The aim of this study was to test whether optional videoconferencing increases adherence to and effectiveness of AUD treatment in an RCT.

We hypothesized that the intervention would decrease the number of patients who dropped out prematurely. We tested this by measuring premature dropout at 6-month follow-up (the primary outcome). Additionally, we hypothesized that the intervention would increase the number of patients terminating their treatment course successfully, increase the proportion of patients still attending a treatment course after 6 months from 45% to 70%, and increase treatment outcome. We tested this by measuring successful treatment termination, treatment duration, and the difference in alcohol characteristics from baseline to 1 year into the treatment course (secondary outcomes).

The study was an RCT with 2 groups: treatment as usual (TAU) and treatment as usual with add-on intervention (TAU+I).

All consecutive patients who attended the public outpatient alcohol clinic in Odense, Denmark, between September 2012 and April 2013 were eligible to participate in the study. Inclusion criteria were age 18 years or older, harmful alcohol use or alcohol dependence syndrome according to the International Classification of Diseases, Tenth Revision (ICD-10), and written informed consent. Exclusion criteria were dementia, psychoses, and lack of sufficient Danish language skills. We registered the study with The Regional Committees on Health Research Ethics for Southern Denmark (S-20110052; Multimedia Appendix 1 [25]).

In the outpatient clinic, an interdisciplinary team of social workers, nurses, and psychiatrists conducts the AUD treatment, based on clinical guidelines [26]. The treatment is free of charge and based on face-to-face therapy sessions and pharmacology, if needed [27]. At the beginning of the treatment course, the patients are offered detoxification and motivational interviewing [28]. When they are free of withdrawal symptoms and if they decide to attend a psychosocial treatment course, the patients undergo an assessment interview using the European version of the Addiction Severity Index (EuropASI) [29,30]. Based on an algorithm using the results of the assessment interview, consultant psychiatrists refer the patients to individual psychosocial treatment. This may consist of cognitive behavioral therapy, supportive consultations, family therapy, or contract treatment [31]; as such, there is no difference in the effect of each offer. Treatment is conducted by well-trained and closely supervised nurses and social workers. The length of each treatment course is individually planned. The duration of a typical successful treatment course is about 7 months. One treatment session lasts between 30 and 60 minutes. Session frequency is 1 to 3 times a week at the beginning of the course and 1 session every other week later in the course [32].

We recruited patients during the assessment interview and systematically offered participation to consecutive patients who decided to attend a psychosocial treatment course and met the inclusion criteria. We included patients who decided to participate in the study and randomly assigned them during the assessment interview; we were aiming at recruiting 140 patients, with 70 in each group. Randomization was carried out by the administrative staff, who were not affiliated with the study. The patients drew a nontransparent envelope packed by an independent person. The envelope contained information about their group placements, either the TAU group or the TAU+I group, with their entailments. Figure 1 shows the recruitment and randomization process.

The Addiction Severity Index (ASI) was developed especially for assessment, treatment, and research in addiction [29,33]. Studies have demonstrated the ASI to be a valid instrument [34,35]. The EuropASI [30] provides sociodemographics, alcohol measures, and composite scores for 9 potential problem areas in the patients’ life circumstances. The composite scores reflect the severity of the problems during the last month preceding the assessment interview. The composite scores range from 0 to 1; the higher the score, the greater the severity [29].

Baseline measures were EuropASI sociodemographics about age, sex, higher/continuing education, employment, and cohabitation; EuropASI alcohol measures regarding dependence, age at onset of excessive use of alcohol (≥5 units a day, at least 3 days a week during the last 30 days), years of excessive alcohol use in life, days of alcohol use in the past month, and days of excessive alcohol use in the past month; and EuropASI composite scores regarding alcohol use, drug use, economic status, employment, legal status, family status, social status, medical status, and psychiatric status. The composite scores were computed according to the EuropASI [30].

The primary outcome was premature dropout at the 6-month follow-up. Secondary outcomes were successful treatment termination, treatment duration (measured by the number of days in treatment), and alcohol characteristics, consisting of 12-month follow-up alcohol use and severity (measured by the number of days of alcohol use and excessive alcohol use in the past month preceding the interview), and composite scores regarding alcohol use, employment, legal status, family status, medical status, and psychiatric status.

Baseline data were collected by the therapists at the assessment interview at treatment start prior to the randomization. Follow-up data were collected by the therapists as part of the routine treatment course as long as the patients attended treatment, at the status sessions at 3, 6, and 12 months, and at discharge. We collected additional 1-year follow-up data. To collect the data, we used letters, telephone calls, and personal contacts to secure the highest possible participation rate [36].

