Community reinforcement and family training (CRAFT) - design of a cluster randomized controlled trial comparing individual, group and self-help interventions

Five types of interventions, aimed at the CSOs, have been described: self-help groups (Al-Anon/Nar-Anon) which are part of the Minnesota treatment family [14], the Johnson Institute Intervention (a very confrontational approach) [15], general unspecific support (typically aimed at supporting the CSO only, but not addressing how to increase the likelihood of getting PWAUD into treatment) [16], The 5-step method [17] and Community Reinforcement and Family Training (CRAFT). So far, CRAFT is the only one of these methods with any evidence to support that it increases the likelihood of the PWAUD seeking treatment [18]. CRAFT was proposed in 1986 by Sisson and Azrin [19] aimed at training CSOs to become involved in the problem drinkers’ choice of treatment, helping CSOs handle situations between themselves and the problem drinkers, and finally helping the CSOs to develop specific strategies to take proper care of themselves in risk situations [20, 21].

CRAFT is aimed at CSOs who struggle, unsuccessfully, to motivate their loved ones to stop drinking and seek treatment. CRAFT promotes active, positive participation from the CSOs in seeking to attract problem drinkers into treatment; changing CSOs’ efforts to help by developing their roles as active collaborators, supportive of the problem drinker. The underlying assumption of CRAFT is, that CSOs already have a detailed knowledge of the problem drinkers’ behaviour and that they are in a strong position to influence PWAUDs’ behaviour because of their concern and personal motivation, including getting PWAUDs to seek treatment [8]. So far, eight randomized or controlled clinical trials on CRAFT have been carried out on CSOs to alcohol dependents and drug users [16, 18‐20, 22‐25]. Three studies focus on alcohol problems [16, 19, 25], three studies on drug users [22‐24], and two studies on both drug users and alcohol dependents [18, 20]. One of the studies on drug users studied CRAFT as a supplement to opioid-dependent adults already in treatment [24]. All studies have been carried out in the USA except one study on alcohol, which was carried out in Germany [25]. The German study compared CRAFT Immediate Intervention with waiting list and found a significantly higher engagement rate (B 1.34 SE 0.6) for the people receiving immediate intervention [25]. The American studies showed a two to three times higher impact in getting PWAUD to attend treatment after four to six CRAFT sessions with CSOs, compared with Al-Anon and Johnson Institute interventions [26]. In general, CRAFT interventions to CSOs lead to more than 60% of the PWAUDs in question to attending treatment in US [16]. Several of these studies have rather small samples from 12 to 40 CSOs [19, 20, 22] The original full intervention of CRAFT consists of 12–14 sessions [19]. Kirby et al. (2017) tested CRAFT in a four-six session intervention, where the only focus in the sessions was on Treatment Entry Training (TEnt) vs. the original CRAFT in 12–14 sessions and Al-Anon/Nar Anon (ANF), and 115 CSOs participated in the study. There was no counseling about e.g. relationship or substance use [18]. The treatment entry rate, after the intervention, was 62% for the CSOs randomized to the full CRAFT and 63% for the ones randomized to Tent, which was significantly higher than the ones receiving Al-Anon/Nar Anon (treatment entry rate 37%). This was to be expected, as the goal of Al-Anon/Nar-Anon is not to encourage treatment entry. No significant differences in mood and functioning were found between the three interventions, even though it was not a subject in the TEnt intervention.

CRAFT has, until now, been examined and shown effective in the USA and Germany. To our knowledge, studies on CRAFT have also been initiated in the Netherlands (ClinicalTrials.​gov ID: NCT02510508) and Sweden (ISRCTN 38220020), but the results are not yet published. In the Netherlands researchers are performing a three-armed RCT with group, self-directed CRAFT or non-intervention addressed to CSOs to alcohol dependents. In Sweden, the effect of a five-week internet-based CRAFT program is tested versus waiting list (ISRCTN 38220020), in addition to another study on an online self-help program combined with a parent-training program for partners suffering from alcohol use disorder, versus a brief psycho-education program [27]. Moreover, a RCT study on CRAFT for CSOs with problem gamblers (CRAFT vs. treatment as usual) is currently being conducted in Sweden [28]. Earlier studies performed are based on small populations, and no study of CRAFT has been performed in Denmark so far.

