Factors Influencing Treatment Outcomes in a Randomized Clinical Trial for Gambling Disorder: A Secondary Analysis

Abstract

Introduction

Sociodemographic background factors, gambling game type, depressive symptoms, and comorbid substance use have been associated with gambling disorder, but the associations of these factors with treatment outcomes remain unclear. This study examined background factors in participants of a clinical trial, identifying factors which were associated with treatment outcomes such as gambling urge and quality of life.

Methods

This was a secondary analysis of a single-center 12-week randomized placebo-controlled clinical trial investigating the effects of as-needed intranasal naloxone combined with psychosocial support in the treatment of gambling disorder. Independent variables were analyzed with linear mixed models, with analyses both including and excluding treatment allocation.

Results

Sociodemographic factors or game types were not associated with treatment outcomes. Adherence to trial medication and high self-perceived readiness to change gambling behavior were associated with greater reduction in gambling urge and severity, and greater increase in gambling self-efficacy and quality of life. Higher scores for depressive symptoms were associated with more severe gambling and a slightly greater decrease in gambling urge, as well as a greater increase in quality of life. When included as a nuisance factor, treatment allocation abolished most statistically significant results.

Conclusion

Participants with high medication adherence had improved treatment outcomes compared to participants with low medication adherence, possibly representing greater motivation and commitment to treatment. More severe depressive symptoms were associated with a greater reduction in gambling urge but not gambling severity.

Introduction

Gambling disorder is recognized as a behavioral addiction in both the DSM-5 and ICD-11 [1, 2]. A global incidence of up to 5.8% for problem gambling underscores its status as a public health concern [3]. The disorder leads to various harms, including financial difficulties, reduced well-being, and strained social relationships, affecting not just the gambler but also their significant others and society [4, 5]. Factors linked to gambling disorder include male gender, low socioeconomic status, lower education, gambling history, and comorbid psychiatric and substance use disorders [6, 7].

Evidence-based treatments primarily involve cognitive-behavioral therapy and motivational interventions [6, 8, 9]. A Cochrane review by Dowling et al. [10] reported preliminary support for opioid antagonists but emphasized the low certainty of evidence, small sample sizes, and methodological limitations across available trials. More recent and comprehensive analyses by Ioannidis and colleagues provide additional context, and Ioannidis et al. [11] demonstrated substantial placebo effects in pharmacological gambling disorder RCTs, with effect sizes exceeding 1.0 in placebo arms and measurable nocebo-related dropout that varied by medication class. Ioannidis et al. [12] using network meta-analysis, found that nalmefene and naltrexone show moderate beneficial effects on gambling severity and quality of life (QoL), whereas naloxone has limited supportive evidence. Furthermore, as opioid antagonists can evoke expectations of adverse effects or carry stigma, nocebo responses may contribute to dropout or reduced adherence, factors relevant to the interpretation of pharmacological trial outcomes. These findings contextualize individual negative or null results, such as those observed in our previous randomized trial of as-needed intranasal naloxone [13].

In addition to pharmacological considerations, several psychosocial and clinical characteristics may influence treatment responses in gambling disorder. The male-to-female ratio among treatment-seeking gamblers is 2:1, with potential gender differences in treatment outcomes [6, 14, 15]. Depressive symptoms correlate with problem gambling, where lower depression is linked to better outcomes [16–18]. Comorbidities such as smoking and alcohol use disorder (AUD) are common, with lower alcohol consumption potentially improving treatment outcomes [17]. Fast-paced electronic gaming machines (EGMs) and table games are more addictive and often favored by those with gambling disorder, affecting severity [19, 20]. However, their influence on treatment outcomes is unclear. Readiness and motivation, particularly autonomous motivation, are vital for successful treatment [21, 22], with self-efficacy also playing a mediating role [23]. However, research on these factors’ impacts on gambling disorder treatment remains limited.

