Objectives: Substance abuse is a risk factor for nonadherence and graft failure after orthotopic liver transplant. This study aimed to evaluate the ability of an internally developed tool, the Rochester Relapse Risk Scale, to predict substance relapse in liver transplant candidates.
Materials and Methods: This single-center, retrospective, observational study included adult patients evaluated for orthotopic liver transplant using the Rochester Relapse Risk Scale. Primary outcome was rate of substance relapse, as measured by the risk scale, which stratified patients into relapse risk levels based on the number of factors present.
Results: In total, 303 patients (71.6% men, 90.4% White, median age of 55 years [interquartile range, 49-60 y]) were included. Median follow-up time was 212 days (interquartile range, 73-661 d). Seventy-four patients (24.4%) relapsed at 127 days (interquartile range, 55-461 d) after evaluation, with 60.8% who relapsed within 6 months. Relapse rates correlated with assigned risk level, with 8.3% relapsing at low, 19.0% at low-moderate, 25.3% at moderate, 33.8% at moderate-high, and 40.0% at high risk. High-risk cohorts had significantly shorter median time to relapse versus low-risk cohorts (104 vs 154 days; P = .001).
Conclusions: Assignment of relapse risk level according to the Rochester Relapse Risk Scale aligned with rates of relapse. Additional studies are needed to refine the tool, assess inter-rater reliability, and confirm findings in prospective, multicenter studies.
Key words : Alcohol, Evaluation, Psychosocial, Selection, Wait list
Orthotopic liver transplant (OLT) is a lifesaving surgery for patients with chronic liver disease, acute liver failure, and certain hepatic malignancies.1 Two of the top 3 indications for liver transplant, liver disease due to alcohol and hepatitis C virus, are associated with substance use, and 30% of such liver transplant candidates have been found to be actively using substances at the time of evaluation.2 Patients with historical or active substance abuse are at particularly high risk for medical nonadherence, potentially resulting in failure of the transplanted organ.3-6 Because of this risk, in the context of scarce organ supply, it is important to carefully screen patients with histories of substance abuse for the likelihood of future relapse.
“Minimum criteria” for liver transplant listing were published in 1997, including at least 6 months of abstinence.4 Although this 6-month criterion is commonly used, it is not well validated and in turn may result in long wait times and risk of death for patients at lower risk for relapse.7,8 In addition, there are multiple other factors that are known to influence relapse. Although there are tools available to assess relapse risk, none are well validated or consistently used within the transplant community.9-12
To the best of our knowledge, there has only been one study in which a model was created using individual risk factors to assess risk of relapse. This European study of 387 liver transplant recipients demonstrated that less than 6 months of abstinence, presence of a psychiatric comorbidity, and High-Risk Alcohol Relapse (HRAR) score greater than 3 were independently associated with relapse to harmful drinking.13 In patients with no risk factors, there was a 5% alcohol relapse rate, whereas patients with all 3 risk factors had a relapse rate of 100%.13 Several other studies have shown that a multidisciplinary approach, including rigorous psychosocial assessment, is the most effective method to identify risk of relapse in patients posttransplant.9-12,14
Given the relative lack of unanimity about relapse risk factors and the absence of a commonly used assessment tool, the intent of this study was to evaluate the ability of the Rochester Relapse Risk Scale (R3S) to predict substance relapse in a liver transplant population, to refine the predictive ability of the tool as necessary, and to advance the process of establishing agreed upon relapse risk factors.
Materials and Methods
This was a retrospective, observational, cohort study of adult patients evaluated for OLT between January 1, 2012, and December 31, 2016, at an 886-bed tertiary academic medical center. The Institutional Review Board of the center approved the study. Patients with a history of substance abuse who were evaluated by transplant psychiatry using the R3S were included. Patients who died during the index hospitalization and who were not transplant candidates for medical reasons and who had no available follow-up data were excluded.
