and less than 6 months abstinence from alcohol.7 The predictive utility of a 6-month abstinence interval, although controversial, is widely used.8 However, rates of relapse after transplantation have mostly been reported in the setting of a 6-month abstinence requirement using retrospective data collection without objective measures of alcohol consumption.9 Furthermore, mandating a 6-month period of pretransplant abstinence is weakly associated with posttransplant abstinence and will result in the deaths of a significant number of individuals who will never drink again.10

A second approach to reducing alcoholic relapse is to treat alcoholism in subjects at risk of relapse or after relapse. To date, the few pilot studies attempting to prophylactically modify drinking behavior after transplantation found that waiting until after transplantation introduced unforeseen obstructions to addiction treatment, whereas treatment before transplantation may be efficacious.11, 12

Consequently, in this study, we attempted to ameliorate the risk of posttransplantation drinking by initiating alcoholism therapy before transplantation. We selected Motivational Enhancement Therapy (MET), an established alcoholism therapy we adapted to address the needs of this population. Our MET protocol was closely based on the Project MATCH (Matching Alcoholism Treatment to Client Heterogeneity) MET manual.13 MET includes personalized feedback of assessment results and a counseling style intended to “produce rapid, internally motivated change by mobilizing the patient's own change resources.” To meet the needs of our population, we increased the number of sessions from 4 to 7 and the duration of treatment from 3 to 6 months, expanded potential targets of intervention to include relevant health-related behaviors for patients with cirrhosis, and included case management to link patients to additional resources.

The effect of MET was compared to referral for local intensive outpatient therapy and Alcoholics Anonymous (AA) or Narcotics Anonymous (NA) meetings, referred to here as “treatment as usual” (TAU).

PATIENTS AND METHODS

This study took place at the Hospital of the University of Pennsylvania (UPenn), Philadelphia, PA, and the University of Wisconsin School of Medicine and Public Health (UW), Madison, WI. Subjects undergoing evaluation for liver transplantation were considered suitable for this study if they met the following criteria: the primary cause of their liver disease was alcohol-related, they had taken at least 1 alcoholic drink within 2 years prior to the initial transplant evaluation, and they understood and agreed to the informed consent document. Subjects who completed formal alcoholism treatment (inpatient, outpatient, or partial hospitalization) in the prior 6 months were excluded. Written informed consent was obtained from each subject. The study conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the institutional review committees of both study centers.

All potential liver transplant patients were screened for alcohol use disorders by the transplant social worker (UPenn) or hepatologist (UW). Subjects with a provisional diagnosis of an alcohol use disorder were evaluated by an addiction specialist via modified SCID (Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition [DSM-IV]) for the diagnosis of Alcohol Dependence versus Alcohol Abuse, assessed for the need for alcoholism treatment, and screened for any other Axis I psychiatric disorder.14 Prospective subjects meeting criteria for study inclusion were invited to participate or were referred to community intensive outpatient programs without research participation by either the principal investigator (R.M.W. at UPenn) or co-principal investigator (M.R.L. at UW). All study patients were listed upon presentation to the selection committee prior to randomization. Compliance with individualized addiction management was monitored within the transplant programs. In the course of the study, no patients were removed from the list for failure to attend addiction therapy that had been recommended by the transplant program.

Prospective subjects were told that declining participation in the study would not affect their transplant candidacy nor did study participation afford special consideration for transplantation. Respectively, the first and final subjects at UPenn were randomized in June 2000 and January 2004 (n = 83), and at UW on August 2002 and January 2004 (n = 8).

Personnel without connection to either center's transplant programs conducted MET sessions and follow-up visits. At baseline and every study visit, a Folstein Mini Mental Status Exam and hepatic encephalopathy rating scale was completed. Subjects with Mini Mental Status Exams of 17 or less, or hepatic encephalopathy levels of greater than stage 1 were rescheduled until improved, if they remained within the window of time stipulated for the visit, or excluded if they were too ill to participate while still in the prerandomization phase. Candid responses were encouraged by reassuring subjects that addictive behaviors would remain confidential except in life-threatening emergencies such as excessive alcohol or illicit drug use or severe psychiatric illness, upon which the principal investigator at the other study site would determine whether the transplant team required notification.

