This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Internet-based interventions targeted at the most at-risk gamblers could reduce the treatment gap for addictive disorders. Currently, no clinical trial has included non–treatment-seeking patients who have been recruited directly in their gambling environment. This study was the first exclusively Internet-based randomized controlled trial among non–help-seeking problem gamblers with naturalistic recruitment in their gambling environment.
The aim of this study was to assess the efficacy of three modalities of Internet-based psychotherapies with or without guidance, compared to a control condition, among problem gamblers who play online poker.
All active poker gamblers on the Winamax website were systematically offered screening. All problem poker gamblers identified with a Problem Gambling Severity Index (PGSI) score of ≥5 were eligible to be included in the trial. Problem gamblers were randomized into four groups: (1) waiting list (control group), (2) personalized normalized feedback on their gambling status by email, (3) an email containing a self-help book to be downloaded with a Cognitive Behavioral Therapy (CBT) program without guidance, and (4) the same CBT program emailed weekly by a trained psychologist with personalized guidance. Efficacy was assessed based on the change in PGSI between baseline and 6 weeks (end of treatment) or 12 weeks (maintenance) and supported by player account-based gambling data automatically collected at the three time points.
All groups met high attrition rates (83%), but the group with guidance had a significantly higher dropout rate than the other three groups, including the control group. Although all groups showed some improvement, with a mean decrease of 1.35 on the PGSI, no significant difference in efficacy between the groups was observed. One-third of the problem gamblers fell below the problem gambling threshold at 6 weeks.
Guidance could have aversively affected problem gamblers who had not sought help. Despite the lack of significant difference in efficacy between groups, this naturalistic trial provides a basis for the development of future Internet-based trials in individuals with gambling disorders. Comorbidities, natural course of illness, and intrinsic motivation seem to be critical issues to consider in future designs.
ANSM 2013-A00794-41
Despite guidelines for responsible gambling standards [
Targeted Internet-based interventions among the most at-risk online gamblers could enhance the efficacy of existing measures and broaden the range of existing sources of help [
A fully Internet-based randomized controlled trial is an emerging design that could be particularly pertinent and acceptable in this population, for whom the Internet is the medium of their addictive behavior [
The aim of this study was to assess the efficacy of treating online problem gamblers, in particular poker players, with three Internet-based psychotherapy modalities with or without guidance: (1) personalized normalized feedback on their gambling status by email, (2) a self-help book to be downloaded with a CBT program with no guidance, and (3) the same CBT program emailed weekly by a trained psychologist with personalized guidance. Our first hypothesis was that the three Internet-based modalities would be more efficient than the control condition. Our second hypothesis was that less severe problem gamblers would benefit more than more severe ones from modalities requiring less time investment and with no guidance, that is, the personalized normative feedback and the CBT program with no guidance. Our third hypothesis was that more severe gamblers would benefit more from the heaviest intervention requiring more time investment and with guidance.
All active poker gamblers on the poker gambling service provider, Winamax, were offered screening for problem gambling, that is, a score of ≥5 on the Problem Gambling Severity Index (PGSI). Those identified as problem poker gamblers were proposed to be included in an exclusively Internet-based randomized controlled trial.
Subjects were considered for enrollment when they started a poker session during the inclusion period from November 13, 2013, to January 16, 2014 (subjects could be included only once). The additional inclusion criteria were age ≥18, completion of registration (ie, an identification card was sent to Winamax to confirm their age), and registration for ≥30 days. On the day after players first opened a poker session during the inclusion period, they were automatically sent an email with a link that they were invited to click and were redirected to an online survey platform hosted by Winamax, where data were collected and then provided to the investigator. Thus, it was a closed survey. The email presented the research team and the perspectives of the study, namely to help identify problem gamblers and offer them an intervention to control their gambling behavior. Before completing the online process, the subjects read a page that contained clear information about the study phase in which they were to be included. The subjects had to read the page to confirm that they agreed and that they understood the study to proceed to the survey. Gamblers were invited to complete the Canadian Problem Gambling Index’s PGSI. If they scored ≥5, they were informed by mail that their scoring could mean they had a gambling problem. They were then invited for inclusion in the randomized control trial with all information on the randomization process and the allocation groups. As a result, the included gamblers were poker gamblers with problem gambling—we use the term “problem poker gamblers” in the paper.