In addition, we collected data on the actual use of videoconferencing, and from questionnaires on nonparticipation and satisfaction, and semistructured interviews with participants in the TAU+I group, therapists, and collaborators. We used these additional data for separate analyses of nonparticipation, satisfaction, patient perspectives, and therapist perspectives on the use of videoconferencing.

We conducted the analyses in this study using SAS Enterprise Guide 7.1 (SAS Institute Inc) and Stata v14 (StataCorp LLC).

The power was calculated in Stata. The primary outcome was the number of patients still attending treatment 6 months after the assessment interview. The calculation was based on numbers from the clinic in 2010 showing that the proportion of patients still attending treatment after 7 months was 45%. We expected the intervention to increase the proportion of patients still attending a treatment course after 6 months from 45% to 70%. For a significance level of 5% and a power of 80%, 140 patients should be included, with 70 in each group.

Baseline variables, sex, higher/continuing education, employment, cohabitation, and dependence were categorical; hence, we used Pearson chi-square tests for relationships between variables. The variables higher/continuing education and dependence had an expected frequency of 5 or less in one cell; hence, we used Fisher exact test. The rest of the variables were continuous; hence, we used the Shapiro-Wilk W test for normal data to check for normally distributed data. None of the continuous variables were normally distributed; hence, we used 2-sample Wilcoxon rank sum (Mann-Whitney) tests to compare the means of not normally distributed interval-dependent variables for 2 independent groups.

We tested the primary outcome, premature dropout at 6-month follow-up, by means of a Kaplan-Meier survival analysis using Wilcoxon test statistics.

Secondary outcomes were regarding successful treatment termination, treatment duration, and alcohol characteristics. We tested successful treatment termination by means of a logistic regression analysis. We tested treatment duration by means of a Kaplan-Meier survival analysis at 6-month and 12-month follow-ups. We tested differences in alcohol characteristics from baseline to 1-year follow-up using per-protocol analyses; however, due to missing data, we used last observation carried forward. The variables were continuous; hence, we used the Shapiro-Wilk W test for normal data to check for normally distributed data. The variable employment was normally distributed; hence, we used a 2-sample t test with equal variances to compare means. The rest of the continuous variables were not normally distributed; hence, we used 2-sample Wilcoxon rank sum (Mann-Whitney) tests to compare means. We made no corrections for multiple comparisons.

Our goal was to recruit 140 participants, but we succeeded in recruiting only 71 participants. As the flowchart in Figure 1 shows, 128 consecutive patients entered psychosocial AUD treatment during the period of recruitment. A total of 47 patients declined to participate in the study, 3 patients were assigned to another treatment, and 1 patient was discharged before giving notice. After inclusion, a further 4 patients withdrew from the study, and data were missing for 2. Hence, only 71 patients completed the baseline assessment interview and were randomly assigned: 39 patients to the TAU group and 32 patients to the TAU+I group.

Table 1 shows baseline characteristics of the participants. The 2 groups received the same variation of treatment offers. The average participant was about 47 years old, most were male, and more than half were cohabiting. A majority had higher/continuing education but less than 50% were employed. More than 80% had a diagnosis of alcohol dependence syndrome according to the ICD-10. The 2 groups did not deviate from each other according to EuropASI sociodemographics, alcohol measures, and composite scores. It seems that the allocation of patients to the 2 groups resulted in 2 similar groups with regard to sociodemographic and alcohol characteristics. Therefore, we assumed that the randomization was successful.

Records of the use of the intervention showed that 16 of the 32 patients (50%) in the TAU+I group used the laptop with videoconferencing for a total of 60 treatment sessions; however, 37 (62%) of the sessions had technical problems. Mostly, these problems consisted of poor sound quality, which was solved by using telephones for the sound.

The termination status of the patients in this study fell into 2 categories. The first category is premature dropout, consisting of patients who did not appear at the discharging session but were expected to return, patients who were discharged after not appearing at the treatment sessions, and patients who were discharged by their own wish (at a time considered by the therapist as being too early).

Figure 2 shows premature dropout by the means of a survival analysis. The plot shows the number of days the TAU group and the TAU+I group stayed in treatment or the number who successfully terminated treatment. After 180 days in treatment, 2 of 32 patients (6%) in the TAU+I group and 12 of 39 patients (31%) in the TAU group had dropped out prematurely. The difference is significant (P=.008). After 365 days, 8 patients (25%) in the TAU+I group and 17 patients (44%) in the TAU group had dropped out prematurely. The difference is significant (P=.02).