Furthermore, to our knowledge, only one effectiveness study, performed in operating treatment institutions and as part of routine praxis, has been conducted [29]. The study from Dutcher et al., 2009, tested CRAFT in a community treatment center in the USA. Altogether 99 CSOs were concerned about alcohol abusers, whereas, for all CSOs, 55% of the treatment-refusing abusers entered treatment after 6 months. Among the CSOs, who completed at least four sessions on CRAFT, or the ones who engaged the abuser to treatment, 65% entered treatment. Further, effectiveness studies outside the USA are still essential before large-scale implementation.

The efficacy studies and the effectiveness studies performed, so far, indicate that CRAFT is effective for CSOs towards getting the drinking person into treatment and to improve the quality of life of the CSO and the relationship between the drinking person and the CSO [16, 18‐20, 22‐25]. Whether CRAFT delivered in group, individual or as self-help material is equally effective is, however, still unknown. The study of Manuel et al. indicated, that CRAFT in group condition may be just as effective as individual CRAFT, but the study was indeed small and did not compare the two settings directly. In the study of Manuel et al., 40 CSOs were randomized to either group CRAFT or self-directed CRAFT, and 60% of the CSOs in group CRAFT had their loved ones enter treatment, and for the self-directed CRAFT the result was 40% after six months [20]. Hence, the findings were promising, but not conclusive or significant.

Furthermore, groups can be organized as closed groups or open groups. In closed groups, all CSOs start at the same time and no new members are enrolled, once the treatment has begun. An open group can start when a minimum of two members are enrolled in the study and new members are included continually until the maximum of group members has been reached. When testing CRAFT in a closed group format, Manuel et al. experienced challenges with the start-up, because it took up to one month to gather enough CSOs to start a group [20]. Compared to closed groups, open groups can be joined without a waiting period. Furthermore, an open group format may create opportunities for senior members in the group to share experiences and advice with newcomers [20]. However, an open group format may also be negatively affected by a constant influx and outflux of people in the group, in addition to not all group members receiving the sessions in a logical order. A general strength of group therapy, in proportion to individual and self-help, may be that the CSOs meet like-minded individuals and are able to share similar experiences and feelings and support each other.

The aim of this study is to implement CRAFT interventions into the daily routine of Danish community alcohol treatment centers, and investigate whether 6 week-individual CRAFT, 6 week-open group-based CRAFT or based on CRAFT self-help material only, is efficient in getting problem drinkers to seek treatment for their alcohol problems – and which of the three interventions (individual, group or self-directed CRAFT) is the most effective.

The study is a cluster randomized controlled trial, carried out in public alcohol treatment facilities. All public alcohol treatment centers in Denmark have been invited to participate in a partnership with the Unit of Clinical Alcohol Research at the University of Southern Denmark. In all, 61 institutions were invited to participate, and 17 institutions agreed to participate in the study. The treatment facilities have been randomized to offer one of the three following CRAFT formats:

Based on the data from Manuel et al. [20], we expect 40% of CSOs receiving self-help material and 60% of the CSOs receiving either Group CRAFT or Individual CRAFT to be able to motivate the drinkers to enter treatment. Furthermore, the effect of CRAFT is one-sided as the intervention of CRAFT cannot make the situation inferior for either the CSO or the drinker [8]. Based on these expectations, 106 participants in each group are needed to be able to detect a 20%-point difference with an α level at 5%, a power of 90% and an ICC of 0.05. As we foresee a dropout rate at approximately 10% and, furthermore, a loss to follow-ups rate at 10%, we need to include at least 135 CSOs in each group. This gives an expected total of 405 CSOs.

Consecutive CSOs, who contact a center randomized to conduct either individual CRAFT or open group CRAFT, will be offered the intervention within two weeks of an intake interview. The intervention consists of six sessions with 7–10 days between each session (Fig. 1).

Consecutive CSOs, who contact a center randomized to the control condition (self-directed CRAFT), will be offered self-help material, only. After three months, and when the primary outcome has been measured, the CSOs in this intervention format will, however, have the possibility of an individual follow-up session with a therapist to make sure that the CSOs are helping to move on.

As can be seen from the design, the control group consists of CSOs randomized to self-help intervention. The self-help intervention is chosen as control condition instead of either Treatment as usual or waiting list. Treatment as usual was disregarded since the interventions, currently being offered to the CSOs in the participating treatment centers, differ a lot one from the other, ranging from nothing, to brief advice on the phone, psycho-education, or personal support. A waiting list is disregarded as control condition since participants, waiting for an intervention or treatment, do not act ‘naturally’, but simply wait and become worse than they would have outside the study [30], potentially leading to an overestimation of the effectiveness of the experimental condition. Instead, offering self-help material to the control group serves as a minimal intervention and, for some CSOs, it may even be sufficiently helpful. Should that be the case, it would constitute an interesting finding and be added to the overall implementation of CRAFT nationwide.