The included gamblers were randomized into four groups following a computer-based randomization process: (1) waiting list (control group), (2) personalized normalized feedback on their gambling status by a preprogrammed email with blanks automatically filled based on the PGSI score, (3) email with a self-help book in a PDF file to be downloaded, containing a CBT program with no guidance, and (4) the same CBT program emailed weekly by a trained psychologist with personalized guidance. The gamblers randomized to the waiting list received an email explaining that they were registered on a waiting list and could contact the research team at the end of the 12-week period if they wanted to benefit from one of the three treatment modalities. The personalized normalized feedback group received an email returning their PGSI score, explaining the corresponding gambling category, and gave prevalence information corresponding to this category, derived from the only available French prevalence data at that time [
All emails regarding screening, recruitment, intervention, and assessment, except for Group 4 intervention (ie, exchange with the therapist) were automatically generated by the Winamax platform.
Subject consent was obtained as required by local French laws and regulations. The trial has been prospectively registered to the French Medicine Agency, ANSM (ID No. RCB: 2013-A00794-41). The study was authorized by the Comité de Protection des Personnes, as is required for medical intervention research in France. The subjects did not receive any compensation for their participation in this study. Subject anonymity was established and maintained throughout the course of the study, except for Group 4, who agreed to share their email addresses to be able to benefit from the guidance. Before completing the online process, the subjects read a page that contained clear information about the study. The subjects had to read the page to confirm that they agreed and that they had understood the study to proceed to the survey. The subjects received no incentive to respond.
The sample size for the interventional phase was 992 patients, assuming a standard deviation of PGSI of 8.4, an expected delta of 3 points, and a dropout rate of 50%. We systematically recruited gamblers to be included in the study until we attained the desired sample size for the interventional phase (see
After gamblers opened a gambling session on the Winamax website, they were automatically emailed an invitation for inclusion in the study. The assessment of enrolled subjects was completed exclusively online. Player account–based gambling data were prospectively collected automatically at baseline, 6 weeks, and 12 weeks by Winamax and were then retrospectively extracted for the 30-day period before the inclusion day, before week 6 and before week 12. We collected the PGSI at the two endpoints by additional email invitations. Data management and analysis were conducted by the authors. Winamax was contractually commissioned to collect the data, but the authors analyzed the data independently of Winamax.
The only additional data collected online involved the PGSI [
Flow chart.
Basic sociodemographic and routinely recorded data were extracted from the Winamax player account–based dataset at the three time points. The gambling variables used have previously been reported to be good indicators of problem and pathological gambling (7), and this information is routinely recorded by Winamax. We selected the gambling variables based on ease of their extraction from the Winamax player account–based dataset and according to their reproducibility among other online gambling providers. These criteria were multitabling (playing on multiple tables in the same time) in the past 30 days (yes/no), compulsivity (yes/no) (defined by at least 3 deposits in a period of 12 hours), amount of total deposit in the past 30 days (euros) (an initial deposit is required upon opening the gambling account, which implies that some gamblers could have a null deposit during the study period), mean loss per gambling session including the rake (euros), loss in the past 30 days including the rake (euros), total stakes (euros), number of gambling sessions in the past 30 days, number of gambling days in the past 30 days, and time gambled (hours) in the past 30 days. Two criteria could not be correctly assessed because of technical limits related to a lack of automatic disconnection from the app on some wireless devices, particularly smartphones and tablet computers: time gambled (hours) in the past 30 days and number of gambling days in the past 30 days. We therefore excluded these two criteria from the analysis.
Included and non-included gamblers were compared using Student’s
Included and non-included screened problem gamblers characteristics and comparisona.
|
Included (n=1122) | Non-included (n=1441) |
|
Age in years, mean (SD) | 34.7 (10.1) | 32.55 (9.6) | <.001 |
Gender, male, n (%) | 1033 (92.07%) | 1339 (92.92%) | .4 |
Deposal €, mean (SD) | 293.4 (805.7) | 212.9 (638.3) | .01 |
Total loss in €, mean (SD) | 180.6 (740.3) | 77.3 (652.7) | .0002 |
Mean loss per gambling session in €, mean (SD) | 4.0 (16.7) | 2.7 (13.8) | .04 |
Total stake in €, mean (SD) | 1736.1 (5662.6) | 1588.1 (11675.8) | .7 |
Number of gambling sessions, mean (SD) | 61.8 (78.6) | 58.7 (73.1) | .3 |
Multitabling, yes, n (%) | 898 (80.04%) | 1110 (77.03%) | .06 |
Compulsivity, yes, n (%) | 91 (8.11%) | 99 (6.87%) | .23 |
PGSI total score (on 27), mean (SD) | 9 (4.7) | 7.73 (3.9) | <.001 |
aStudent’s
Very few gamblers completed the PGSI assessment at the end of the interventions (188/1122, 16.76%; see
Dropout rate in the randomization groups at 6 and 12 weeks.
