Alcohol misuse and dependence is a major problem among opioid- and cocaine-dependent patients [1–4]. This association is linked to lower response to methadone substitution [3–5], more frequent overdose [6–9], lower quality of life [10], and higher risk of impulsive behaviors [11]. Alcohol consumption is common among cocaine users [12]. It can trigger cravings for cocaine and thus relapse [12, 13] and may reduce anxiety and dysphoria induced by cocaine withdrawal [12, 13], leading to alcohol use. This clinical evidence increases interest in specific alcohol interventions for patients with cocaine dependence. Despite the importance of alcohol-related problem among patients with opiate and cocaine dependence, however, few studies have evaluated specific interventions within this group.

Brief intervention (BI) is one such intervention that could be of particular interest for its simplicity and accessibility. Among excessive drinkers who were treated in primary care settings, BI reduced the percentage of excessive drinking, the quantity and frequency of alcohol consumption, and the negative consequences associated with alcohol use [14–16]. These results were confirmed in settings such as emergency services [17] and practitioners' clinics [18], as well as in various adolescent populations [19]. In addition, positive results were published in the BRAINE study, a randomized clinical trial that evaluated BI in 197 intravenous drug users [2].

Several less favorable results have also been shown, however, in studies of alcohol-dependent patients [18, 20], dual diagnosis patients [21], and patients in various other settings, such as emergency [22] and general hospitals [23]. These disparities among studies are possibly due to differences in the patient and setting characteristics.

Although the efficacy of BI in different settings remains controversial, it seems to be a simple and relatively efficient intervention [24, 25]. The aim of the present randomized controlled study was to evaluate the impact of screening with the Alcohol Use Disorders Identification Test (AUDIT)[26]--a 10-item self-assessment questionnaire for alcohol abuse and dependence--and of BI on alcohol use among patients treated for opioid and cocaine dependence in a specialized outpatient clinic.

Three hundred patients were contacted for inclusion. Among them, 32 declined study participation and 14 were not included due to their engagement in an alcohol-related treatment. Finally, 254 patients (mean age: 35 years; range: 18-56; 72.3% male) signed written informed consent, completed the first evaluation, and were included in the study (Figure 1; Table 1).

Some patients were treated for opiate dependence with methadone substitution (56.2%) or diacethyl morphine (heroin treatment; 12%). Other patients had no opioid substitution and were treated for opiate or cocaine dependence (31.7%). Acceptance to participate in the study was similar (broad; wide) for both groups (patients with or without substitution). Most patients with cocaine dependence or with opiate dependence also had tobacco or cannabis dependence. Most patients had one or more concomitant psychiatric disorders (mood disorder, 35.6%; personality disorder, 34%; anxiety disorders, 14.7%; psychotic disorders, 9.4%). Logistic regressions showed that the type of drinker and treatment group did not explain the missingness of data. At T3, the odds ratio was 1.03 with confidence intervals of [0.46; 2.29] for type of drinker and 1.10 [0.50; 2.43] for treatment groups. At T9, the odds ratio was 0.78 with confidence intervals of [0.35; 1.75] for type of drinker and 1.22 [0.55; 2.70] for treatment groups. Hence, these variables displayed no particular pattern, meaning that the data for excessive drinkers and for alcohol-dependent patients, as well as for the control group and the intervention group, were equally likely to be missing.

The AUDIT shows an excellent internal consistency in this population (Cronbach α = 0.90). According to AUDIT cutoffs, 142 patients (55.9%) had no alcohol problems, whereas 112 (44.1%) were considered problematic drinkers. At inclusion, among those with problematic alcohol consumption, 43.8% were classified as excessive drinkers and 56.2% as alcohol dependents.

In the present study, significant decreases in AUDIT scores were observed at T3 and T9 in both treatment groups, without gender differences, as previously found elsewhere [14, 18, 20, 31]. In our study, gender differences were not linked to treatment effect but to a more general pattern of alcohol consumption. Similarly, a significant decrease in alcohol consumption was observed at T3. The main changes in alcohol consumption (AUDIT scores, number of drinks, and frequency) are observed at T3, but changes in frequency and number of drinks do not persist at T9. This is possibly due to a reduction of the intervention effect over time.

Of interest, these results are in accordance with other studies founding a significant decrease in the consumption of alcohol following BI [14, 15], including among a population of intravenous drug users [2]. Astonishingly, dependent subjects improved better than excessive drinkers in AUDIT scores between T0 and T3. This difference between types of drinker was not observed for the number of drinks consumed per week. In the present setting, BI did not have an additional effect in comparison with AUDIT screening. This is possibly due to a lower impact of BI in this target population [20, 21, 32]. BI has previously been found to be less efficient among alcohol dependents [14, 24]. Furthermore, lack of differences between interventions may simply show that screening alcohol use with AUDIT and providing feedback about the score has a positive impact on alcohol-drinking behaviors [33]. This phenomenon has been evoked by different studies and could correspond to what Jenkins et al. call reactivity to assessment [34]. The results were also possibly due to spontaneous adaptations of the naturalistic "treatment as usual" received by both groups following AUDIT screening or BI.

In this study, screening with the AUDIT, as well as the BI, induced changes in alcohol consumption among heroin and cocaine users in treatment. This result possibly means that in an addiction clinic, both strategies were useful. In the absence of a naturalistic or waiting list group, we cannot fully exclude a spontaneous favorable evolution linked to time or to other concomitant factors such as treatment as usual. It is probable, however, that in the absence of a specific intervention targeting alcohol consumption, patients treated for opiate or cocaine addiction may not usually receive sufficient attention for alcohol misuse.

In fact, one of the limitations of the present study is the lack of data about naturalistic change during treatment as usual. Participation in the study was voluntary and all participants were selected by convenience sampling when they came to the Division of Substance Abuse for treatment. As there is a lack of studies that have evaluated the longitudinal effect of the intervention within this specific group of drug-addicted patients, no a priori sample size calculation had been undertaken. Indeed, the extreme heterogeneity of patient characteristics, health care, and study settings made it difficult to provide reliable effect size estimates for sample size calculation for continuous endpoints. If we had to take, for instance, an effect size of 30% and a correlation among the repeated observations of 0.4, using the equation for sample size calculations given by Diggle et al. [35], with a power of 0.8, would yield a minimum sample size of 82 persons in each arm. This estimate, if a plausible guess, shows that the current study sample size may have proven insufficient to reject the null hypothesis at the 0.05 level. Hence, the study may be underpowered and may have led to discarding the BI as a potentially useful treatment.

Nevertheless, because of the simplicity of the proposed interventions and the relatively important improvements observed (reduction in the number of drinks per week from 20.9 at T0 to 13 at T3 in the control group and from 25 at T0 to 15.4 at T3 in the intervention group), systematic screening with AUDIT, followed by feedback on the score, and/or BI for alcohol misuse among outpatients treated for opiate or cocaine dependence will be promoted and investigated as having promising potential in this target group. The screening, followed by feedback on the problematic use of alcohol, is an interesting strategy to include in a population of heroin and cocaine users. New tools, such as the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST) questionnaire, deserve to be evaluated in future research for screening the use of alcohol and other substances [36].

As previously found [1], another important result of the present study is the excellent internal consistency of AUDIT in a population of patients treated for opiate or cocaine dependence. Using this instrument, we found a 44% rate for problematic alcohol consumption (excessive drinking and dependence), confirming results of other studies on similar populations [2, 3].