In our previous trial on as-needed intranasal naloxone for reducing gambling urges, results were inconclusive, showing no significant difference between naloxone and placebo groups despite some improvements [13]. Given recent evidence on the magnitude of placebo and nocebo responses in gambling disorder pharmacotherapy, traditional assumptions about medication effects warrant reconsideration [11, 24]. Methodological differences may obscure benefits among participant subgroups [25]. By analyzing a wider range of variables, we aimed to uncover potentially overlooked associations. This study extended the findings of the previous study by investigating the associations of other explanatory variables influencing treatment outcomes with the goal of identifying subgroups of participants who may have benefited from the naloxone medication as discussed in the original trial publication [13].

Aims and Hypotheses

In this secondary analysis, we examined if sociodemographic background factors, depressive symptoms, alcohol use, and smoking status affect gambling treatment outcomes in this clinical trial. Furthermore, we assessed if having more severe gambling disorder symptoms at baseline, game type and medication adherence during the trial influence treatment outcomes. Analysis was conducted by pooling the naloxone and placebo groups together in addition to comparing the outcomes based on participant group assignment. The aim of this exploratory analysis was to identify factors which predict and mediate treatment success.

Methods

Trial Outline

This trial was a randomized double-blind trial conducted at the Institution for Health and Welfare in Helsinki, Finland. The clinical phase ran from February 2018 to August 2019.

Principal eligibility criteria included ages 18–75, male or female, South-Oaks Gambling Screen-Revised (SOGS-R) score ≥5 at screening, moderate or severe gambling disorder per DSM-5 criteria, and a minimum of 4 weeks since any previous gambling disorder treatment. A complete list of inclusion and exclusion criteria has been reported earlier [13]. A CONSORT Checklist for this study is available in the supplementary material (for all online suppl. material, see https://doi.org/10.1159/000550527).

Study Treatments

Participants were randomized in a 1:1 ratio using a computer with permuted block randomization by the electronic data capture system. The study drug was a 40-mg/mL naloxone nasal spray, with saline as the indistinguishable comparator. Each dose of 0.1 mL (4 mg naloxone) was to be sprayed in one nostril up to 4 times daily in response to gambling urges or high-risk situations. The investigational medicinal product (IMP) was manufactured by Sharp Clinical Service (UK) Limited and donated by Opiant Pharmaceuticals, Santa Monica, CA, USA, without terms or conditions. All participants received psychosocial support from trained psychologists, utilizing motivational interviewing and the CBT-based self-help manual “Defeating Problem Gambling” (D. Hodgins and Makarchuk, 2002) to enhance treatment compliance and medication adherence.

Study Procedures

Persons interested in participating completed an online SOGS-R questionnaire, with a score of ≥5 indicating eligibility [26]. During the screening, participants detailed their medical and gambling histories, underwent study assessments, and were instructed on the IMP use and e-diary completion. They recorded their gambling frequency, expenditure, IMP use, and any adverse events daily during the study. The trial lasted a maximum of 15 weeks, with a 12-week treatment period. Participants attended four visits at the study clinic (screening, baseline, week 6, and week 12) and received three phone calls (week 3, week 9, and a week 14 follow-up).

Study Measures

The trial aimed to determine if as-needed intranasal naloxone adherence reduces gambling urges, measured by the Gambling Symptom Assessment Scale (G-SAS) [27]. The G-SAS is scored on a scale of 0–48, with higher scores indicating more severe symptoms. Other assessments included DSM-5 criteria for gambling disorder severity (severity: mild 4–5 criteria met, moderate 6–7 criteria met, severe 8–9 criteria met) [1], the Gambling Abstinence Self-Efficacy Scale (GASS, 21 items scored 0–5 with higher scores indicating greater confidence in gambling abstinence) [28], the Alcohol Use Disorder Identification Test (AUDIT, 10 questions scored 0–4; interpretation: 0–7 low risk, 8–15 increasing risk, 16–19 higher risk, 20+ indicating possible dependence) [29], the Montgomery-Åsberg Depression Rating Scale (MADRS, 10 items scored 0–6; scores 7–19 mild depression, 20–34 moderate depression, 35–60 severe depression) [30], and the EUROHIS-QOL 8-item index for QoL (8 items scored on a five-point Likert scale with a score range of 8 to 40) [31]. Safety was evaluated through recorded adverse events, and participants rated their perceived success (0 = not at all successful to 10 = extremely successful) in achieving their treatment goal. Smoking status was assessed with a yes/no question at baseline.