Study Design

A randomized controlled design with follow-up by a research assistant with no clinical responsibilities occurred every 3 months until transplant, and at 1 month, 6 months, and 1 year after transplantation. All subjects were placed on the waiting list prior to randomization. After completing baseline measures, subjects were randomized to receive either 7 individual sessions over 6 months of MET with encouragement to attend AA meetings and case management (MET/CM), each lasting 50 minutes, or to receive TAU, entailing referral to community AA and standard intensive outpatient therapy. Addiction therapists administering MET had Master's degrees and were trained and monitored for treatment adherence by one of the authors (D.V.H.), who graded the integrity of randomly selected audiotapes of MET/CM sessions using a standardized measure.15 Therapists were found not to substantially differ in their approaches or skill in delivering MET. No patients in TAU received MET from community resources. Two years of pretransplant follow-up data were analyzed because less than 43% of the original sample were available for analysis beyond that point (Table 1).

Table 1. Number of Subjects Participating in Either Study Group at Intervals After Randomization
 Week
 01224487296
  1. Two-year follow-up rates (No significant difference: χ2 (1) = 0.08, P = 0.78).

MET (n)463735321813
TAU (n)453734272012
Total N (%)91 (100%)74 (81%)69 (76%)59 (65%)38 (42%)25 (27%)

The primary aim of the study was to evaluate whether patients with alcohol-related cirrhosis who were randomized to receive MET/CM had lower rates of drinking compared to subjects receiving TAU. Secondary endpoints were changes in psychiatric problems as well as improved general health and motivation for alcoholism treatment. Alcohol use was assessed at each study visit by: self-report via time-line follow back16; a validated assessment that uses a calendar with key dates as anchor points to provide daily drinking estimates over a specified time period; collateral interview with a family member or significant other using the Addiction Severity Index (ASI) a semi-structured interview, 17 recent use of alcohol by breathalyzer (Alco-Sensor III; Intoximeter, Inc., St. Louis, MO). Measures of drinking included number of drinks, number of drinking days, and number of drinks per drinking day. One standard drink was defined as 13 g ethanol, which was derived from equating 1.5 oz. distilled spirits = 8.0 oz. wine = 12 oz. beer.

We evaluated moods and anxiety symptoms using the Beck Depression Inventory18 and the Beck Anxiety Inventory19 at baseline and every 3 months from the point of randomization. Because our prior work suggested an association between motivation for alcoholism treatment and treatment acceptance, we used the Stages of Change Readiness and Treatment Eagerness Scale (SOCRATES) model20 to measure our subjects' level of acceptance of their alcohol problem and willingness to make changes toward attaining and maintaining alcohol abstinence. The SOCRATES model discriminates among those who lacked recognition of alcoholism as a problem to those who endorsed “Ambivalence,” “Recognition,” or “Taking steps” toward treatment. Additionally, we report quality of life (QOL) using the Medical Outcomes Study Short Form-12 (MOS SF-12),21 scoring General Health only.