The most important strength of this small RCT is that it was carried out as an effectiveness study in a real-life setting, where consecutive patients seeking ordinary AUD treatment at the outpatient clinic were offered participation in the study. Studies conducted among a treatment-seeking population are unique and rarely seen. If an experimental intervention in a research study is likely to be implemented and upscaled in real-life praxis, it is an advantage that the research has been carried out among alcohol patients attending an operating clinic. Effectiveness studies generate results that can inform clinical practice [45,46], and examination of the intervention’s effectiveness, when implemented within an everyday clinical setting, is an important part of establishing an evidence base for a particular treatment [47].

However, as a consequence, the findings of this study may not be as positive as findings from other studies with other prerequisites. Most of the previous studies on videoconferencing-based AUD treatment were small feasibility studies or randomized studies with paid patient participation. Our sample was, nevertheless, fairly similar to them regarding sociodemographic and alcohol measures [13-15,17]. Since the study is representative, it can be generalized to the extent of treatment-seeking patients with harmful alcohol use or alcohol dependence, at the higher severity end, attending clinics in Denmark and other countries with a similar organization of, and distances between, clinics.

A severe limitation of this study is that we were unable to include the number of patients estimated in the power calculations prior to study start. The relatively low number of participants may have been due to patient rates being lower than expected, compared with the same time period in previous years in the same setting. Also, it may have been due to the participation rate being lower than expected, based on participation rates in other studies conducted in the same setting. In alcohol treatment and research, it is a common challenge to recruit and maintain patients for studies, as well as for treatment [48]. Unwillingness to participate in research studies has been reported as becoming more and more common [36], especially regarding studies performed over the Internet [49]. Thus, more patients than anticipated may simply have declined participation because of the technical element in the study. As a consequence, the small sample size in this study is a limitation for the significance of the results and may, thereby, have consequences for the inferential conclusions that can be drawn from the results.

It is a huge strength that data on premature dropout and treatment termination were available for all but 2 patients; however, it may be a limitation that we have 1-year follow-up data for only 66%. Prior studies have reported follow-up participation rates below 60% with no evidence of bias [36], and the use of last observation carried forward is a conservative approach to secure validity. In comparison, previous studies on videoconferencing-based AUD treatment have reported relatively good session attendance and successful intervention engagement, as well as completion rates similar to face-to-face treatment [13,15,16], completion rates ranging from 50% to almost 100% [14,15,17], and follow-up rates of up to 90% [16,18]. This may be due to the fact that most of the prior studies were small pilot and feasibility studies or RCTs, where participants were even paid to participate. These recruitment processes may have biased participation in the prior studies in a positive direction compared with participation in effectiveness studies like ours, where participants were consecutive patients seeking treatment for alcohol problems in a real-life setting.

It may be a limitation to the study that the results were based on self-reported EuropASI data. Even though general population surveys have found alcohol consumption to be underreported, and the accuracy of an individual’s report may be difficult to determine, the literature suggests that, from a group perspective, self-reports of alcohol use from clinical and nonclinical samples are accurate provided that people are interviewed under the following conditions: alcohol-free when interviewed; given written assurances of confidentiality; interviewed in a setting encouraging honest reporting; asked clearly worded objective questions; and provided with memory aids [50].

Furthermore, it may be a limitation to this study that we analyzed psychosocial treatment as a single treatment approach, even though it consists of 4 different psychosocial treatment forms. Since the offers were equally effective and the 2 groups received the same variation in treatment offers, we did this to limit the different outcome possibilities as opposed to limiting any broad inferences about the effects of offering videoconferencing.

Moreover, it may be both a strength and a limitation to have chosen videoconferencing as an option in order not to force any patient to use it. None of the previous similar studies have used optional videoconferencing; however, blended care is commonly used in psychiatric treatment. Here, patients reported advantages such as having met the therapist before or during the treatment course [12], and optional use of videoconferencing, throughout the treatment course, offers similar advantages.

The aim of this study was to test whether optional videoconferencing increases adherence and effect in AUD treatment. We tested this by examining premature dropout (the primary outcome), as well as successful treatment termination, treatment duration, and alcohol characteristics (secondary outcomes). The results indicate that offering patients optional videoconferencing may prevent premature dropouts from treatment and prolong treatment courses. However, the small sample size precludes conclusions regarding the effect of the intervention, which was not detectable in the patients’ use of alcohol and severity of problems. Even though videoconferencing did not, in this study, seem to lead to a more effective treatment course, it may be a tool to increase adherence. Thus, future research is warranted on how videoconferencing and treatment duration may influence adherence and effect in AUD treatment.