Rochester Relapse Risk Scale
The R3S is a tool that was developed at the University of Rochester Medical Center to assess risk of substance relapse at transplant evaluation. Its use was implemented in 2012 and included risk factors for relapse that had commonly appeared in substance abuse literature. Patients identified by social work with current substance abuse, less than 6 months of sobriety, or concerning features of prior substance abuse history (eg, multiple substances of abuse, recent use amidst a history of problematic use, a significant prior history or abuse of one or more substances) were referred to transplant psychiatry and assessed using the R3S tool. Between 2012 and 2015, the R3S included 9 criteria (R3S 1.0); however, in 2015, it was updated to include 2 additional components (R3S 2.0).15-17 Definitions for each risk factor are included in Table 1. The application of the HRAR was done only for those patients with problematic alcohol use. Assignment of risk level was based on the number of risk factors present (low: ≤2 factors, low-moderate: 3-4 factors, moderate: 5-6 factors, moderate-high: 7-8 factors, high: ≥9 factors). Recommendations on psychosocial candidacy for transplant were guided by risk stratification.
Patient follow-up and substance relapse
All patients were assessed until the first documented relapse, date of last follow-up, death, or through September 30, 2017. Substance relapse was defined as any amount of use, whether large or small, problematic or not, as reported by the patient themselves, a family member, friend, significant other, or member of the health care team or by laboratory confirmation of substance use. Complete abstinence from all substances of abuse (alcohol, marijuana, and illicit drugs) was the explicit expectation for all patients. Screening for alcohol or other substances of abuse was done randomly with patient understanding that failure to complete lab work when requested would be considered the same as alcohol or substance use. Urine samples were used for substance use screening.
Any patient who received a risk designation of moderate or higher after R3S assessment and who had substance use in the prior 6 months was required to complete a chemical dependency program and establish at least 6 months of abstinence prior to transplant listing consideration. Patients deemed low or low-moderate risk, with substance use in the prior 6 months, were required to engage in at least a relapse prevention program and, if possible, establish a minimum of 6 months of abstinence prior to consideration for listing. Given the community-based nature of substance abuse treatment, and the fact that our transplant program receives regional referrals, patients were referred locally for treatment and were not required to select a program affiliated with our institution. Our team ensures patients have made an appointment with a treatment program and completes necessary paperwork to allow for sharing of information between the program and our institution. Once a patient is enrolled in a treatment program, the transplant team receives feedback on the patient’s progress as needed, but this does not occur at a standard frequency. In some instances, patients who were assessed by the R3S to be at or below moderate risk of relapse, but who had substance use in the prior 6 months, would experience significant hepatic decompensation prior to completing the required relapse prevention or chemical dependency program. These patients would be discussed by the team and listed for transplant on a case-by-case basis.
Duration of abstinence prior to consideration for transplant was defined as the most recent of either the time from last reported or last documented substance use until the date of initial psychiatric evaluation.
The primary outcome was rate of substance relapse after assessment by R3S, stratified according to assigned level of relapse risk (low, low-moderate, moderate, moderate-high, and high). Secondary outcomes included a comparison between R3S-assigned risk categories and actual relapse rates in those who underwent transplant, the presence of independent associations for the variables used in the R3S, relapse rates stratified according to duration of abstinence, time to relapse, and survival after relapse.
Participants were identified from preexisting, intradepartmental databases (Organ Tracking Transplant Record, internal liver transplant mixed-use database). Data for this study were extracted from records in the internal mixed-use database on patients with liver failure and supplemented with information not routinely collected. Sources included electronic hospital and transplant records. Study data were collected and managed using Research Electronic Data Capture (REDCap), a secure, Web-based software platform hosted at The University of Rochester.18
Simple descriptive statistics were used to summarize patient demographics and clinical measures. Univariate analyses were used to evaluate differences between patients with and without substance relapse using the Pearson chi-square or Fisher exact test for categorical variables and Mann-Whitney U test for continuous variables. Variables with a P value < .20 (total cohort) or P < .10 (R3S 2.0 only) in the univariate analysis were considered for inclusion in the multivariate model. Multivariate analysis was performed using the Cox proportional hazards regression model. Time to relapse by assigned risk level was assessed using a Kaplan-Meier curve with log-rank pairwise comparisons. Tests were 2-tailed, and P < .05 represented statistical significance. All analyses were performed using SPSS software version 25.