Statistics

In the absence of prior studies of this type, the study was planned around expected observable differences that might have clinical impact. Thus, for sample size calculations, a Type I error of 0.01 was taken. The sample size of 100 per group was thought attainable and large enough to detect a difference of 0.7 standard deviations (SD) (0.7 effect size) in means between the 2 groups (eg, a 0.7 SD of the percent days drinking or of the Beck Depression Inventory), a difference of 17% in proportions, and a difference of 4 months in survival times, with 80% power, while accounting for alpha = 0.01. The projected sample size of N = 200 was not accrued, but the actual sample of 91 was large enough to detect a difference of 0.6 SD in comparisons of means, a 28% difference in proportions (eg, proportion who used drugs: 50% versus 78%), and a 6-month difference in median survival times, with 80% power at a 5% significance level. When numbers of subjects are too small for accurate statistical comparison, descriptive data only are provided. The proportions for drinking in each group were compared using a chi-square test. Measures of quantity and frequency of drinking among subjects who drank were based on negative binomial regression models, incorporating offsets to account for differential follow-up time. We used linear mixed effects to compare the 2 groups on repeated mood responses, including ASI composite scores, Beck Depression Beck Anxiety scores, over the first 2 years on the wait list. Explanatory variables in these models included the baseline version of the response, a binary indicator for intervention group, terms for linear and quadratic time effects, and terms for group by time interactions. We included random intercepts and random slopes in all models.

Abbreviations:

AA, Alcoholics Anonymous; ASI, Addiction Severity Index; BDI, Beck Depression Inventory; CI, confidence interval; CTP, Child-Turcotte-Pugh; df, degree of freedom; IOP, intensive outpatient program; IQR, interquartile range; MELD, Model for End-Stage Liver Disease; MET, motivational enhancement therapy; MET/CM, motivational enhancement therapy/case management; NA, Narcotics Anonymous; QOL, quality of life; SCID, structured clinical interview for DSM-IV; SD, standard deviation; SE, standard error; SOCRATES, Stages of Change Readiness and Treatment Eagerness Scale; TAU, treatment as usual; Upenn, University of Pennsylvania; UW, University of Wisconsin.

RESULTS

The demographics of the 2 treatment groups are shown in Table 2. In an effort to increase recruitment, UW was added as a second site; however, limited time to recruit before the grant terminated led to the discrepancy between the 2 sites. There were no significant differences in age, sex, racial origin, or severity of liver injury as estimated by Model for End-Stage Liver Disease (MELD) score.

Table 2. Baseline Demographics
 METTAUP Value
  • *

    CTP values only; MET (n = 22), TAU (n = 26). In this table, continuous variables are summarized as (median [IQR]). The P values for binary variables are based on chi-square tests, whereas those for continuous variables are based on Wilcoxon tests.

Subjects4645
Sex (M/F)39/737/80.74
Age (years)50.5 (42.0-56.0)48.0 (43.0-52.0)0.50
Race: % white; % other85%; 15%78%; 22%0.39
Years of education12.0 (11.0-12.0)12.0 (12.0-13.0)0.17
Employed (full or part time) n (%)23 (50%)28 (62.2%)0.35
MELD scores14.5 (12.0-17.0)14.0 (11.0-16.0)0.21
CTP*8.5 (8.0-11.0)8.0 (7.0-10.0)0.19
Drinking within 30 days of randomization, n (%)5.0 (10.87%)4.0 (8.89%)0.95
Months since most recent drink9.0 (6.0-15.0)9.0 (4.0-18.0)0.91
Lifetime years alcohol abuse20.5 (10.0-28.0)15.0 (10.0-20.0)0.16
Received alcohol treatment (non-detox)18 (39.13%)22 (48.89%)0.34
Received drug treatment (non-detox)8 (17.39%)10 (22.22%)0.56
SOCRATES – Ambivalence10.0 (7.0-15.0)10.0 (4.0-15.0)0.93
SOCRATES – Recognition17.0 (13.0-28.0)19.0 (11.5-28.0)0.26
SOCRATES – Taking Steps36.0 (30.0-39.0)36.0 (30.5-40.0)0.39
Beck Depression Inventory10.0 (7.0-15.0)11.0 (4.0-16.0)0.89
Beck Anxiety Inventory7.5 (3.0-12.0)10.0 (4.0-15.0)0.11
SF-124.0 (3.0-5.0)4.0 (3.0-4.0)0.09

