Readiness to Change Predicts Drinking: Findings from 12-Month Follow-Up of Alcohol Use Disorder Outpatients

Alcohol and Alcoholism, 217, 52(1) 65 71 doi:.93/alcalc/agw47 Advance Access Publication Date: 28 July 216 Article Article to Change Predicts Drinking: Findings from 12- Follow-Up of Alcohol Use Disorder Outpatients Jacques Gaume*, Nicolas Bertholet, and Jean-Bernard Daeppen Alcohol Treatment Center, Department of Community Health and Medicine, Lausanne University Hospital, Avenue de Beaumont 21 bis, Batiment P2, 11 Lausanne, Switzerland *Corresponding author: Lausanne University Hospital, Alcohol Treatment Center, Avenue de Beaumont 21 bis, Batiment P2, 11 Lausanne, Switzerland. Tel.: +41-21-314-73-53; Fax: +41-21-314-5-62; E-mail: jacques.gaume@chuv.ch Received 4 April 216; Revised June 216; Accepted 25 June 216 Abstract Aim: To test whether measures of readiness to change (RTC) re-assessed every 3 months had predictive value for change in alcohol use over 12 months in a sample of adult outpatients with alcohol use disorder (AUD). Methods: Of the case, 78 outpatients were followed monthly over one year and averaged 9. interviews each (total observations = 74). Alcohol abstinence days and heavy drinking days were assessed monthly using a 3-day timeline follow-back procedure. RTC was assessed using 3 readiness rulers (importance, readiness, and confidence to change, measured on a visual analog scale). The effect of RTC on alcohol use over time was tested every 3 months using negative binomial generalized estimating equations (GEE), controlling for gender, age, baseline alcohol dependence severity and AUD treatment status (ongoing vs. ceased). Results: GEE models showed highly significant effects of readiness and confidence to change on respective alcohol outcomes. Effects of importance to change were weaker. Conclusion: As hypothesized, higher RTC scores were associated with improved alcohol use outcomes in this longitudinal study. The strongest effects were for confidence to change. Finding significant predictive validity prospectively is consistent with a theoretical view of RTC as a dynamic construct. Further research might clarify how AUD treatment could actually elicit or increase RTC. INTRODUCTION Motivation to change is a complex concept that covers many diverse aspects of the process of intentional behavior change (DiClemente et al., 24). The trans-theoretical model of change proposes that change occurs in a series of stages (Prochaska and Velicer, 1997). to change (RTC) is more generic in scope than the stages concept, and typically indicates a willingness or openness to engage in a particular process or to adopt a particular behavior (DiClemente et al., 24). It has been conceptualized by some as a combination of patients perceived importance of the problem and confidence in their ability to change (Rollnick, 1998; Miller and Rollnick, 22). In a recent meta-analysis encompassing 39 studies and 8238 patients, Norcross et al. (211) evaluated the ability of stages of change and related RTC measures to predict psychotherapy outcomes. There was a mean significant medium effect size of d =.46 (95% confidence interval (CI):.35.58). Among the 39 studies, 14 tested addiction outcomes using baseline RTC, resulting in a mean effect of d =.37 (95% CI =.23.52). However, some research suggests that RTC may operate differently in various populations, with some sub-dimensions or measures of RTC playing a more prominent role than others in predicting future changes (e.g. Williams et al., 27; Bertholet et al., 29; Kaysen et al., 29). For example, in individuals with unhealthy alcohol use, evidence has been mixed regarding the predictive validity of RTC. There were positive associations of RTC and alcohol use (higher RTC related to more drinking, as in Palfai et al., 22; The Author 216. Medical Council on Alcohol and Oxford University Press. All rights reserved 65

