Introduction
Approximately 2.3% of the Swedish population aged 16–85 years are considered to be either problem gamblers or moderate-risk gamblers (Swedish National Institute of Public Health
2010), as measured by the Problem Gambling Severity Index (PGSI) (Ferris and Wynne
2001). Problem gambling is associated with significant harm for the gambler such as economic difficulties, psychological distress, comorbid substance abuse, suicidality and physical health problems (Langham et al.
2016; Moghaddam et al.
2015; Petry
2009; Swedish National Institute of Public Health
2010).
A Swedish study estimated that 18% of the general population in Sweden could be considered as concerned significant others (CSOs) of someone who is or has been a problem gambler (Svensson et al.
2013). Problem gambling produces significant harm and distress for the CSOs of the problem gambler. Problem gambling is also known to cause financial problems for the CSOs of problem gamblers, as well as psychological and physical ill health (Downs and Woolrych
2010; Kalischuk et al.
2006; Patford
2009). Many CSOs report that their relationship with the problem gambler is severely affected, and they also describe disturbed relationships with other family members and friends (Shaw et al.
2007; Wenzel et al.
2008).
Various treatment programs have been evaluated for problem gambling and many of the successful treatment approaches are based on knowledge gained within the framework of treating substance-use disorders (Ferentzy and Turner
2013). This seems logical given the overlap in symptoms between problem gambling and substance-use disorders, as well as a possible parallel biological dysfunction and a substantial degree of comorbidity (Frascella et al.
2010; Goudriaan et al.
2006; Lorains et al.
2011; Petry
2009). In a Cochrane review (Cowlishaw et al.
2012) of psychological treatments for those with a gambling disorder, the authors found support for the efficacy of
cognitive behavior therapy (CBT). Based on the results from seven randomized controlled trials (RCTs), the authors concluded that CBT reduced gambling behaviors (Gambling symptom severity; Cohen’s
d pre–post-test: −1.82; 95% CI −2.61 to −1.02) and depression and anxiety symptoms compared to a control condition. Two other meta-analyses have reached similar conclusions (Gooding and Tarrier
2009; Pallesen et al.
2005) and two RCTs finding support for CBT interventions have been performed in a Swedish context (Carlbring et al.
2010,
2012).
Despite the accumulation of negative consequences associated with problem gambling, and despite the promising results of CBT for problem gambling, a mere 5–12% of problem gamblers ever seek treatment (Slutske
2006; Swedish National Institute of Public Health
2010). This has generally been attributed to stigma, a lack of accessibility to treatment, and/or an unwillingness to admit to the problem and a desire to handle problems oneself (Bellringer et al.
2008; Clarke et al.
2007; Suurvali et al.
2009). Among those who do seek treatment, adherence is low and attrition rates are high; a systematic review found that, on average, 42% of participants drop out of psychological interventions for problem gambling (Melville et al.
2007). However, including CSOs in the treatment increases gambler retention and merely having a CSO seems to increase the odds of successful treatment (Ingle et al.
2008; Kourgiantakis et al.
2013). Furthermore, concerns of CSOs have been identified as one of the main reasons for problem gamblers entering treatment (Hing et al.
2012; Ingle et al.
2008; Tepperman et al.
2006). In addition, the CSOs may have a limited understanding of—and may be less aware of—the full extent of the gambling problem (Tepperman et al.
2006), a situation which could unintentionally enable further gambling (Patford
2009).
Data from the Swedish National Gambling Helpline shows that roughly half of the helpline contacts are with the CSOs of gamblers, which indicates that among CSOs, there is an unmet need for support (Stockholm Centre for Psychiatric Research
2016). Parents and partners make up the majority of CSO contacts at roughly 30% each, while friends, siblings and relatives make up approximately 10% each. Children, including adult children of gamblers, make up less than 5% of the helpline contacts. However, research on interventions for problem gambling involving CSOs is scarce. A study of CBT group treatment where family members were invited to participate in the treatment found that CSO involvement was associated with a higher relapse rate (Jimenez-Murcia et al.