Employment patterns in the 2 groups were similar. In the MET group, 24 (52%) people were employed fulltime, 5 (11%) [5/46 = 0.1087] were employed part-time, whereas the remaining 17 (37%) [17/46 = 0.3696] were retired or disabled. In the TAU group, there were 18 (40%) in full-time employment, 5 (11%) in part-time employment, and 22 (49%) were retired or disabled. There was no significant difference between the groups on rates of employment versus retirement/disability (Table 2). Income for the past month came from some combination of employment income, pensions, some public assistance, and from other family members. The distribution of incomes for the previous 30 days were similar for the 2 groups (TAU: median = $924, interquartile range [IQR] = [$513-$1690]; MET: median = $1024, IQR = [$550-$2528]; Wilcoxon test P value = 0.34)

The first drinking variable in Table 2 gives the presence or absence of drinking in the 30 days prior to randomization. In the MET group, 2 people self-reported drinking in the prior 30 days. Of these, 1 drank to intoxication on 27 days of the 30, whereas the other drank lightly for 1 day. For the TAU group, 3 people self-reported drinking. One drank to intoxication on 10 days, and less heavily on 5 other days, whereas the other 2 drank lightly on 1 or 2 days. The other prerandomization drinkers were identified by blood alcohol content level or by collateral report, and we have no quantity/frequency data for them. As part of the inclusion criteria for the study, subjects had to have used alcohol within the 2 years before the study. Table 2 shows that the average times to most recent drink were less than 1 year for each group. Lifetime alcohol abuse is defined as the years in which alcohol has been used to intoxication. Overall, the median number of years of abuse was approximately 17 years, with no significant difference between the 2 groups. Approximately 45% of the sample had received formal treatment, not including detoxification visits, for alcohol problems. A smaller number had also received formal treatment for drug abuse problems.

Twenty-nine subjects underwent transplantation (MET, n = 13; TAU, n = 16). No association was found between study group and transplant (chi-square [1 degree of freedom {df}] = 0.56, P = 0.46). Sixteen subjects died (MET, n = 9; TAU, n = 7). No association was found between study group and death (chi-square [1 df] = 0.25, P = 0.62). Of the subjects who did not die, 18 of 37 (49%) dropped out of the study in the MET group and 24 of 38 (63%) dropped out of the TAU group. Although the TAU proportion is higher, the difference was not statistically significant (chi-square [1 df] = 1.07, P = 0.30).

Follow-up indicated that 69 patients completed 24 weeks of observation, and 25 subjects completed 96 weeks (Table 1). There was no difference in research study visit attendance among MET and TAU subjects.

Treatment Adherence

MET (Intervention) Services Received

Data at weeks 12 and 24 were available for 36 of 46 (78%) MET subjects' sessions, of whom 30 (83%) attended at least 1 session whereas 18 (50%) attended 4-7 sessions (Table 3). The average number of completed MET sessions was 3.83 of 7. Failure to attend >4 MET sessions were because of illness (n = 6), transplantation (n = 2), transportation problems (n = 3), drop-out (n = 3), or “other” (n = 4). Six MET subjects failed to attend any MET sessions because they were either too ill (n = 4), received transplantation (n = 1), or had transportation problems (n = 1).

Table 3. Treatment Engagement
 METn/Total (%)TAUn/Total (%)P Value
  • *

    Data available: n = 36/46 (78%) (MET) weeks 12 and 24.

  • **

    Data available: n = 40/45 (89%) (TAU) weeks 12 and 24.

Attended at least 1 session MET or intensive outpatient program30/36* (83%)16/40** (40%)chi-square (1 df) = 9.36, P = 0.00
Attended at least 1 AA meeting11/34 (32%)20/40 (50%)chi-square (1 df) = 2.35, P = 0.13

Therapist Adherence to MET Therapy Protocol

A total of 20% of audiotapes were reviewed for monitoring therapists' adherence to the MET protocol. If a therapist was below threshold, he or she received feedback and the next tape from the same patient was reviewed in addition to the usual 20% of available tapes. In addition to the formal ratings of adherence, approximately 35% of recordings were reviewed in the course of clinical supervision. Of 34 tapes that were formally rated, 1 was rated below threshold (2.9% of all formally rated tapes) and 3 met the overall threshold but were marked as having some scales below the threshold.