66 Alcohol and Alcoholism, 217, Vol. 52, No. 1 Shealy et al., 27 and Harris et al., 28; importance to change as in Williams et al., 27). On the other hand, there were also negative associations (higher RTC related to decreased drinking, as in Carey et al., 27; Bertholet et al., 212 and Merrill et al., 215; confidence to change as in Williams et al., 27). Finally, no associations were found by Borsari et al. (29) and Barnett et al. (2). To elucidate the nature of the association of RTC with drinking behavior, Merrill et al. (215) postulated that it was important to use longitudinal designs to establish temporality among the variables, since one s stage or level of readiness is theoretically dynamic (Prochaska and Velicer, 1997) and drinking patterns among individuals also tend to fluctuate a great deal daily and weekly (e.g. Maggs et al., 211; Daeppen et al., 213; Voogt et al., 214). However, such designs are rare. In a sample of female college students, Kaysen et al. (29) examined whether RTC predicted subsequent weekly drinking behavior over 11 weeks. They found that during weeks when students reported more RTC relative to their average levels, they also reported intentions to drink less in the future, as well as actual reductions in drinks per week during the following week. Merrill et al. (215) tested the associations of RTC with weekly alcohol use and consequences in a sample of 96 college students. As hypothesized, higher RTC in any given week (relative to one s average RTC level) was negatively associated with alcohol use and consequences the following week. To our knowledge, such longitudinal analyses have not been conducted in a sample of alcohol treatment patients. rulers, or ladders, have been developed as RTC measures to provide a quick overview of client RTC (e.g. Biener and Abrams, 1991; Sobell et al., 1996; Miller and Rollnick, 22). These rulers are used both as clinical tools (to elicit RTC and facilitate behavior change) and as research assessment methods (to measure baseline RTC, shifts in RTC as an outcome, or as a mediator of behavior change). Three dimensions can be measured: RTC, of changing and in ability to change (Rollnick, 1998). In the present study, we tested whether these three dimensions, assessed every 3 months, had predictive value for changes in alcohol use over a 12-month follow-up of outpatients with an alcohol use disorder (AUD). Our hypothesis was that higher scores on the three readiness rulers would be related to improved alcohol use outcomes. METHODS Sample Data for the present study were derived from the CONTROL project (Daeppen et al., 213), a prospective single-center, observational cohort study. All consecutive new patients assessed at the Alcohol Treatment Center (ATC) outpatient clinic, inpatient clinic and liaison consultations at Lausanne University Hospital, Lausanne, Switzerland, were offered participation in the research. Patients were excluded if they did not meet criteria for alcohol dependence based on the Mini International Neuropsychiatric Interview questionnaire (Lecrubier et al., 1997), were under age 18, were confused or delirious, did not speak French or had no contact address. Participants provided written informed consent. For the present study we limited the sample to patients of the ATC outpatient clinic. This excluded patients from the inpatient program or those referred to the ATC following hospitalization at the general hospital/ed department (liaison consultations). This was done to ensure a more homogeneous sample that is more representative of common frontline AUD practice, and to realize higher rates of follow-up. The CONTROL study was approved by the Ethics committee for clinical research at Lausanne University Medical School (protocol 127/9). Irrespective of their participation in the cohort study, all new patients assessed at the ATC are offered standard care provided by physicians, nurses and/or psychologists, as required under Switzerland s universal healthcare. This includes an initial evaluation consisting of alcohol use history, severity and repercussions, and a diagnostic assessment including AUD, internal medicine and psychiatric diseases. Individuals are then invited to participate in the standard care program, and to choose their own objective pertaining to their drinking. They are offered a combination of medical treatment, motivational interviewing, relapse prevention measures and/or group support sessions. The actual combination, frequency and length of treatment are adjusted according to the needs of each patient. Measures Face-to-face interviews at baseline, 3-, 6- and 9-month follow-ups were conducted by a trained psychologist. In addition, short telephone assessments were conducted at follow-up