Measurements were taken at baseline, week 3, week 6, week 9, and week 12, except for QoL, gambling severity, and alcohol use, which were measured at baseline, week 6, and week 12. At baseline, participants identified a treatment goal (abstinence or reduced gambling) and assessed their readiness to change using a 10-cm visual analogue scale (VAS) [32]. Based on readiness scores, participants were divided into three groups. Participants were classified by their primary game type based on the games they played most in the past month, including lottery-type games, betting games, and online gambling; casino games were excluded due to no reported participation.

Medication adherence was measured through self-reported e-diary data, comparing reported gambling behavior with IMP use. Days were classified as adherent or non-adherent based on adherence criteria, resulting in three adherence groups: low (<50%), medium (50–75%), and high (75–125%) adherence. Participants who exceeded 100% of intended doses were also noted. Groups were further divided by treatment allocation (naloxone/placebo), totaling six groups, with cut-off percentages chosen by the authors.

Statistical Considerations

Associations of explanatory variables with the mean profile of response variables were estimated using a linear mixed model, along with their associations at week 12. The marginal mean profile is the estimated average curve of the considered total sum scores (e.g., total sum scores of the G-SAS variable) when the estimation is done by the linear mixed effect model. The mean level at the week 12 is the estimated average value of the total sum scores (e.g., G-SAS variable) at the week 12 when the estimation is done by the linear mixed effect model. The mean level at week 12 was chosen as the decision-making estimand to emphasize the results at the end of the study. Explanatory variables achieving statistical significance (p < 0.05) for week 12 were further examined. The first analysis reports testing results (p values; Table 2) regarding the statistical significance of the considered explanatory variable when there was only one considered explanatory variable in the linear mixed model in addition to the quadratic time variable. The second analysis reports testing results (p values; Table 3) regarding the statistical significance of the considered explanatory variable when there were both the treatment variable and the considered explanatory variable in the linear mixed model in addition to the quadratic time variable. Hence, reported p values in Table 3 are measuring the additional statistical significance of the considered explanatory variable on the response variable when the effect of the treatment allocation variable and the quadratic time variable are already taken account.

In this study, the considered variables contained relatively few missing values, hence multiple imputation was not applied, and thus, missing values were assumed to be missing at random. Prior to the study, the sample size calculation was carried out by setting the effect size to the 0.7 level and by assuming that the standard deviation for the total score would be 5.502 and the dropout rate would be 30%. A detailed description of statistical methods used is provided in the online supplementary statistical.

Results

Sociodemographic Background Factors

One participant out of 127 withdrew consent before randomization, resulting in 126 participants (62 naloxone and 64 placebo). Of these, 106 (84%) completed the trial. Non-completion was due to consent withdrawal (12 subjects, 9.5%) and loss to follow-up (8 subjects, 6.3%), with higher rates in the placebo group (23.4% vs. 8.1% for naloxone). Nine participants who discontinued before week 6 were excluded from analysis due to having little to no recorded data points, leaving 118 participants.

The analyzed groups were representative of typical treatment-seeking gamblers and mostly followed a normal distribution, except for the age of female participants (p = 0.047), likely due to the small sample size. Other differences were not statistically significant. The median age was 43.5 years in the naloxone group and 45.0 years in the placebo group. Sociodemographic factors are presented in Table 1.

Table 1.