(Nonintervention) Services Received

Data at weeks 12 and 24 were available for 40 of 45 (89%) TAU subjects. There were no inpatient visits or hospitalizations for alcohol treatment reported for either group, so the services received comprised outpatient visits or AA/NA meetings attended. Of the 40 subjects in the TAU group, 11 (28%) reported neither outpatient visits nor attendance at AA/NA meetings; 9 (23%) reported outpatient visits only, 13 (33%) reported attendance at AA/NA meetings only, whereas 7 (18%) reported receiving outpatient visits and attendance at AA/NA meetings. For the 36 people in the MET group, 21 (58.33%) reported neither outpatient visits (apart from the MET intervention), nor attendance at AA/NA meetings; 2 (6%) reported outpatient visits only, (apart from the MET intervention), 10 (28%) reported attendance at AA/NA meetings only, whereas 1 (3%) reported receiving outpatient visits (apart from the MET intervention) and attendance at AA/NA meetings. There was no difference in the proportions of TAU versus MET reporting attendance at some AA/NA meetings (chi-square [1 df] = 2.35, P = 0.13). Among people who reported at least 1 outpatient visit, the mean number of visits per month was 3.94 (SD = 4.73) for the 3 people in the MET group, compared to 2.21 (SD = 2.28) per month for the 16 people in the TAU group. This difference was not statistically significant (P = 0.65). Among people who reported attending at least 1 AA/NA meeting, the mean number of meetings per month was 5.49 (SD = 6.58) for the 11 people in the MET group, compared to 5.53 (SD = 8.21) per month for the 20 people in the TAU group. This difference was not statistically significant (P = 0.85). Thus, more people in the TAU group attended AA/NA meetings, but the rate per month was the same for the 2 groups. Lifetime AA meeting attendance was not measured.

Drinking Behavior

Drinking was measured by patient self-report (ASI, Timeline Follow-Back interview), collateral informant (MATCH), or breathalyzer. Whereas 9 (9.9%) subjects reported consumption of some alcohol in the 6 months prior to the baseline visit (prerandomization), 23 subjects (25% of entire sample) consumed alcohol after randomization and before transplantation. The prevalence of drinking was virtually identical in both groups: 12 of 46 MET subjects drank (26%), whereas 11 of 45 subjects in TAU drank (24%) (chi-square [1 df] = 0.00, P = 0.95). In MET, almost half the episodes of drinking were self-reported (n = 14) and the remaining episodes were identified by collateral informant and/or breathalyzer (n = 15). In TAU, approximately 80% of all drinking was identified by self-report.

We used logistic regression models to investigate whether baseline characteristics predicted drinking. The variables chosen to predict drinking were treatment group, presence of prerandomization drinking, race (white/other), sex, age, Beck Depression score, Total score from Beck Anxiety Inventory, SF-12 scale, years of education, and the Ambivalence, Recognition, and Taking Steps scales from the SOCRATES. When considered individually, race, drinking in the 30 days prior to randomization, the Beck Depression score, and the SOCRATES scales had significant effects at the 5% significance level. Being nonwhite, drinking prior to randomization, and having higher scores on the mood, anxiety, and SOCRATES scales were all associated with increased risk for postrandomization drinking. Under different model selection approaches (forward, backward, and stepwise selection with P = 0.05 threshold for inclusion), we found that the best model comprised the Beck Depression score and the SOCRATES ambivalence scale, with all other variables being excluded from the model. Higher scores on the Beck Depression scale (odds ratio = 1.10, 95% confidence interval [CI] = [1.02, 1.18], P = 0.01) or on the SOCRATES Ambivalence scale (odds ratio = 1.22, 95% CI = [1.08, 1.38], P = 0.001) were associated with higher risk of drinking.