months 1 and 2, 4 and 5, 7 and 8 and and 11. The following measures were used in the present study. Alcohol outcomes were assessed each month from baseline through the 11 follow-ups using a 3-day Timeline Follow-Back (TLFB; Sobell and Sobell, 1995). Based on TLFB data, we derived (a) the number of abstinence days (ABD) over the 3-day period, and (b) the number of heavy drinking days (HDD), defined as days with six standard drinks or more, i.e. 6 grams of pure alcohol or more. Independent variables were three RTC measures on visual analog scales, also known as readiness rulers and labeled, and (e.g. Miller and Rollnick, 213). They were assessed during face-to-face interviews at baseline, 3, 6 and 9 months. Because some participants might have changed their alcohol use by follow-up, questions were worded to include importance/readiness/confidence to change drinking or maintain abstinence. Exact wording can be translated from French as To what extent is it important for you today to change your drinking habits/maintain abstinence? means not important at all and extremely important ; Suppose now that you ve decided to stop drinking, what confidence do you have in your ability to actually do it? or What confidence do you have in your ability to maintain abstinence? means not at all confident and means you are certain to be able to do it ; Finally, to what extent are you ready to change your drinking habits/maintain abstinence? means not ready at all and means absolutely ready. Additionally, we used four potentially confounding variables as control variables in our models. These variables were chosen a priori, based on their recognized relationship with alcohol use in the literature. The first three, gender, age and alcohol dependence severity, were measured at baseline. Gender and age were asked with standard closed questions, while alcohol dependence severity used the Alcohol Dependence Scale (ADS; Skinner and Horn, 1984). The fourth variable, AUD treatment status, was measured at each trimester (i.e. baseline and follow-ups at 3, 6 and 9 months) in face-to-face interviews. AUD treatment was a dichotomous variable defined as ongoing treatment vs. ceased treatment by termination or dropout.

Alcohol and Alcoholism, 217, Vol. 52, No. 1 67 Table 1. Sample description and follow-up measures Baseline 1-month 2-month 3-month 4-month 5-month 6-month 7-month 8-month 9-month -month 11-month N 78 75 73 66 61 59 61 55 55 61 58 55 % followed-up 96.2 93.6 84.6 78.2 75.6 78.2 7.5 7.5 78.2 74.4 7.5 Gender 52 (66.7%) Male a Age b 48.9 (11.5) Severity 14.8 (7.2) ADS score [ 47] b In AUD treatment a 78 (%) 47 (71.2%) 31 (5.8%) 21 (34.4%) # Abstinence days b 14.4 (11.9) 2.8 (12.1) 23. (11.) 2.3 (12.2) 19.3 (12.6) 19.4 (13.1) 19.7 (12.6) 19.7 (12.4) 2.4 (12.6) 19. (12.7) 2.5 (11.9) 21. (11.7) # HDD b. (12.) 4.3 (9.) 2.9 (7.5) 3.8 (7.5) 4.4 (9.2) 4.3 (9.9) 4.2 (8.5) 4.4 (9.3) 4.2 (9.2) 3.8 (8.5) 2.5 (6.5) 3.7 (8.2) [ ] b 8.6 (2.5) 7.8 (3.) 7.9 (3.3) 7.2 (3.6) [ ] b 8.5 (2.8) 7.3 (3.3) 7.4 (3.7) 6.2 (3.9) [ ] b 7.3 (2.4) 7.4 (2.5) 7.5 (2.6) 7.2 (2.6) a Categorical variables: descriptive statistics are N (%). b Continuous variables: descriptive statistics are means (standard deviation) ADS, Alcohol Dependence Scale; AUD, alcohol use disorder. Statistical analysis We first calculated standard descriptive statistics to describe the sample. Then, we tested if there were any attrition patterns by using correlations between continuous baseline measures and the number of follow-up observations, and a t-test to compare the number of follow-up observations by gender. We used negative binomial generalized estimating equations (GEE) to analyze the effect of RTC on alcohol use over time. GEE (see e.g. Liang and Zeger, 1986) offers an extension of regression analysis to the case of correlated observations over time (here repeated alcohol measures). It also allows for specification of the assumed correlation structure (set here as exchangeable, i.e. responses for individual patient are assumed to be equally correlated) as well as the appropriate link functions (set here as log) to handle dependent variables with different distributions (set here as negative binomial, due to over-dispersed count data). We first computed six models (3 readiness rulers by two alcohol outcomes), all adjusted for gender, age, baseline