Sociodemographic background factors of trial participants

	Naloxone (N = 60)	Placebo (N = 58)	Combined (N = 118)
Age (median: Q1, Q3)	43.5 (30.0, 54.0)	45.0 (34.5, 61.5)	44.0 (32.0, 59.0)
Male, n (%)	44 (73.3)	38 (65.5)	82 (69.5)
Female, n (%)	16 (26.7)	20 (34.5)	36 (30.5)
Marital status, n (%)
Single	18 (30.0)	14 (24.1)	32 (27.1)
Married/cohabiting	31 (51.7)	33 (56.9)	64 (54.2)
Divorced	11 (18.3)	10 (17.2)	21 (17.8)
Widow/widower	0	1 (1.7)	1 (0.8)
Living status, n (%)
Alone	25 (41.7)	24 (41.4)	49 (41.5)
With family	33 (55.0)	34 (58.6)	67 (56.8)
With friends/rooming house	2 (3.3)	0	2 (1.7)
Education, n (%)
Up to lower secondary	15 (25.0)	11 (19.0)	26 (22.0)
Vocational qualification	33 (55.0)	31 (53.4)	64 (54.2)
University degree	12 (20.0)	16 (27.6)	28 (23.7)
Employment status, n (%)
Entrepreneur/professional expert	5 (8.3)	1 (1.7)	6 (5.1)
Office worker/clerk	32 (53.3)	34 (58.6)	66 (55.9)
Unemployed	7 (11.7)	2 (3.4)	9 (7.6)
Pensioner	11 (18.3)	18 (31.0)	29 (24.6)
Student	3 (5.0)	3 (5.2)	6 (5.1)
Other	2 (3.3)	0	2 (1.7)
Smoking, n (%)
Never	17 (28.3)	21 (36.2)	38 (32.2)
Currently	34 (56.7)	19 (32.8)	53 (44.9)
Formerly	9 (15.0)	18 (31.0)	27 (22.9)
Alcohol use, n (%)
Never	1 (1.7)	1 (1.7)	2 (1.7)
Currently	55 (91.7)	53 (91.4)	108 (91.5)
Formerly	4 (6.7)	4 (6.9)	8 (6.8)
Adherence to IMP, n (%)
Low (<50%)	11 (18.3)	9 (15.5)	20 (16.9)
Medium (50–75%)	21 (35.0)	24 (41.4)	45 (38.1)
High (75–125%)	28 (46.7)	25 (43.1)	53 (44.9)

The sociodemographic background information distribution of the participants included in this analysis. The sociodemographic background information was acquired from the participants during the screening visit.

Summary of Linear Mixed Model Results

The naloxone and placebo group had similar outcomes during the study: both groups had a similar improvement in scores in the gambling-related study measures used. In addition, both groups had a similar reduction in depressive symptoms and a similar increase in QoL scores. When treatment allocation was added as a variable to the linear mixed model, most results were statistically insignificant.

The effects of sociodemographic factors on outcomes are presented in Table 2. Female participants showed a greater increase in gambling self-efficacy during the study than males, and married or cohabiting participants had slightly higher QoL scores than singles. Living with family was also associated with higher QoL compared to living alone. Unemployment and male gender were associated with higher AUDIT scores, while alcohol use status was only significant for AUDIT scores. Tobacco use reported at baseline did not reach significance with any variables.

Table 2.

The test results of the linear mixed models used for testing the associations with the marginal mean profile and mean level at week 12

Response variable	G-SAS		QoL (EUROHIS-QOL)		GASS		Depressive symptoms (DSM-V criteria)		AUD Identification Scale
	p value (mean profile)	p value (week 12)	p value (mean profile)	p value (week 12)	p value (mean profile)	p value (week 12)	p value (mean profile)	p value (week 12)	p value (mean profile)	p value (week 12)
Treatment allocation	0.02	0.48	0.19	0.07	0.52	0.03	0.049	0.633	0.246	0.096
Age	0.07	0.01	0.21	0.13	0.01	0.08	0.094	0.16	0.003	0.074
Sex	0.4	0.23	0.44	0.28	0.06	0.01	0.376	0.502	0.072	0.021
Marital status	0.07	0.37	0	0	0.09	0.01	0.484	0.658	0.097	0.464
Living	0.09	0.33	0.05	0.02	0.53	0.1	0.1	0.231	0.058	0.548
Education level	0.52	0.53	0.53	0.41	0.08	0.44	0.683	0.655	0.346	0.651
Employment	0.54	0.23	0.64	0.54	0.51	0.11	0.017	0.376	0.001	0.011
Readiness to change	0.01	0.01	0.01	0	0.03	0	0.017	0.048	0.275	0.603
Adherence	0	0	0.16	0.09	0.01	0	0.002	0.001	0.316	0.3
Game type	0.07	0.16	0.11	0.15	0.26	0.18	0.326	0.082	0.267	0.092
Money spent	0	0.56	0.02	0.27	0.12	0.57	0.19	0.052	0.307	0.011
Smoking	0.04	0.39	0.47	0.45	0.18	0.07	0.444	0.388	0.004	0.136
Alcohol use	0.67	0.3	0.23	0.07	0.62	0.32	0.115	0.288	0.001	0.005
MADRS	0.05	0.05	0	0	0.38	0.67	0.001	0.018	0.452	0.284