Six of the 9 people who drank in the 30 days prior to baseline also drank after randomization. However, the prerandomization drinking variable did not get selected into the best model because it was correlated with the Ambivalence score. The Ambivalence score was a highly significant predictor of postrandomization drinking and was included in the best model on the second step of the model selection. Once the Ambivalence score was in the model, inclusion of the drinking variable did not lead to a significant improvement of fit (P = 0.09) and therefore was not included.

The 23 subjects drank a total of 64 episodes over 2 years of pretransplant observation, and quantity and frequency data were available for n = 17: 8 MET subjects and 9 from TAU. The median within-subjects values and IQR are reported in Table 4. The median number of drinks per drinking day was 3.75 (IQR = 2.5-4.8) in MET compared to 4.3 (IQR = 3.4-8.0) for TAU. Similarly, median total drinking days over two years of observation were similar in both groups. The heaviest drinkers, defined as >5 drinks/drinking day, comprised 3 MET subjects who reported drinking at 6 study visits compared to 7 TAU subjects who reported drinking at 13 study visits. Eight subjects reported drinking on more than 4 study visits over the entire two years of observation.

Table 4. Self-Reported Quantity and Frequency Pretransplant, Postrandomization Drinking (n = 17)
  MET Group With OutlierMET Group Without Outlier
 TAU (N = 9)MET (N = 8)P Value (N = 17)MET (N = 7)P Value (N = 16)
  1. Entries are median (IQR) for maximum 108 weeks follow-up (allowing 1 month window past 2-year follow-up point). P values based on Negative Binomial regression models, n = 16 after exclusion of extreme outlier in MET.

Weeks of follow-up96 (72-108)60 (24-96)0.1648 (24-96)0.16
Total number of drinks59 (8.0-136.0)7.0 (3.5-9.0)0.697.0 (2-8)0.003
Total number of drinking days7.0 (2.0-11.0)2.0 (1.0-2.0)0.612.0 (1.0-2.0)0.004
Drinks per drinking day4.3 (3.4-8.0)3.75 (2.5-4.8)0.0253.5 (2.0-5.0)0.036

We compared the groups on the responses in Table 4 using Negative Binomial regression. In Table 4 and in the analyses, we account for 1 MET subject who reported drinking more than 875 standard drinks in the 2-year period. No other member of the MET group reported more than 10. Although the data from this subject have very little effect on the median and IQR summaries given in Table 4, they have a very strong influence on the overall group comparisons in the regression models. We felt the number of reported drinks was unusually large, and we therefore excluded that subject from our main analyses. There was no significant difference in the length of follow-up time for the 2 groups (P = 0.15). However, because of the small sample sizes and low power for the comparison of follow-up times, we used an offset of log (weeks of follow-up) to adjust the comparisons of total number of drinks, and total number of drinking days, for number of weeks of follow-up. The MET drinkers had significantly fewer total number of drinks than the TAU group (beta = −2.51, 95% CI = [−4.14, −0.87], P = 0.003), and significantly fewer drinking days than TAU (beta = −2.62, 95% CI = [−4.40, −0.83], P = 0.004). We found similar results, with smaller P values, when we did not include an adjustment for follow-up time. Finally, MET drinkers had lower drinks per drinking day than TAU drinkers (beta = −0.70, 95% CI = [−1.36, −0.04], P = 0.035).

If we include the MET subject with more than 875 drinks reported, we see similar trends of lower drinking among MET drinkers relative to TAU drinkers, but the comparisons are not significant for total number of drinks (P = 0.69) or total number of drinking days (P = 0.61), and are essentially the same as before for drinks per drinking day (P = 0.025).