alcohol dependence severity and AUD treatment status, to predict alcohol use over the next 3 days (TLFB measured at next month assessment). For this analysis, we used all data from the trimester assessments (baseline, 3-, 6- and 9-month) for independent and confounding variables, and the next month (i.e. 1-, 4-, 7- and -month) for the outcome variables. For this analysis we had 78 subjects and a total of 248 observations (3.2 observations per subject on average). To take advantage of the entire set of data (i.e. TLFB measured each month), we repeated the latter analysis while imputing the readiness rulers values collected at each trimester interview to the following 2 months interviews (e.g. if a given patient had a value of 3 for confidence to change at the 3-month interview, we imputed a 3 on this scale at 4- and 5-month follow-up). Because we imputed readiness rulers values using values collected the first month of each trimester, rulers were used to predict the next month alcohol use as in the first analysis, but also the next 2 months (e.g. confidence to change at 3-month was used to predict alcohol use at 4-month, but also alcohol use at 5-month and 6-month). For this analysis we had 78 subjects and a total of 74 observations (9. observations per subject on average). Since distributions were largely left-skewed in our outpatient population, in order to test linearity and to better describe the effects of each readiness ruler on alcohol outcomes, we trichotomized the 3 rulers by recoding each into three categories: low ( 5), medium (6 9) and high (). This categorization allowed sub-categories of comparable sizes (respectively, N = 57, 64 and 146 for importance; N = 73, 63, 131 for readiness; and N = 79, 116, 72 for confidence). We first tested these variables in the same GEE models as described above, and also generated two-way graphs to depict the effects of the three measures on both outcomes over time. As the 3 rulers were importantly correlated (importance-readiness: r =.83; confidence-readiness: r =.52; importance-confidence: r =.39), we did not compute a multiple regression model. RESULTS Among the 79 patients included, one patient had no follow-up data and thus no valid outcome data, resulting in N = 78 patients in all analyses. Follow-up rates varied between 96.2% and 7.5% over the 12 months of research (Table 1, first two rows). The sample was comprised of 66.7% males having a mean age (standard deviation) of 48.9 (11.5) years. Average ADS scores were 14.8 (7.2),

68 Alcohol and Alcoholism, 217, Vol. 52, No. 1 corresponding to intermediate levels of alcohol dependence and a high likelihood of psychological problems related to drinking, indicating outpatient care. Over the course of the study, the proportion of participants receiving AUD treatment was % (recruited at baseline while attending the ATC outpatient clinic), 71.2% at 3-months, 5.8% at 6-months and 34.4% at 9-months. The average number of ABD at baseline was 14.4 (11.9), which increased to 2.8 (12.1) at 1-month and to 23. (11.) at 2-months, then varied from 19. (12.7) to 21. (11.7). The average number of HDD at baseline was. (12.), which lessened to 4.3 (9.) at 1-month, then varied from 2.5 (6.5) to 4.4 (9.3). As expected among individuals seeking treatment, RTC was high overall, averaging 7 8 on the three rulers. There were no significant attrition patterns (all P values >., correlations <.2) on the variables presented above, except for importance to change which was slightly but significantly related to more follow-up observations (r =.24, P =.3). The negative binomial GEE models in Table 2 show highly significant effects of RTC and confidence to change on both alcohol outcomes. to change was a significant predictor of change in HDD, and its link with ABD was in the expected direction, but did not reach significance (P =.15). Imputing the readiness rulers values collected at each trimester interview to the following 2 months interviews showed the same pattern of effects, only slightly reduced (Table 3). When trichotomized to test linearity of the readiness rulers effects on alcohol outcomes, we observed similar patterns, as shown in Table 4. In all models, there was a gradient in effects, which were stronger for high vs. low scores than for medium vs. low scores. Significance patterns were