The associations of the explanatory variables with the mean profile of the response variables. “Readiness to change” refers to self-perceived readiness to change one’s gambling behavior (see Section 2.3). Full test results are given in online supplementary Table 4.

MADRS, Montgomery-Åsberg Depression Rating Scale.

When treatment allocation was included in the model, most results became statistically insignificant (Table 3). Some mean profile changes, like living and marital status being associated with gambling urge (p = 0.009 and p = 0.021), were found significant but lost significance at week 12, rendering them clinically insignificant. Test results for mean profiles and levels at week 12 are displayed in Table 3.

Table 3.

The test results of the linear mixed models used for testing the associations with the marginal mean profile and mean level at week 12 with treatment allocation included in the model

Response variable	G-SAS		QoL (EUROHIS-QOL)		GASS		Depressive symptoms (DSM-V criteria)		AUD Identification Scale
	p value (mean profile)	p value (week 12)	p value (mean profile)	p value (week 12)	p value (mean profile)	p value (week 12)	p value (mean profile)	p value (week 12)	p value (mean profile)	p value (week 12)
Age	0.233	0.546	0.31	0.372	0.062	0.318	0.469	0.403	0.223	0.103
Sex	0.272	0.586	0.101	0.54	0.627	0.383	0.326	0.581	0.256	0.041
Marital status	0.023	0.675	0.231	0.627	0.364	0.513	0.358	0.067	0.238	0.097
Living	0.009	0.665	0.296	0.422	0.559	0.047	0.589	0.544	0.547	0.313
Educational level	0.253	0.669	0.488	0.531	0.597	0.062	0.363	0.629	0.436	0.055
Employment	0.011	0.274	0.133	0.04	0.428	0.517	0.574	0.582	0.156	0.374
Readiness to change	0.003	0.692	0.156	0.256	0.036	0.544	0.016	0.603	0.444	0.017
Adherence	0.33	0.408	0.272	0.035	0.163	0.122	0.147	0.281	0.272	0.014
Game type	0.095	0.411	0.005	0.599	0.189	0.619	0.008	0.426	0.648	0.367
Money spent	0.103	0.588	0.117	0.138	0.122	0.282	0.491	0.226	0.005	0.006
Smoking	0.233	0.325	0.518	0.554	0.609	0.55	0.652	0.565	0.204	0.064
Alcohol	0.08	0.503	0.036	0.015	0.322	0.142	0.566	0.608	0.274	0.103
MADRS	0.045	0.349	0.238	0.24	0.608	0.416	0.142	0.538	0.296	0.084

Complete test results are presented in online supplementary Table 5.

Medication Adherence

Participants were divided into three groups based on medication adherence. The high-adherence group had the lowest baseline G-SAS scores and showed a greater decrease in G-SAS at 12 weeks (indicating a greater reduction in gambling symptoms) compared to the other groups, with significant differences between low- and high-adherence groups (mean G-SAS 23.9 vs. 13.6; p < 0.001, G-SAS score range 0–48) and low- and medium-adherence groups (p = 0.003). Results for G-SAS scores in different adherence are presented in Figure 1. The low-adherence group experienced the smallest reduction in gambling severity, significantly differing from medium- (p = 0.004) and high-adherence groups (p = 0.001), and had the smallest increase in self-efficacy, with significant differences noted for medium- (p = 0.05) and high-adherence groups (p = 0.001). The online supplementary Figure 1 presents the Gambling Self-Efficacy Scale results for different adherence groups.