2015) and the authors recommended that CSOs and gamblers should be given separate interventions. Another pilot study involving 18 couples studied the impact of
congruence couple therapy, involving both the gambler and a partner in couple therapy. Contrary to the study by Jimenez-Murcia et al., the latter showed significant changes in gambling symptoms and in psychological distress for the CSO (Lee and Awosoga
2015).
A handful of studies have studied interventions involving solely the CSOs of problem gamblers. Three studies have investigated
community reinforcement and family training (CRAFT) for gambling (Hodgins et al.
2007; Makarchuk et al.
2002; Nayoski and Hodgins
2016). CRAFT was originally aimed at working with the CSOs of people with alcohol and substance problems, but has been adapted to suit the CSOs of problem gamblers. The major aim is to get “treatment-refusing” gamblers into treatment, and while this has been a successful intervention for substance abuse (Roozen et al.
2010), CRAFT for gambling has so far failed to increase treatment engagement, but has produced significant results regarding the number of days gambled (among the gamblers) and also program satisfaction for the CSOs. Another study investigating the impact of coping skills training for CSOs found significant reductions in the symptoms of depression and anxiety among CSOs, but there were no changes in treatment entry or gambling (Rychtarik and McGillicuddy
2006).
CSO involvement in clinical trials targeting alcohol and substance abuse has been somewhat better studied. One treatment approach that has been successful in treating the person with an addiction as well as involving the spouse in the treatment is
behavioural couples therapy (BCT) (Meis et al.
2012). BCT resembles CBT approaches to substance abuse, such as traditional CBT and CRAFT, and incorporates CBT techniques targeting substance abuse such as functional analysis, relapse prevention and behavioral activation with interventions targeting relationship functioning (O’Farrell and Fals-Stewart
2006). BCT involves both the user and a CSO and has two main goals: (1) to build support for abstinence and (2) to improve relationship functioning. The hypothesized mechanism of change is that improved relationship functioning will promote relationship behaviors that are conducive to abstinence (O’Farrell and Fals-Stewart
2006). A meta-analysis of 12 BCT RCTs, of which 8 targeted alcohol problems and 4 targeted other substances, showed better outcomes for BCT than for individual-based treatments, with a mean overall between-group effect size of Cohen’s
d = 0.44 in favour of BCT (Powers et al.
2008). BCT has been tested for different types of relationships (e.g. heterosexual couples, same-sex couples and the parent–child relationship) and for different types of substances (e.g. alcohol, illegal drugs and methadone). BCT is without doubt the therapy involving significant others aimed at addiction with the most robust research support to date (Fletcher
2013). As mentioned above, there is already some support for the notion that CSO involvement in problem gambling treatment produces better outcomes for the gambler in terms of gambling and for CSOs in terms of psychological distress, even though the results are somewhat ambiguous. It thus seems appropriate to investigate the potential effects of BCT on problem gambling.
According to the Swedish Gambling Authority, more than half (55%) of the gamblers in Sweden play online (Swedish Gambling Authority
2015). Figures from the Swedish National Gambling Helpline reveal that the three most common problem games among gamblers contacting the helpline were Internet-related: Internet casinos, Internet betting and Internet poker (Stockholm Centre for Psychiatric Research
2016). Not only are the gambling problems increasingly Internet-associated, but a growing number of contacts to the helpline are made through chat or e-mail. Internet-delivered treatments have been proposed as potential treatment alternatives for problem gambling, in part because they are flexible, anonymous and available nationwide, and thus could lower barriers to treatment. Carlbring et al. (
2012) tested Internet-delivered CBT in Sweden and found significant reductions in gambling problems as well as in comorbid disorders. The Swedish studies have also been replicated in a Finnish context (Castren et al.
2013). Several other studies also suggest that Internet-delivered interventions could be viable treatment options (Canale et al.
2016; Myrseth et al.
2013; Rodda et al.
2013). Given the above-mentioned issues regarding the reluctance among problem gamblers to seek and remain in regular face-to-face treatment, in combination with the lack of studies investigating the role of CSOs in problem gambling treatment, we found reason to further develop and study treatment approaches for problem gambling.