Intervention group | Dropout on PGSI at 6 weeks, n (%) | Dropout on PGSI at 12 weeks, n (%) |
Waiting list (n=264) | 199 (75.4)a | 219 (83.0) |
Feedback email (n=293) | 228 (77.8)a | 252 (86.0) |
Self-help CBT book (n=264) | 220 (83.3)a | 245 (92.8) ( |
Weekly emailed CBT (n=301) | 287 (95.3)a | 293 (97.3)a |
a
The mean PGSI total score decreased significantly between baseline and 6 weeks in the overall sample and within each group, except in the CBT group with weekly emailed guidance (Group 4): mean change -1.35 points (SD 3.8) for the overall sample (see
PGSI variation score between baseline and 6 weeks by intervention group and by severity.
Intervention group | Mean (SD) | ||
All | 5≤ PGSI <8 (n=98) | 8≤ PGSI (n=91) | |
Waiting list (n=65) | -1.32 (3.1) | -0.75 (2.4) | -2.03 (3.7) |
Feedback email (n=65) | -1.06 (4.1) | -0.74 (2.2) | -1.43 (5.5) |
Self-help CBT book (n=44) | -1.73 (4.2) | -0.48 (3.3) | -3.1 (4.7) |
Weekly emailed CBT (n=14) | -1.64 (3.9) | -1.25 (3.6) | -1.8 (4.2) |
No significant difference was found in the other gambling variables between the groups at 6 and 12 weeks (
Gambling variables changes at 6 weeks by intervention group (n=1122).
Gambling variables (last 30 days) | Minimum | Median | Maximum | Mean (SD) or % |
|
|
|
||||||
|
Waiting list | -5180.00 | 0.00b | 1970.00 | -9.52 (436.9) | .7 |
|
Feedback email | -3206.00 | 0.00 b | 2487.00 | -69.10 (477.4) | .01 |
|
Self-help CBT book | -4774.00 | 0.00 b | 9540.00 | 33.03 (903.8) | .6 |
|
Weekly emailed CBT | -3645.00 | 0.00 b | 4300.00 | -11.99 (474.4) | .70 |
|
||||||
|
Waiting list | -5210.95 | 0.48 b | 3348.89 | -18.93 (693.7) | .7 |
|
Feedback email | -2377.80 | -0.06 b | 7518.17 | 89.64 (953.1) | .11 |
|
Self-help CBT book | -5727.30 | 3.26 b | 13855.52 | 99.27 (1146.0) | .16 |
|
Weekly emailed CBT | -2614.05 | 0.00 b | 12285.52 | 93.83 (988.3) | .1 |
|
||||||
|
Waiting list | -206.35 | 0.01 b | 91.12 | -1.90 (23.1) | .2 |
|
Feedback email | -69.02 | -0.04 b | 174.03 | 1.51 (16.2) | .11 |
|
Self-help CBT book | -162.36 | 0.02 b | 127.93 | 1.65 (22.3) | .2 |
|
Weekly emailed CBT | -54.54 | 0.00 b | 148.02 | 1.74 (15.8) | .06 |
|
||||||
|
Waiting list | -15027.92 | -10.88 b | 12126.09 | -77.70 (2594.7) | .6 |
|
Feedback email | -29009.16 | -4.00 b | 42677.77 | -153.72 (4718.2) | .6 |
|
Self-help CBT book | -39197.70 | -30.50 b | 129684.50 | 588.24 (9967.4) | .3 |
|
Weekly emailed CBT | -22094.43 | -2.00 b | 55041.05 | 400.32 (4741.9) | .14 |
|
||||||
|
Waiting list | -448.00 | -3.00 b | 277.00 | -5.18 (70.3) | .23 |
|
Feedback email | -255.00 | -2.00 | 221.00 | -4.85 (50.9) | .10 |
|
Self-help CBT book | -452.00 | -5.50 | 403.00 | -9.25 (70.9) | .04 |
|
Weekly emailed CBT | -563.00 | -3.00 | 362.00 | 0.69 (68.3) | .9 |
|
||||||
|
Waiting list | - | - | - | -14% | <.001 |
|
Feedback email | - | - | - | -6% | .03 |
|
Self-help CBT book | - | - | - | -13% | <.001 |
|
Weekly emailed CBT | - | - | - | -9% | .001 |
|
||||||
|
Waiting list | - | - | - | -0.4% | .83 |
|
Feedback email | - | - | - | -2.4% | .2 |
|
Self-help CBT book | - | - | - | -1.2% | .6 |
|
Weekly emailed CBT | - | - | - | 2.3% | .09 |
|
||||||
|
Waiting list (n=65) | -13.00 | -1.00 | 6.00 | -1.32 (3.1) | <.001 |
|
Feedback email (n=65) | -17.00 | -1.00 | 8.00 | -1.06 (4.1) | .04 |
|
Self-help CBT book (n=44) | -17.00 | -2.00 | 8.00 | -1.73 (4.2) | .004 |
|
Weekly emailed CBT (n=14) | -11.00 | -1.00 | 6.00 | -1.64 (3.9) | .11 |
aA negative value is a worsening, and a positive value is an improvement for the participant.