Fig. 1.

Estimated marginal mean profiles of G-SAS scores in groups with different levels of medication adherence. The difference between the low-adherence group compared with the medium- and high-adherence groups was statistically significant (p = 0.003 and p < 0.001, respectively). Estimated marginal fitted values are presented in the online supplementary Table 1.

When further divided by treatment status (naloxone/placebo), differences between groups became insignificant. The G-SAS scores at 12 weeks for the high-adherence naloxone group were 12.18 versus 15.18 for the placebo group (p = 0.28, G-SAS scoring range 0–48). The results are presented in Figure 2. Background information and marginally fitted values for adherence groups by treatment status are detailed in online supplementary Tables 2 and 3, respectively. Higher medication adherence was associated with lower AUDIT scores at week 12 (p = 0.014), though the high-adherence naloxone group had slightly higher AUDIT scores than the placebo group, with minimal changes across all trial groups.

Fig. 2.

Estimated marginal mean profiles of G-SAS scores in groups with different levels of medication adherence. The differences between the naloxone and placebo groups were not statistically significant (p = 0.277 for the corresponding high-adherence groups at week 12). Estimated marginal fitted values are presented in the online supplementary Table 1.

The increase in QoL was greatest in the high-adherence group; however, the difference between the low-adherence group and the high-adherence group was not statistically significant (p = 0.127). The predicted mean values for response variables are shown in the online supplementary Table 1.

Depressive Symptoms

The group with the highest baseline depressive symptoms according to MADRS had a mean G-SAS score of 19.3 at 12 weeks, compared to 12.9 in the lowest MADRS group (G-SAS is scored on a scale of 0–48 with higher scores indicating more severe gambling symptoms). Higher MADRS scores at the baseline predicted a greater decrease in gambling symptoms (p = 0.049) but a smaller decrease in gambling severity (p = 0.018). Additionally, higher MADRS scores were associated with a significant increase in QoL (p < 0.001). Associations of MADRS scores on G-SAS scores are presented in Figure 3 and associations with QoL in online supplementary Figure 2. When exploring associations with treatment allocation included in the model, a significant difference in gambling symptoms in the mean profile was observed (p = 0.045), but not at week 12 (p = 0.502).

Fig. 3.

Estimated marginal mean profiles of G-SAS scores in groups with different levels of depressive symptoms (measured on the MADRS). The difference between the MADRS = 1 and MADRS = 23 groups was statistically significant (p = 0.049). Estimated marginal fitted values are presented in the online supplementary Table 1.

Self-Perceived Readiness to Change Gambling Behavior

Higher baseline readiness to change gambling behavior (VAS) predicted a greater decrease in gambling symptoms, with G-SAS scores at week 12 significantly lower in the highest readiness group (mean scores: 12.7 vs. 23.9; p < 0.001). Greater readiness to change also was associated with reduced gambling severity (DSM-5 criteria) (p = 0.048). Participants with a high readiness to change showed the greatest increase in self-efficacy, while those with low readiness returned to baseline levels (p = 0.003). Higher readiness was linked to better baseline QoL scores and significant posttreatment improvement (p < 0.001), presented in online supplementary Figure 3. When treatment allocation was included in the model, significant associations for readiness were observed regarding gambling symptoms, severity, and self-efficacy, but not for week 12 response variables.

Game Types

We analyzed whether participants’ primary game type predicted a reduction in gambling urge, but no significant differences were found across game types. While game type approached significance for changes in gambling severity when including treatment group allocation (p = 0.08), these differences were not significant at 12 weeks (see Table 3).