consistent with those of continuous variables, except for medium RTC, which was not significantly related to more abstinence days at follow-up (P =.21). Replicating this analysis by imputing the readiness rulers values collected at each trimester interview to the following 2 months interviews showed the same pattern of effects, with two exceptions. Medium levels of importance and readiness were significantly related to lower HDD than low levels, but high levels on these scales were not significantly different from medium levels (data not presented, but available on request to the first author). Graphs presented in Fig. 1 confirm the patterns described above. A clear gradient appears for confidence to change on both outcomes. This gradient also appears for RTC, but is milder. A trend can be found for importance to change, but symbols are more intermixed. Table 2. Effect of readiness rulers (linear measures) on alcohol outcomes measured the month after four trimester interviews B SE z P [95% CI] Outcome: Abstinence days.3.2 1.45.15.1.8.5.2 2.83.5.2.9.6.2 2.88.4.2.11 Outcome: HDD.11.3 4.21 <.1.16.6.14.2 6.57 <.1.19..25.2.21 <.1.3.2 N = 78, with 248 observations of alcohol outcome measured the month after four trimester interviews. Models are six negative binomial GEE, set with exchangeable correlation between alcohol outcome measures, and adjusted for gender, age, baseline alcohol dependence severity (ADS score), and AUD treatment status (in treatment vs. not). SE, standard error; CI, confidence interval. Table 3. Effect of readiness rulers (linear measures) on alcohol outcomes measured 1-, 2- and 3-month after the four trimester interviews DISCUSSION B SE z P [95% CI] Outcome: Abstinence days.1.1.74.46.2.4.3.1 2.91.4.1.5.4.1 3.3.1.2.7 Outcome: HDD.5.2 3.28.1.9.2.8.1 5.84 <.1.11.5.15.2 9.5 <.1.18.12 N = 78, with 74 observations of alcohol outcome measured each month post baseline (1-month to 12-month follow-up). rulers values collected at each trimester interview were imputed to the following 2 months interviews. Models are six negative binomial GEE, set with exchangeable correlation between alcohol outcome measures, and adjusted for gender, age, baseline alcohol dependence severity (ADS score) and AUD treatment status (in treatment vs. not). SE, standard error; CI, confidence interval. Table 4. Effect of readiness rulers (trichotomized) on alcohol outcomes B SE z P [95% CI] Outcome: Abstinence days Med. [6 9] vs. Low [ 5].11.17.65.52.22.44 High [] vs. Low [ 5].26.16 1.62.11.6.57 High [] vs. Med. [6 9].15.13 1.12.26.11.41 Med. [6 9] vs. Low [ 5].21.17 1.27.21.12.54 High [] vs. Low [ 5].44.15 2.97.3.15.73 High [] vs. Med. [6 9].23.14 1.64..5.51 Med. [6 9] vs. Low [ 5].24.12 1.9.6.1.48 High [] vs. Low [ 5].44.16 2.71.7.12.75 High [] vs. Med. [6 9].2.15 1.35.18.9.49 Outcome: HDD Med. [6 9] vs. Low [ 5].2.2.99.32.58.19 High [] vs. Low [ 5].76.19 3.98 <.1 1.13.39 High [] vs. Med. [6 9].56.16 3.51 <.1.88.25 Med. [6 9] vs. Low [ 5].5.2 2.51.1.89.11 High [] vs. Low [ 5] 1.13.18 6.27 <.1 1.49.78 High [] vs. Med. [6 9].64.18 3.59 <.1.98.29 Med. [6 9] vs. Low [ 5].81.15 5.28 <.1 1.11.51 High [] vs. Low [ 5] 2.65.27 9.93 <.1 3.18 2.13 High [] vs. Med. [6 9] 1.84.26 7.17 <.1 2.34 1.34 N = 78, with 248 observations of alcohol outcome measured the month after four trimester interviews. Models are negative binomial GEE, set with exchangeable correlation between alcohol outcome measures, and adjusted for gender, age, baseline alcohol dependence severity (ADS score) and AUD treatment status (in treatment vs. not). SE, standard error; CI, confidence interval; Med., Medium. As hypothesized, higher scores on the importance, readiness and confidence to change rulers were associated with improved alcohol use outcomes overall (irrespective of AUD treatment, age, gender and