Of the 29 people who underwent transplantation, we were able to ascertain that 7 had used alcohol after the transplant, and 6 of these subjects reported quantity/frequency data. Four of these subjects were in the MET group and 2 were in the TAU group, but the numbers of episodes and subjects are too small for formal statistical analyses. Of the 6 subjects, 3 had also used alcohol prior to their transplant, whereas none had reported drinking in the 30 days prior to randomization. In total, the 6 subjects reported posttransplant drinking on 9 separate occasions. For these occasions, the number of drinking days reported ranged from 1 day to approximately 2 months, and the reported drinks per drinking day ranged from 1-6 standard drinks.

Illicit Drug Use

Baseline assessments of prescribed and illicit drug use were obtained by technician administered ASI for the previous 30 days and urine toxicology. Only illicit drug use will be reported here. At baseline, 9 subjects used marijuana and 1 used heroin. For post-baseline drug use, assessments included the ASI, urine toxicology, and Timeline Follow-Back interview. Illicit drug use consisted of a total of 7 patients who used marijuana, n = 4 in the MET group and n = 3 in the TAU group.

Psychosocial Measures

Beck's Anxiety and Depression Inventories

The groups had similar baseline scores on the mood responses. Overall, Beck Anxiety Inventory scores did not change significantly over time (F (1,54) = 0.75, P = 0.39). On the Beck Depression Inventory, MET showed lower total scores over the 2 years than TAU (b = −4.2, standard error [SE] = 1.86, P = 0.02). The difference emerged in the first 12 weeks, but after that TAU showed a slightly higher rate decline than MET for total Beck Depression Inventory, although the differences were not statistically significant. Overall, mood responses did not change much over the first 2 years on the wait list.

MOS-SF 12 Quality of Life

QOL scores were similar in both groups and did not change significantly over time (F[1,52] = 0.70, P = 0.41).

Stages of Change (SOCRATES) Scores

Although MET and TAU had similar Ambivalence scores at study entry (b = −1.21, SE = 1.21, P = 0.32), MET had a small but significant decrease (1.2 units every 6 months) in Ambivalence scores across time compared to TAU (F[1,49] = 4.29, P = 0.04). Recognition scores were similar in MET and TAU at baseline (b = −1.87, SE = 2.00, P = 0.35), although MET showed a significant decrease (of approximately 2.1 units every 6 months) in the Recognition scores across time (F[1,48] = 5.87, P = 0.02). Regarding Taking Steps scores, MET and TAU were similar at entry (b = 0.93, SE = 2.29, P = 0.68), and did not show a significant change across time (F[1,49] = 2.69, P = 0.11).

DISCUSSION

Our previous studies showed that liver transplant recipients with alcoholism were reluctant to undertake alcoholism treatment2, 12 because of being preoccupied with posttransplant management, endorsing no alcohol craving, and feeling apprehensive about the potential for liver damage from naltrexone, a medication approved for the treatment of alcoholism. These findings led us to test psychotherapies to reduce posttransplant drinking and improve general health prior to transplantation in liver transplant candidates suffering from alcoholism. We chose MET as the psychological intervention. MET is an established alcoholism treatment regimen we adapted to address the specific needs of this population, and compared it to our standard practice referral for local intensive outpatient therapy and attendance at AA meetings. We called the latter option TAU.

This study was initially designed to evaluate the posttransplant effects of MET administered in the pretransplant period; however, the classification system for liver allocation changed while the study was in progress. When we started, donor livers were allocated to patients according to the recipient's severity of liver disease as estimated by the Child-Turcotte-Pugh (CTP) score. Because the CTP has just 3 graduations (A, B, and C), time on the waiting list was used to discriminate between candidates in the same CTP class. Thus, placement on the transplant list relatively early was encouraged in order to accumulate time on the list. In 2002, the allocation process changed to the MELD score, a more sophisticated estimation of urgency. Thus, many of our subjects with low MELD scores did not progress to transplantation and fewer received transplantation at study termination than we had planned. Consequently, this report emphasized pretransplant measures of functioning.