Aim
The primary aim with the current study was to investigate whether the involvement of CSOs in treatment would affect treatment response among problem gamblers in an Internet-delivered pilot study comparing two conditions: BCT involving both the gambler and a CSO versus CBT for the gambler only. Further, a secondary aim of this pilot study was to investigate the feasibility of the program before conducting a full-scale RCT.
Treatment Arms
After randomization, the participants gained access to a treatment website containing their respective treatment programs. The BCT as well as the CBT program consisted of one chapter/module each week, containing text material, short films and tasks related to a specific treatment component. Each module was, on average, 5–10-pages long. In the BCT condition, the gambler and the CSO were given 10 modules each. In the CBT condition, the gambler was given 10 modules but the CSO was not given any modules. The content in the treatment modules for the gamblers was constructed to be as similar as possible, regardless of the trial arm. Each treatment arm contained 10 modules, thus lasting 10 weeks. Participants were, however, given the opportunity to complete the program during a 12-week time frame to increase flexibility. The modules were made available to the participants one at a time as participants advanced in the treatment.
Although the CBT arm and the BCT arm for the involved gamblers were similar to a large degree, the gamblers in the BCT condition were asked to collaborate with their CSO throughout treatment and several exercises in each module were designed to involve both the gambler and the CSO.
The modules were complemented with scheduled telephone and e-mail support from their assigned therapist. E-mail communication was administered via an online messaging system that is built into the treatment platform. The gamblers and the CSOs had their own log-ins and were not able to read each other’s content or communications.
Feedback from the Therapists
The two therapists were asked general questions on their experience of working on this pilot study. While an exhaustive account of their experiences is beyond the scope of this article, a brief résumé of their answers provides an insight into the feasibility of conducting a full-scale RCT. The questions were as follows:
-
What is your general experience of working with “Gambling Free Together”?
-
What would you change in “Gambling Free Together” if you had the opportunity?
-
What were the greatest challenges in working with “Gambling Free Together”?
-
What was most time-consuming when working with “Gambling Free Together”?
The therapists generally emphasized the positive aspects of the study: the length of the treatment, the content in the modules and the satisfaction of working with participants who benefited from the interventions.
In terms of the negative aspects, both therapists stressed that they spent a lot of time trying to reach participants who did not answer their e-mails or telephone calls. According to the therapists this could be attributed to a lack of motivation on behalf of the participants, but perhaps also to the participants’ chaotic life circumstances. One therapist described this chasing as “tiresome and demoralizing.” In many cases, the two participants in the BCT condition would complete the modules at a very different pace. This posed a problem to the therapists, but also to the treatment program itself, since the program builds on the assumption that participants will cooperate and perform certain exercises together. One therapist suggested that some exercises concerning relationship functioning should be provided earlier in the treatment program to give the participants “a common ground and a common start.”
One therapist asked for more tailored interventions that could be adapted to suit the different needs of the participants. Both therapists concluded that a challenge to this intervention was when the CSO was more motivated than the gambler and could be assumed to have pushed the gambler into treatment.
Treatment Results
Both treatments successfully lowered the symptoms of problem gambling and measures of depression and anxiety for the gamblers. Both groups went from NODS levels corresponding to pathological gambling to levels corresponding to a mild but subclinical risk for problem gambling. The NODS findings were consistent with results from the TLFB-G indicating rapid decreases in time spent and money lost on gambling after the intervention had started. There were no clear differences between the two conditions on any gambling-related outcome measure, since both treatments significantly lowered all outcomes related to gambling. Most gamblers in both groups seemed to abstain from gambling altogether during the treatment period, but it is unclear if this represents a permanent change in gambling behavior. This makes it difficult to draw conclusions regarding the relative efficacy of the treatments.