Alcohol and Alcoholism, 217, Vol. 52, No. 1 69 Outcome: Average abstinence days (ABD) over the next 3 days Next month ABD 3 2 1 2 3 4 5 6 7 8 9 11 Next month ABD 3 2 1 2 3 4 5 6 7 8 9 11 Next month ABD 3 2 1 2 3 4 5 6 7 8 9 11 Outcome: Average heavy drinking days (HDD) over the next 3 days Next month HDD 15 5 1 2 3 4 5 6 7 8 9 11 dependence severity) in a cohort of adults with AUD recruited at an outpatient clinic. The strongest effects were for confidence, which consistently predicted the two alcohol outcomes (increases in number of abstinence days and decreases in number of HDD) in all analyses. also significantly predicted both outcomes, but further analyses showed small inconsistencies (only high levels versus low levels predicted the number of abstinence days; no difference between medium and high levels on HDD when considering alcohol use over 3 months after readiness measure). The predictive validity of importance to change was weaker. The associations were in the expected direction, but were significant only for the number of HDD, and further analyses showed small inconsistencies across levels of this measure. Overall, effects were similar (even if somewhat smaller) when predicting alcohol use the month after RTC measure (i.e. four measures, one at each trimester), and when predicting alcohol use the 3 months after RTC measure (i.e. trimester measures imputed to the next 2 months). This suggests that predictive value of readiness rulers, when observed, might be quite consistent over time. Despite the small sample size of 78 individuals, an important strength of this study is the longitudinal design. The three rulers were assessed four times, and outcomes were measured each month over 12 months (74 observations, 9. per patient on average, with no significant attrition patterns, except for importance to change which was lightly related to more follow-up observations). Finding significant predictive validity over time is consistent with the theoretical view of RTC as a dynamic construct (Prochaska and Velicer, 1997). Similarly, drinking patterns also tend to fluctuate over time (e.g. Maggs et al., 211; Daeppen et al., 213; Voogt et al., 214). The results herein are in line with those of Kaysen et al. (29) and Merrill et al. (215). Moreover, we observed this association between RTC measures and alcohol use over time in a different Next month HDD 15 5 1 2 3 4 5 6 7 8 9 11 Next month HDD 15 5 1 2 3 4 5 6 7 8 9 11 Fig. 1. Average alcohol use over time, by categories of importance, readiness and confidence to change. ABD/HDD: abstinence/heavy drinking days over the next 3 days (3-day TLFB assessed at next month follow-up). Shaded bars correspond to months during which readiness rulers were measured (i.e. trimester interviews). These measures were imputed to the following next 2 months and thus symbols on non-shaded lines represent effects on alcohol use 2 and 3 months, respectively after the measure. sample (adult AUD-treatment patients) than the college students in Kaysen et al. (29) and Merrill et al. (215). Additionally, our timeframe of 12 months with monthly assessments was longer than the 11 weeks with weekly assessments in Kaysen et al. (29) and Merrill et al. (215). Regarding RTC dimensions, the strongest effects were for confidence to change, which consistently predicted both alcohol outcomes. This dimension seems to be a reliable predictor of improved outcomes. Similar results were observed among primary care patients with unhealthy alcohol use (Williams et al., 27) and among Swiss young men from the general population (Bertholet et al., 212). Conversely, importance to change showed contrasted findings. Interestingly, the importance to change evidence in the literature is even more incongruous. For example, the importance ruler was significantly related to lower odds of heavy drinking among Swiss young men from the general population (Bertholet et al., 212), as opposed to reflecting severity of use and being related to poorer outcomes, as in several other studies (Maisto et al., 1999; Williams et al., 27; Bertholet et al., 29). One limitation of the present analysis is the restricted range of data available on RTC and related dimensions in the parent dataset. Only the three readiness rulers were assessed. In addition, one of these measures is itself called readiness, while all three are aimed at measuring the construct of RTC. Following Rollnick (1998), RTC is a combination of importance and confidence. However, the restricted range of variables and the limited sample size prevented us to conduct more in depth analyses about these relationships and potential redundancy among these measures. We clearly see this work as a first step in testing RTC using longitudinal data, which would require further investigation and replication in order to address the theoretical issues evoked here.