Table 2 shows the baseline characteristic of the 303 included patients. Most patients were White (90.4%) and men (71.6%), with median age of 55 years (interquartile range [IQR], 49-60 y). Median follow-up from date of psychiatric evaluation was 212 days (IQR, 73-661 d). Forty-two patients (13.9%) had less than 30 days of follow-up, and 109 patients (36.0%) died during the study period. Alcoholic cirrhosis and hepatitis C virus were the most common causes of end-stage liver disease. Median Model for End Stage Liver Disease sodium (MELD-Na) score was 17 (IQR, 11-24) at evaluation, and 30.7% of patients had hepatocellular carcinoma (HCC). Ninety-two percent of psychiatric evaluations occurred in the outpatient setting, and past or current substances of abuse included alcohol (97.0%), marijuana (49.5%), and cocaine (31.7%). Sixty-seven patients (22.1%) were listed by the usual process, with team consideration of medical/surgical and psychosocial data, and 31 patients (10.2%) underwent liver transplant during study follow-up. One hundred and twenty-one patients (39.9%) were not listed for transplant due to concerns identified during their psychiatric evaluation, 109 (90.1%) of which were recommended to complete a substance abuse treatment program in order to be reconsidered for transplant evaluation. The percentage of patients excluded from transplant listing due to psychiatric concerns increased in concordance with increasing risk assignment, with no patients in the low-risk group being excluded for this reason. The median follow-up duration for these patients was 133 days (IQR, 44-476 d), and 45 patients (37.2%) died during the study period.
The most common risk factors present at evaluation were defensiveness (65.0%), history of substance abuse in a first-degree relative (61.4%), and <6 months of abstinence (56.1%) (Table 3). The prevalence of each risk factor was similar across both the R3S 1.0 and 2.0 cohorts, with the exception of dependence diagnosis (36.2% vs 56.5%; P = .001). Frequencies of risk category assignment were as follows: 4.0% were low, 39.9% were low-moderate, 30.0% were moderate, 24.4% were moderate-high, and 1.7% were high risk of relapse (Table 4).
During the study period, 74 patients (24.4%) relapsed (Figure 1). Relapse increased across risk category assignment (8.3% of those at low risk, 19.0% at low-moderate, 25.3% at moderate, 33.8% at moderate-high, and 40.0% at high). The median time to relapse was 127 days (IQR, 55-461 d) after psychiatric evaluation (Figure 2). Median time to relapse was shorter in high-risk cohorts versus low-risk cohorts, at 104 days (IQR, 55-279 d) and 154 days (IQR, 86-599 d), respectively (P = .001), when risk categories were combined (low included low and low-moderate, moderate included moderate, and high included moderate-high and high). With regard to relapse events, 74% occurred within 12 months of psychiatric evaluation and 36.5% within the first 3 months. Rates of relapse were inversely correlated with duration of abstinence: 38.6% relapsed at <3 months of abstinence, 19.3% relapsed at 3 to 6 months, 18.2% relapse at 6 to 12 months, and 10% relapse at >12 months (Figure 3). When a relapse occurred, the most common substances used were alcohol (59.5%), marijuana (31.1%), cocaine (10.8%), and heroin (2.7%), with relapse most often reported by the patient (53.4%) or family member (13.7%). Seven patients (9.5%) relapsed with use of multiple substances, but most patients (52.7%) used alcohol alone. Laboratory findings to identify or support relapse (blood or urinary alcohol levels, ethyl glucuronide, or positive urine drug screen) were available in 58.1% of patients.
Of 31 study patients who underwent transplant, 2 patients (6.5%) had a relapse event after transplant. Both patients were White men, 53 and 65 years old, and had a diagnosis of alcohol-related cirrhosis. They were evaluated using the R3S model 1.0 and were both categorized as having moderate risk of relapse, with the recommendation to complete a chemical dependency program. One patient underwent urgent transplant due to deteriorating clinical status prior to involvement in a chemical dependency program. He had been sober 13 days at the time of transplant and attended a chemical dependency program after transplant; however, it is unclear whether he completed this program. He relapsed 15 months after transplant as determined by a urine drug screen positive for amphetamines. No other relapses have been noted, and liver function tests have remained stable. The second patient was sober 26 months and completed a chemical dependency program prior to transplant, as recommended by the transplant team. Thirteen months after transplant, intermittent use of alcohol was reported by both the patient and his wife over the previous 6 months. The patient thereafter recommitted to sobriety, and no other relapses have been noted. Liver function tests have remained within normal limits.