Using the TLFB-G to measure gambling behavior every day provides us with the opportunity to closely monitor the progress of the participants. The amount wagered seemed to be rather unevenly distributed, where short episodes of “binge gambling” were succeeded by longer periods of complete abstinence. During a binge episode, gamblers often spent large sums of money, while little or no money was spent during periods of abstinence. This poses a challenge in terms of how the data should be analyzed statistically, but also what it means clinically. It is not clear at what level of spending participants could be classified as recovered, and whether a sharp decrease or increase in spending is mirrored by a similar change in problem gambling symptoms overall. Put differently, is 5000 SEK lost on gambling five times worse than 1000 SEK lost on gambling in regards to harm and the degree of addiction? There are also many different ways in which to measure the TLFB-G: either as money lost per day, as number of days gambled, as a proportion of days gambled or as money lost on gambling days. This relates to an overarching issue on how we define recovery and harm in association with gambling and what it means for problem gambling. When has someone recovered and when does gambling become harmful? While it is possible to provide a general answer to these questions such as cut-off scores on NODS or number of DSM-5 criteria, it is a challenge to pinpoint the exact thresholds for recovery and harm when it comes to its relation to time or money spent on gambling.
As for many other studies on interventions for problem gambling (Melville et al.
2007), attrition was high, especially among the gamblers. Since this was merely a pilot study, the initial number of participants was already small, making it even harder to draw conclusions, as is illustrated by some of the large CIs. It is difficult to know whether those who dropped out differed in any substantial way from the completers. They could possibly have had more ongoing gambling problems, but they could also have felt that they did not need treatment any longer. Comparisons between included participants and prospective participants that signed up but never initiated treatment indicate that there were differences between these two groups. The latter group had significantly worse gambling problems and had a higher score on AUDIT; a measure of alcohol consumption and harm from alcohol use. This could indicate that this type of treatment appeals less to gamblers with more complex problems. The study’s design, where gamblers and CSOs are recruited pairwise, is modeled to increase gambler retention in treatment, given results from earlier studies identifying CSO involvement as positive for treatment participation and retention. However, there may also be a risk that unmotivated gamblers are pressured into treatment by their CSO, which could possibly make the gambler even less motivated in terms of participating in treatment.
Measures for anxiety and distress indicate that gamblers in both groups, as well as the CSOs in the BCT group, improved during the course of treatment. The CSOs in the CBT group, however, did not seem to improve substantially during the course of treatment. This seems logical since they did not receive any intervention, and could provide indirect support for the possible benefit of CSO directed treatment. However, this also indicates that the CSOs do not necessarily benefit solely from improvements in the gambler’s gambling behavior. Given that the main reason for seeking treatment is to target the gambling problems of the gambler, this may either be regarded as somewhat surprising, or as an indication of the severe negative effects that gambling may exert on significant others. While the exact reasons for these results remain to be uncovered, one interpretation is that the CSOs of problem gamblers need interventions tailored specifically for them, regardless of how the gambling problem develops. One reason for this could be that CSO-tailored interventions give insights into how problem gambling develops, how it is maintained and how it can be overcome. Since many problem gamblers have a history of alternating periods of abstinence and relapse, the CSOs may not feel sufficiently reassured by recent changes in gambling behavior alone.
Feasibility and Acceptability
The participants receiving treatment generally rated it very favorably, especially those who received BCT. While all BCT gamblers but one gave the treatment the highest rating, the CBT gamblers had a slightly more mixed experience. A few of them mentioned their own lack of self-discipline as a reason for not being completely satisfied with the program. This could be interpreted as an advantage for the BCT condition, where CSO involvement could encourage treatment engagement. The CSOs who did not receive any treatment were generally less positive and some of them expressed disappointment at not receiving any modules or therapist support. However, as with any intervention affected by attrition, and the fact that this is a limited pilot study, these results should be interpreted with caution. Participants who were not satisfied with the treatment program could also be less motivated to complete the treatment and to fill out the treatment evaluation form.
The therapists were also generally positive about the study design regarding involving both the gambler and the CSO in the treatment. However, they both concluded that a substantial amount of their time was used up in reminding patients to complete modules and in arranging and re-arranging scheduled telephone calls. They also remarked on how some of the CSOs seemed considerably more motivated to participate than the gambler did, which became obvious in the screening process. This posed a particular challenge when they were randomized to the CBT condition where the CSOs did not receive any treatment. Some of these issues relate to the feasibility of conducting Internet-based interventions targeting more than one participant rather than to the feasibility of interventions targeting gambling.