7 Alcohol and Alcoholism, 217, Vol. 52, No. 1 Nevertheless, the findings that RTC (particularly confidence to change) is predictive of positive alcohol outcomes over time might be of major clinical interest, especially for treatments targeting shifts in RTC and/or elicitation of self-efficacy. Nevertheless, an additional empirical step is needed to establish whether treatment can actually elicit or increase RTC. We were not able to test this in the present study because treatment modalities were not recorded in detail and there was no control group or treatment comparison in the parent trial (Daeppen et al., 213). There is some evidence that pretreatment RTC is related to AUD treatment outcome (e.g. Isenhart, 1997; Project MATCH Research Group, 1997; Hernandez-Avila et al., 1998; Project MATCH Research Group, 1998; Demmel et al., 24; Adamson et al., 29). However, evidence is scarce and less consistent regarding whether AUD treatment might actually elicit or increase RTC (e.g. Borsari et al., 29; Barnett et al., 2; Heather et al., 213). Further research should focus on the full causal relationships between pre-treatment RTC, AUD treatment components, post-treatment RTC, and alcohol outcomes. A handful of studies that have done this have produced some promising results (Stein et al., 29; Cook et al., 215). ACKNOWLEDGEMENT We thank Dr George Danko for the careful editing of the manuscript. FUNDING The CONTROL cohort study and this work were supported by Lundbeck SAS. CONFLICT OF INTEREST STATEMENT This study was funded by Lundbeck SAS. The funder had no influence on data analysis and drafting of the manuscript. The authors have no other conflicts of interest to declare in relation to this article. REFERENCES Adamson SJ, Sellman JD, Frampton CM. (29) Patient predictors of alcohol treatment outcome: a systematic review. J Subst Abuse Treat 36:75 86. Barnett NP, Apodaca TR, Magill M, et al. (2) Moderators and mediators of two brief interventions for alcohol in the emergency department. Addiction 5:452 65. Bertholet N, Cheng DM, Palfai TP, et al. (29) Does readiness to change predict subsequent alcohol consumption in medical inpatients with unhealthy alcohol use? Addict Behav 34:636 4. Bertholet N, Gaume J, Faouzi M, et al. (212) Predictive value of readiness, importance, and confidence in ability to change drinking and smoking. BMC Public Health 12:78. Biener L, Abrams DB. (1991) The Contemplation Ladder: validation of a measure of readiness to consider smoking cessation. Health Psychol : 36 5. Borsari B, Murphy JG, Carey KB. (29) to change in brief motivational interventions: a requisite condition for drinking reductions? Addict Behav 34:232 5. Carey KB, Henson JM, Carey MP, et al. (27) Which heavy drinking college students benefit from a brief motivational intervention? J Consult Clin Psychol 75:663 9. Cook S, Heather N, McCambridge J, et al. (215) Posttreatment motivation and alcohol treatment outcome 9 months later: findings from structural equation modeling. J Consult Clin Psychol 83:232 7. Daeppen JB, Faouzi M, Sanglier T, et al. (213) Drinking patterns and their predictive factors in CONTROL: a 12-month prospective study in a sample of alcohol-dependent patients initiating treatment. Alcohol Alcohol 48:189 95. Demmel R, Beck B, Richter D, et al. (24) to change in a clinical sample of problem drinkers: relation to alcohol use, self-efficacy, and treatment outcome. Eur Addict Res :133 8. DiClemente CC, Schlundt D, Gemmell L. (24) and stages of change in addiction treatment. Am J Addict 13:3 19. Harris TR, Walters ST, Leahy MM. (28) to change among a group of heavy-drinking college students: correlates of readiness and a comparison of measures. J Am Coll Health 57:325 3. Heather N, McCambridge J, UKATT Research Team (213) Post-treatment stage of change predicts 12-month outcome of treatment for alcohol problems. Alcohol Alcohol 