bAs the variance is huge on the monetary variables, median is more meaningful than the mean.
The self-help book group had the highest responder rate: 15% (10/65), 17% (11/65), 25% (11/44), and 14% (2/14) in Groups 1, 2, 3, and 4, respectively (no significant difference). Age was the only variable predictive of responder status (
We found no significant difference in the financial subscore of the PGSI between the groups at 6 and 12 weeks.
This randomized controlled trial among non–treatment-seeking online problem poker gamblers showed no between-group difference of efficacy of Internet-based interventions compared to placebo. The group with guidance had the highest dropout rate.
Given the low treatment-seeking status in problem gambling, we ambitiously chose to propose accessing the health care system by proactively inviting problem gamblers screened in their gambling environment to participate in exclusively Internet-based interventions. However, we found limits to the acceptability of these interventions. The dropout rate was very high, although Internet-based randomized trials usually have high dropout rates [
The proposed interventions could lack an intrinsic motivational component owing to their non-face-to-face nature. Learning during skills-based psychosocial treatments has been shown to be influenced by the intrinsic motivating properties of the treatment context in mental disorders [
Some trials have proposed financial compensation to lower the dropout rate, through increasing the extrinsic motivation. However, these methods are questionable in patients with a gambling disorder, for whom a monetary gain could interfere with gambling behavior [
Khadjesari has already indicated that online trial designs evaluate access to therapeutic material rather than engagement in using it [
Gambling variables changes at 12 weeks by intervention group (n=1122).
Gambling variables (last 30 days) | Minimum | Median | Maximum | Mean (SD) or n (%) |
|
|
|
||||||
|
Waiting list | -5180.00 | 0.00 | 3000.00 | -17.67 (573.55) | .03 |
|
Feedback email | -12800.00 | -3.00 | 2150.00 | -136.69 (884.45) | <.001 |
|
Self-help CBT book | -4058.00 | -10.00 | 9300.00 | -33.41 (801.36) | <.001 |
|
Weekly emailed CBT | -2600.00 | -10.00 | 2710.00 | -54.22 (365.61) | <.001 |
|
||||||
|
Waiting list | -4990.51 | 2.31 | 8333.44 | -3.30 (852.54) | .12 |
|
Feedback email | -2315.58 | 0.68 | 26580.10 | 193.78 (1792.40) | .10 |
|
Self-help CBT book | -4051.70 | 8.00 | 181410.22 | 737.06 (11252.29) | .02 |
|
Weekly emailed CBT | -2304.00 | 2.00 | 10348.32 | 63.47 (803.72) | .14 |
|
||||||
|
Waiting list | -1062.72 | 0.00b | 176.63 | -5.64 (70.02) | .9 |
|
Feedback email | -87.89 | 0.01 | 227.21 | 0.68 (19.09) | .5 |
|
Self-help CBT book | -134.81 | 0.08 | 3003.66 | 11.90 (187.26) | .2 |
|
Weekly emailed CBT | -73.15 | 0.03 | 96.21 | 0.72 (14.10) | .3 |
|
||||||
|
Waiting list | -20160.14 | -63.50 | 90062.29 | 143.79 (6345.74) | .007 |
|
Feedback email | -54205.38 | -33.21 | 26282.27 | -318.87 (4284.30) | .002 |
|
Self-help CBT book | -62492.55 | -36.84 | 50231.28 | -5.36 (7083.45) | .003 |
|
Weekly emailed CBT | -20276.97 | -46.50 | 122346.52 | 576.16 (8490.70) | <.001 |
|
||||||
|
Waiting list | -448.00 | -7.00 | 285.00 | -11.91 (71.56) | .005 |
|
Feedback email | -262.00 | -8.00 | 318.00 | -5.25 (66.27) | <.001 |
|
Self-help CBT book | -381.00 | -6.00 | 356.00 | -8.96 (71.42) | .001 |
|
Weekly emailed CBT | -501.00 | -8.00 | 365.00 | -6.46 (71.46) | .002 |
|
||||||
|
Waiting list |
|
|
|
-18% | <.001 |
|
Feedback email |
|
|
|
-13% | <.001 |
|