Univariate analyses demonstrated that less than 6 months of abstinence at time of OLT evaluation, psychiatric comorbidity, dependence diagnosis, and more than 1 substance of abuse were associated with relapse (Table 5). Multivariate Cox proportional hazards regression confirmed an independent association of less than 6 months of abstinence at the time of OLT evaluation (adjusted hazard ratio [HR] of 2.57; 95% CI, 1.53-4.33) (Table 6). For R3S models 1.0 and 2.0, multivariate Cox proportional hazards regression demonstrated independent associations between less than 6 months of abstinence at time of OLT evaluation for both cohorts and HRAR score of >2 in the R3S 2.0 cohort only (Table 7).
Rate of death among patients who relapsed was 32.4%, at a median of 170 days (IQR, 34.5-385.5 d) after relapse. There was no trend seen between rate of death and assigned risk category. Two deaths occurred in those who received OLT, one due to asystole during surgery and one due to ischemic injury secondary to postoperative bleeding.
This study reported relapse outcomes in patients with end-stage liver disease assessed at the time of transplant evaluation by a tool that incorporates multiple potential risk factors. Prior studies have investigated individual risk factors, but few have evaluated the ability of those factors to predict relapse when considered together.12,13 Our results demonstrated that, with increasing risk level assignment, relapse rates increased proportionally.
The design of this study was unique in that patients were assessed at transplant evaluation and relapse events were captured both before and after transplant. A report by De Gottardi and colleagues described similar predictors of relapse, including less than 6 months of abstinence, presence of psychiatric comorbidity, and HRAR score greater than 3, but assessment was conducted only in patients who received a transplant.13 Our tool is reflective of a larger population and explores relapse in patients involved in the transplant program whether undergoing transplant or not, rather than focusing solely on those who underwent transplant.
Despite recommendations to maintain sobriety, 24.4% of patients relapsed after initial assessment, in comparison to rates of 14.6%, 11.9%, and 25% described in previous reports.6,13,19 The rate of relapse captured in our study is higher than those previously reported; however, we believe this is a result of inclusion of a higher risk population, all substances of abuse, and a broader definition of relapse events (including any reported substance use during follow-up).6,13 Similar to DiMartini and colleagues, we chose to consider multiple sources of input to identify relapse events as we believe this provides a more thorough assessment and increases the likelihood of detecting relapse, especially because denial and deception can occur in this patient population.17 Furthermore, our decision not to use a specific test or laboratory assay is reflective of an environment typically faced by transplant programs, rendering these findings potentially more relevant in day-to-day transplant practice.
Similar to the report by De Gottardi and colleagues,13 our multivariate analysis demonstrated an independent association of less than 6 months abstinence with increased risk of relapse (HR of 2.57; 95% CI, 1.53-4.33; P ≤ .001). Duration of sobriety has been identified as a predictor of alcohol relapse in several publications; however, the appropriate duration of sobriety has not been thoroughly investigated and an inappropriately long duration of required abstinence could unnecessarily deny or delay transplant for patients who are at lower risk of relapse.5,7,8,12,19-23 This issue is increasingly more pertinent as alcoholic-related cirrhosis as an indication for liver transplant becomes more prevalent. Interestingly, most of the relapse events in our study occurred in the first 3 months after evaluation (36.5%). There was no difference in substance relapse between patients who abstained from substance use for 3 to 6 months and those who abstained for 6 to 12 months (19.3% vs 18.2%; HR of 1.22; P = .629); however, we did find a difference in those abstinent for <3 months versus 6 to 12 months (38.6% vs 18.2%; HR of 2.72; P < .01). These results are similar to a study published by Hartl and colleagues that found a higher rate of relapse in patients with <3 months of abstinence at time of transplant compared with those with a pretransplant abstinence duration of >3 months (63.6% vs 11.4%; P = .012).22 When considered together, these studies suggest that there is more work to be done to discern the optimal duration of abstinence prior to OLT. Moreover, it is possible that the duration of abstinence must be considered in the context of other factors.
In addition, an independent association between HRAR score of greater than 2 (R3S 2.0 cohort) and risk of relapse was identified, similar to that described by De Gottardi and colleagues.13 Our threshold HRAR score of greater than 2 was lower than that used previously.13 We chose this threshold (rather than >3) to increase the sensitivity of the assessment. It is important to note that, although the R3S is designed to assess risk of relapse to any substance, HRAR score is specific to alcohol only, and was therefore only applied to patients with current or prior alcohol use in our study. Of note, this occurred in a large percentage of our patients; however, the R3S-assigned risk level in these patients may have been higher due to this additional risk factor being assessed.