48:329 36. Hernandez-Avila CA, Burleson JA, Kranzler HR. (1998) Stage of change as a predictor of abstinence among alcohol-dependent subjects in pharmacotherapy trials. Subst Abus 19:81 91. Isenhart CE. (1997) Pretreatment readiness for change in male alcohol dependent subjects: predictors of one-year follow-up status. J Stud Alcohol 58:351 7. Kaysen DL, Lee CM, Labrie JW, et al. (29) to change drinking behavior in female college students. J Stud Alcohol Drugs s16:6 14. Lecrubier Y, Sheehan DV, Weiller E, et al. (1997) The Mini International Neuropsychiatric Interview (MINI). A short diagnostic structured interview: reliability and validity according to the CIDI. Eur Psychiat 12:224 31. Liang K-Y, Zeger SL. (1986) Longitudinal data analysis using generalized linear models. Biometrika 73:13 22. Maggs JL, Williams LR, Lee CM. (211) Ups and downs of alcohol use among first-year college students: number of drinks, heavy drinking, and stumble and pass out drinking days. Addict Behav 36:197 22. Maisto SA, Conigliaro J, McNeil M, et al. (1999) Factor structure of the SOCRATES in a sample of primary care patients. Addict Behav 24: 879 92. Merrill JE, Wardell JD, Read JP. (215) Is readiness to change drinking related to reductions in alcohol use and consequences? a week-to-week analysis. J Stud Alcohol Drugs 76:79 8. Miller WR, Rollnick S. (22) Motivational Interviewing: Preparing People for Change, 2nd edn. New York, NY: Guilford Press. Miller WR, Rollnick S. (213) Motivational Interviewing: Helping People Change, 3rd edn. New York, NY: Guilford Press. Norcross JC, Krebs PM, Prochaska JO. (211) Stages of change. J Clin Psychol 67:143 54. Palfai TP, McNally AM, Roy M. (22) Volition and alcohol-risk reduction: the role of action orientation in the reduction of alcohol-related harm among college student drinkers. Addict Behav 27: 39 17. Prochaska JO, Velicer WF. (1997) The transtheoretical model of health behavior change. Am J Health Promot 12:38 48. Project MATCH Research Group. (1997) Matching alcoholism treatments to client heterogeneity: project MATCH Posttreatment drinking outcomes. J Stud Alcohol 58:7 29. Project MATCH Research Group. (1998) Matching alcoholism treatments to client heterogeneity: project MATCH three-year drinking outcomes. Alcohol Clin Exp Res 22:13 11. Rollnick S. (1998), importance and confidence: critical conditions of change in treatment. In Miller WR, Heather N (eds), Treating Addictive Behaviour, 2nd edn. New York: Plenum. Shealy AE, Murphy JG, Borsari B, et al. (27) Predictors of motivation to change alcohol use among referred college students. Addict Behav 32:2358 64. Skinner HA, Horn JL. (1984) Alcohol Dependence Scale: Users Guide. Toronto: Addiction Research Foundation. Sobell LC, Cunningham JA, Sobell MB, et al. (1996) Fostering self-change among problem drinkers: a proactive community intervention. Addict Behav 21:817 33. Sobell LC, Sobell MB. (1995) Alcohol consumption measures. In Allen JP, Columbus M (eds), Assessing Alcohol Problems: A Guide for Clinicians and Researchers. Rockville, MD, USA: National Institute on Alcohol Abuse and Alcoholism, 55 73.

Alcohol and Alcoholism, 217, Vol. 52, No. 1 71 Stein LA, Minugh PA, Longabaugh R, et al. (29) to change as a mediator of the effect of a brief motivational intervention on posttreatment alcohol-related consequences of injured emergency department hazardous drinkers. Psychol Addict Behav 23:185 95. Voogt C, Kuntsche E, Kleinjan M, et al. (214) Using ecological momentary assessment to test the effectiveness of a web-based brief alcohol intervention over time among heavy-drinking students: randomized controlled trial. J Med Internet Res 16:e5. Williams EC, Horton NJ, Samet JH, et al. (27) Do brief measures of readiness to change predict alcohol consumption and consequences in primary care patients with unhealthy alcohol use? Alcohol Clin Exp Res 31:428 35.