Self-help CBT book |
|
|
|
-18% | <.001 |
|
Weekly emailed CBT |
|
|
|
-17% | <.001 |
|
||||||
|
Waiting list |
|
|
|
0.0% | 1 |
|
Feedback email |
|
|
|
-3.1% | .07 |
|
Self-help CBT book |
|
|
|
-0.8% | .7 |
|
Weekly emailed CBT |
|
|
|
1.7% | .3 |
|
||||||
|
Waiting list (n=45) | -10.00 | -3.00 | 0.00 | 1.00 (12.00) | .02 |
|
Feedback email (n=41) | -14.00 | -3.00 | -1.00 | 0.00 (6.0) | <.001 |
|
Self-help CBT book (n=19) | -9.00 | -4.00 | -3.00 | 0.00 (7.0) | .09 |
|
Weekly emailed CBT (n=8) | -8.00 | -2.50 | -1.00 | 0.50 (5.0) | .4 |
aA negative value is a worsening, and a positive value is an improvement for the participant.
bAs the variance is huge on the monetary variables, median is more meaningful than the mean.
Even if the sample size has been calculated for the primary outcome (PGSI) for which we endorsed a substantial loss to follow-up, the lack of efficacy is supported by the lack of between-groups difference in the secondary criteria, in a very large sample (three time larger per intervention group than the Hodgins’ trial [
There are several explanations for the lack of efficacy of the Internet-based CBT interventions in this trial. First, this trial included non-help-seeking problem gamblers recruited in their gambling environment with no initial involvement in treatment. It has been shown that problem gamblers with higher external motivation for change were less likely to be farther along the stage of change continuum [
In regard to the less severe patients, we chose a conservative threshold of 5 for the PGSI, whereas many trials include patients with a threshold of 3. However, the PGSI is a screening instrument and does not provide a clear diagnosis of gambling disorder. This threshold choice could have biased the results because low-risk or low-problem gamblers have few reasons to change their behavior.
It is also possible that the program is not effective in the gambling population selected in this trial. The program itself could present limits if proposed to any problem gambler; the proposed design is deeply naturalistic, and there were no exclusion criteria except for the legal age limit. Additional psychiatric conditions could have limited the impact of the proposed program, namely, depression and anxiety are frequent comorbidities in problem gamblers and could interfere with work on the cognitive distortions as proposed in our program [
This first Internet-based randomized controlled trial among non–help-seeking online problem poker gamblers showed a lower acceptability of the modality including guidance compared to the other modalities including placebo. This was possibly due to an aversive effect of guidance in this particular population. We found no significant difference in efficacy between the Internet-based CBT modalities, with or without guidance, compared to the control condition. This naturalistic trial provides a basis for developing future Internet-based trials for individuals with gambling disorders. The natural course of gambling disorders is still poorly documented, and spontaneous changes are a challenge for future assessment of therapeutic interventions. Although Internet-based CBT may enhance access to treatment, it should include intrinsic motivational components to increase engagement in treatment.
cognitive behavioral therapy
Problem Gambling Severity Index
The study was funded by Winamax, a commercial online gambling service provider.
MLT and ML received funds for this study from Winamax. Other authors have no conflicts of interest to declare.