In our study, psychiatric comorbidity and defensiveness did not predict substance relapse, which differs from some prior reports.10,11,13 This may be an example of how these terms have been applied differently across studies due to a lack of uniformity in definitions and application. Our study defined psychiatric comorbidity as a psychiatric issue that impacted health care (eg, inability to attend visits due to panic disorder with agoraphobia, personality disorders impeding the ability to form relationships with health care providers, schizophrenia with chronic symptoms despite optimized treatment, and eating disorders with symptoms that impact medical health such as electrolyte disturbances or inaccurate medication levels due to purging) rather than merely based on the presence of psychotropic medications or a psychiatric diagnosis. We did not use a scale or additional tools for this item. With this definition, our study identified that 9.0% of patients had a “psychiatric comorbidity” in R3S model 1.0 and 14.8% in R3S model 2.0. Given the high concordance with substance use disorder and psychiatric pathology, our incidence is lower than that described in other studies, but we believe this is a more idealistic reflection of candidates with concerning features of their psychiatric history. Similarly, the definition of defensiveness can vary substantially. In this study, defensiveness was strictly evaluated as it applied to substance use or abuse. These factors, and others, did not prove significant in predicting relapse in this population. For this reason, their removal as potential predictive factors will be considered in subsequent versions of the tool.
This study focused on relapse while patients were involved with the transplant program, whether undergoing transplant or not. How our findings correlate to solely the posttransplant population, which may arguably be a more meaningful metric for transplant programs deciding upon candidacy, is not clear. Although the overall rate of relapse was 24.4%, our posttransplant relapse rate was 6.5%, which is lower than previous reports.6,13,23 One potential reason could be that, in our cohort, despite the 5-year scope, only 31 patients underwent transplant procedures and, of those, only 2 relapsed. This low number of patients transplanted reflects our center’s selection criteria (ie, requirement for a substance abuse treatment program in all patients at moderate or high risk and absence of high-risk patients who met listing criteria), with 61.3% of transplanted patients in the low risk for relapse category. Because of these small numbers, we were unable to specifically assess the ability of the R3S to predict relapse in posttransplant patients.
An assessment of substance relapse and attributable mortality was limited in this study by the small number of relapse cases. In total, 24 patients (32.4%) who relapsed died during follow-up. Conclusions regarding whether a return to substance use was associated with mortality could not be made given the incomplete data.
The design of this study includes several strengths and limitations that merit consideration. One strength is the inclusion of various risk factors that appear in the general substance abuse literature to create a composite score and risk categorization, which were investigated within a population of patients being evaluated for transplant. By examining multiple factors, our approach is similar to the Alcohol Relapse Risk Assessment scoring system but different in the fact that we used relapse risk factors explored in the general substance abuse population.24 Our study findings related to relevant risk factors are consistent with those of De Gottardi and colleagues,13 which, as such, could help to narrow the breadth of possibilities of substance abuse relapse predictors in transplant populations.
There are several limitations to our tool that are important to consider. First, we defined relapse as any use, regardless of amount or whether use was problematic. This definition was chosen given the retrospective nature of the study and the inability to comprehensively assess the significance of relapse for all participants as a result of limited documentation. For this reason, our study described more relapse events than some prior investigations. However, the lack of routine, objective assessment of relapse could have affected our primary endpoint in an opposite manner. Our approach, however, is reflective of what we believe to be a real-world clinical practice setting. In addition, all patients were assessed by a single psychiatrist who assigned risk category. Some risk factors assessed in the R3S are subjective and rely on information from patients or family members; therefore, responses may be impacted by recall bias and/or patient deception. These aspects limit the external validity and more widespread utilization of the tool; however, assessment by the same psychiatrist led to consistent evaluation and application of the criteria. With regard to our assessment of the criterion of more than 1 substance of abuse, we did not distinguish between recent and remote use, nor did we quantify use. All prior use was considered similarly, which is recognized as a limitation of our study. In addition, the percent of patients deemed not to be candidates for transplant was highest in the moderate-high-risk and high-risk groups. These patients were likely lost to follow-up in many cases, as they were no longer receiving care from our institution; therefore, our data related to their relapse were less robust. Furthermore, the median duration of follow-up was significantly greater for patients assigned to the low-risk and low-moderate-risk categories compared with those assigned to the moderate-high-risk and high-risk categories. Given this deficiency in data, it is possible that the true relapse rates in our higher risk categories are higher than those reported. Monitoring for relapse in patients evaluated for transplant was naturalistic and did not follow a standardized frequency or format. Because of this and the fact that urine drug screening was the more commonly used modality, it is possible that cases of relapse may have been missed that would have been otherwise identified by more frequent monitoring or use of serum laboratory detection. Finally, although adherence to clinic visits, blood work, and medications may be an endpoint of interest relative to relapse, we did not include this in either the R3S model 1.0 or model 2.0.