The participants of the study are concerned significant others (CSO) to a person with an AUD. There is no consensus definition on a CSO in the literature, but it could, for example, be a spouse, a daughter, a son, a cousin, a friend or a colleague. The CSO must fulfill the criteria listed below to be included in the study. The therapist, who includes the CSO, estimates whether the CSO fulfills the inclusion and exclusion criteria.

Inclusion criteria (CSO):

Exclusion criteria (CSO):

Informative leaflets and posters have been distributed to participating local authorities, who are committed to distribute the leaflets via social services departments, departments for children and adolescents, and to general practitioners. Additionally, we recruit participants through advertisements in local newspapers and via videos and posts on Facebook, linking to the participating alcohol treatment centers’ websites and their Facebook pages. Further, we inform about the project on national websites for counseling about alcohol problems such as www.​hope.​dk and the national telephone hotline ‘Alkolinjen’.

Data will be analyzed using a mixed effect logistic regression model, mixed effects Poisson regression, and univariate statistical models, including t-tests, and chi2-tests, will be used for descriptive statistics. For some of the quantitative secondary outcomes a mixed effect model (for normally distributed data) or a random effect quantile regression model will be used. Strategy for missing data will be based on multiple imputations with special emphasis on the sensitivity of the results of various imputation schemes, as the missing mechanism is likely to be missing not at random.

In view of the fact that the CRAFT intervention has proven highly effective in the US, and since living close to a problem drinker is such a burden to the individual and, finally, because the Danish National Guidelines strongly recommend the implementation of CRAFT, we find it un-ethical and problematical not to offer CRAFT in some form to all the participants, even to the control condition. However, since we expect less effect of the control condition (CRAFT as self-help material), for ethical reasons, the participants in the control group will be offered an individual CRAFT session three month after enrollment. Training significant others in new communication strategies and new ways of acting, in relation to the problem drinker, may lead to increased tension in the family set-up. Domestic violence is already relatively more frequent in families with problem drinking, thus, an increased risk may be anticipated. It is, therefore, highly important to address the risk of domestic violence in therapy. Addressing this risk is mandatory in the interventions and includes developing safety plans together with the CSO. It is, however, important to bear in mind that the risk of abuse and suffering, on the part of the CSOs, is even higher if the problem drinking continues unaddressed. We expect no risks or side-effects for the participants.

Data will be collected in Research Electronic Data Capture (REDCap). Baseline data will be collected directly on an iPad. Data from the follow-up points (three months and six months after baseline) will be collected by means of Web-based questionnaires. All data will be handled and stored by Odense Patient Data Explorative Network (OPEN) [43] and safe SharePoint. The researchers RH, RB, ASN, CE, one data manager, and one statistician will have access to the data and final dataset. The researcher RH and RB make interim analysis and present interim analysis for the scientific advisory board. The researcher RH and RB are responsible for recording adverse events and other unintended effects of the trial intervention.

Only the investigators have the right to publish the results. Firstly, the results will be published in international peer-reviewed journals. Afterwards, the results will be shared in the press and presented at international conferences. Moreover, some of the results will be a part of a Ph.D. thesis. The results will also be communicated to the participants and the therapists in the study.

The study will be carried out by the Unit of Clinical Alcohol Research (UCAR), The Clinical Research Institute, and The University of Southern Denmark. UCAR is governed by two complementary functions: A Research Office and a Scientific Advisory board.

The UCAR Research Office manages the daily research operations of UCAR and comprises UCAR’s director, the leading Principal Investigators and administrative staff. The Research Office is responsible for the general administration of research projects, including the present study, and provides specialized services to researchers. Regular meetings will take place between The Research Office, key-persons and leaders of the treatment centres, to inform about progress of the study. Furthermore, a monthly newsletter will be sent to all parties involved in the study.

The Scientific Advisory board meets three times a year, to discuss the progress in study. The scientific Advisory board have the right to make the final decision to terminate the project. The sponsors have no influence on the process.

The first four months of the inclusion-period (from January 2018) have been considered as the pilot phase, in which all procedures, from inclusion of the first participants in the study to the first follow-up interview after 3 months, are tested and evaluated. Together with the participating treatment centers, we have evaluated all the procedures. We have also estimated the duration of the entire inclusion period of 405 participants. Regardless that we only have included 60 CSOs in the first four months, we decided to invite more treatment institutions to the project. This resulted in an additional seven institutions to be trained, randomized and commence the inclusion of CSOs from the first of September 2018.