Surprisingly, we discovered more subjects drank, and more serious drinking took place, in the pretransplant period than was previously recognized. This is disappointing because all study subjects underwent extensive psychosocial evaluation and were counseled to avoid all alcohol. The observation of drinking by subjects awaiting transplantation is probably enhanced both by the use of multiple methods to measure drinking and by the provisions we instituted to encourage candor about drinking on the part of subjects. This corroborates our belief that patients are more likely to admit to drinking if they do not risk permanent removal from the waiting list.22

In the present study, 25% of subjects waiting or in preparation for liver transplantation relapsed to some alcohol use. This proportion is similar to alcohol use observed prospectively in patients with alcohol-related cirrhosis after portal decompressive therapy following variceal hemorrhage.23 These data suggest a powerful urge to drink exists in some patients with alcohol-related cirrhosis at risk of life-threatening decompensation. Apparently, neither variceal hemorrhage, distal splenorenal shunt, transjugular intrahepatic portal systematic shunt, or risk of losing access to liver transplantation are sufficient, of themselves, to induce and sustain abstinence. Furthermore, if the compulsion to drink is due to an internal drive, these observations cast doubt on previous data in which we found a lack of alcohol craving in liver transplant candidates.2 These data suggest that more than 1 population of these patients exists in relation to their risk of relapse.

Although similar prevalences of drinking before transplant were found in both the MET and TAU treatment groups, our study suggests that MET may be effective in reducing the quantity and frequency of pretransplant alcohol consumption. We also found that higher scores on the Beck Depression Inventory or on the SOCRATES Ambivalence scale were associated with higher risk of drinking. However, given the limited amount of quantity/frequency data in our analysis, these findings must be interpreted cautiously. Data on the Beck scales of depression and anxiety, QOL, and stages of change (SOCRATES), showed no significant differences between MET and TAU groups over the study period. We chose to study MET because it was the intervention in the largest-ever study of alcoholism treatment that targeted the prominent lack of motivation for alcoholism treatment in our population.24 Careful measurement of adherence to MET in this study showed that nonadherence was explained by the severity of illness and social impediments common in this population.

Limitations

Conclusions concerning the results of our study must be tempered by its limitations. One limitation is our inability to provide detailed information concerning the lack of treatment attendance in the TAU group. We wanted the behavior of the TAU group to be as close as possible to what naturally happened on the transplant service. This meant that we allowed the transplant teams' Social Workers and nurse coordinators to monitor adherence to treatment and alcohol abstinence. Unfortunately, this approach did not yield significantly detailed information about adherence to allow us to investigate why subjects did not receive the amount of treatment that we planned for them to receive.

Another limitation of our study is that half of the MET group attended fewer than 4 of the allotted 7 MET sessions. The most common reason reported for this lack of treatment adherence in our study was that patients were too ill to attend. Although it could be argued that low treatment adherence makes it impossible to attribute any positive outcomes of the study to MET, it could also be argued that because MET was originally designed to consist of 4 sessions in total and the remaining half of our study patients attended 4-7 sessions, MET may have conferred some benefit to them.

In summary, it appears that alcohol-dependent liver transplant candidates struggle with total abstinence before and after transplantation. Unfortunately, our capacity to identify patients at greatest risk of relapse remains poor. Similarly, treatment interventions to effect robust and lasting sobriety in this population are limited. Although this study does provide some degree of support for MET to limit the amount and frequency of pretransplant alcohol consumption among liver transplant candidates with alcohol use disorders, the limitations of our study did not allow for us to glean any information on the potential benefits of MET in the posttransplant period. It is also important to recognize that the severity of illness in this population of patients is so substantial, that rigorously studying them is quite difficult. Future treatment research in alcohol-dependent liver transplant patients will need to take these difficulties into consideration by designing studies that accommodate patients with this degree of debility. Perhaps the use of newer treatment technologies such as telemedicine could improve rates of adherence to treatment and research visits, thereby providing more comprehensive information to more thoroughly analyze and interpret.

Acknowledgements

We thank Maria De Piano, Kathy Massoth, and Aaron Shields for their role as study coordinators.

Ancillary