Substance relapse risk levels assigned using the comprehensive R3S correlated with rates of substance relapse in a population of liver transplant candidates. Of the 11 risk factors included in this assessment, less than 6 months of abstinence was independently associated with relapse in both the R3S model 1.0 and model 2.0, whereas HRAR score greater than 2 was independently associated with relapse in the R3S model 2.0 only. Further studies are needed to assess accuracy of the R3S tool when administered by other practitioners and to confirm these findings in prospective, multicenter studies.
For our program, implementation of the R3S has facilitated the candidate selection process. It is often difficult to determine candidacy for transplant, which relies on an accurate prediction of the likelihood of relapse, while simultaneously advocating for the patient and the scarcity of donated organs. Although there is no means of ascertaining the actuality of future relapse, this tool, which has shown higher rates of relapse in those estimated to be at higher relapse risk, helps ease difficult team decisions about candidacy by reducing the degree of uncertainty in relapse prediction. For those patients estimated to be at higher relapse risk, patients are not listed unless or until there is demonstration of active abstinence management and a period of ongoing documented sobriety, as described earlier in this paper. The R3S also assists the team in deciding how to proceed with candidates who have shorter periods of abstinence and have HCC (which allows accrual of MELD points) or who otherwise require urgent transplant. Although length of pretransplant abstinence may be a relapse risk predictor, it is not clear whether it is a predictor across all potential transplant candidates; that is, there may be some candidates presenting for urgent transplant without prolonged abstinence who may in fact undergo transplant and then maintain posttransplant sobriety. It seems likely this is the case. Mathurin and colleagues, for example, reported on early transplant for severe alcoholic hepatitis in 26 patients who had less than 6 months of abstinence; of these, only 2 patients relapsed over 2 years posttransplant.26 By using the R3S, our team has proceeded toward transplant in patients who have shorter abstinence and lower estimated relapse risk levels. Finally, the R3S has helped our team advocate for the integrity of its selection process by applying the same parameters across candidates and by reducing emotionally driven candidacy decisions.
Given the results of this study, our center will consider implementation of a simplified R3S. In general, future research is necessary to continue refining these factors and their interrelationships and to establish an agreed upon and consistently applied set, which will help augment existing information and narrow the scope of this endeavor. Refinement of relapse risk assessment is becoming increasingly important as transplant for alcohol-related cirrhosis becomes more prevalent and the field considers transplant of patients with acute alcoholic hepatitis.
Volume : 19
Issue : 9
Pages : 919 - 927
DOI : 10.6002/ect.2021.0034
From the 1Department of Pharmacy, the 2Division of Transplantation, the 4Department of Surgery, and the 5Department of Psychiatry, University of Rochester Medical Center, Rochester; and the 3Wegmans School of Pharmacy, St. John Fisher College, Rochester, New York, USA
Acknowledgements: The authors have not received any funding or grants in support of the presented research or for the preparation of this work and have no declarations of potential conflicts of interest.
Corresponding author: Samantha Leistman, Department of Pharmacy, University of Rochester Medical Center, 601 Elmwood Avenue, Box 638, Rochester, NY 14642, USA
Phone: +1 585 602 6123
Table 1. Risk Factors and Corresponding Definitions Included in the Rochester Relapse Risk Scale
Table 2. Patient Demographics
Table 3. Presence of Individual Risk Factors
Table 4. Risk Categorization After Assessment With the Rochester Relapse Risk Scale
Figure 1. Relapse Rate According to Assigned Risk Level
Figure 2. Time to Relapse by Risk Assignment Category
Figure 3. Rate of Relapse According to Duration of Abstinence
Table 5. Univariate Analysis of Factors Associated With Relapse
Table 6. Multivariate Predictors of Relapse (Total Cohort)*
Table 7. Multivariate Predictors of Relapse by Rochester Relapse Risk Scale Model