Discussion

This study further analyzed which background factors influence treatment outcomes for gambling disorder in a clinical trial setting and whether allocation to the naloxone treatment group was associated with the relationship between these factors and the outcomes. Participants with higher medication adherence experienced a greater reduction in gambling symptoms, irrespective of treatment allocation. Depressive symptoms were linked to more intense gambling symptoms at baseline and a significant reduction in gambling symptoms, but not in gambling severity. Higher self-perceived readiness to change was associated with reduced gambling symptoms and severity, along with improved self-efficacy after treatment. Allocation to the naloxone group did not have any notable associations with the measured outcomes in this study.

When comparing high-adherence naloxone and placebo groups, few statistically significant differences were observed in trial outcomes. The high adherence group had lower Gambling Abstinence Self-Efficacy scores at 12 weeks, a statistically significant difference, but adherence rates did not yield significant results between naloxone and placebo treatment groups. Our trial found no statistical difference between participants using the IMP and those receiving placebo. As participants were divided into multiple subgroups for this analysis (two treatment groups, three adherence groups), achieving statistically significant results is challenging. Opioid antagonists naltrexone and nalmefene have shown promise in gambling disorder treatment, while naloxone could be considered an inferior candidate due to its shorter half-life and lack of supporting evidence [12]. The hypothesis was that the rapid occupation of brain mu opioid receptors by intranasal naloxone, reported in Johansson et al. [33] in response to gambling urge would provide beneficial effects in the treatment of gambling disorder [13]. Future trials may benefit from rigorous adherence monitoring or a regular dosing schedule in addition to careful consideration on which opioid antagonist to investigate.

Despite no clear outcome differences, medication adherence showed positive associations with gambling-related and non-gambling-related outcomes, consistent with previous research linking adherence to better results in opioid antagonists for AUD [34]. Our adherence rate was 44.9%, comparable to 55.89% reported in a previous AUD study [35]. The issue of adherence in as-needed formulations in addictive disorders remains underexplored. Notably, a Cochrane review excluded addiction interventions due to adherence challenges [36].

The placebo effect has significantly influenced pharmacological trials for gambling disorder [13, 37]. Recent reviews indicate a substantial placebo effect, notably in GD trials, influenced by various factors which must be taken into consideration when reviewing results [11, 24]. Fava et al. argued that the traditional placebo model fails to capture the complexities of treatment effects, suggesting that patient engagement and expectations play crucial roles. Nocebo effects related to intranasal naloxone may have played a role in premature patient withdrawal from the study [11].

In our study, participants were supported through regular psychosocial contact, a CBT-based self-help manual, and the use of an e-diary. These elements may themselves have functioned as important interventions that supported recovery or helped participants reach their goals, for example, by fostering greater adherence. While it is true that up to 40% of gamblers may recover naturally, a factor future studies need to account for, our findings suggest that structured psychosocial support, such as coaching participants to recognize and act on urges, can be therapeutic [38]. This combination of psychosocial support (contact online or face-to-face), self-monitoring, and guided self-help may not only enhance engagement but also improve the overall effectiveness of psychological interventions. Clinically, this points to the value of integrating low-threshold, structured self-help tools and ongoing supportive contact into routine practice to strengthen treatment adherence and outcomes.

Pickering et al. stressed the variability and complexity in measuring treatment outcomes for gambling disorder, noting that null results can mask benefits varying by subgroup, advocating for a nuanced approach to efficacy evaluation. The interplay between adherence, the therapeutic process, and the placebo effect underscores the complexity of treatment responses, suggesting a need for adapting traditional models in future research. Overall, insights into medication adherence, the placebo effect, and the therapeutic process can inform and improve treatment approaches for gambling disorder.

Research indicates that sociodemographic factors like lower educational level, living alone, and financial issues are linked to gambling [7]. In this study, we found that these factors generally were not associated with treatment outcomes, though marital status and cohabitation showed some inconsistent positive effects. While most participants were male, consistent with prior studies [39], no gender differences were observed in our sample, aligning with findings that the gender gap in gambling is narrowing [40, 41]. Overall, sociodemographic factors may influence gambling severity but their impact on treatment outcomes appears inconsistent.

Psychosocial methods have been utilized to enhance adherence in clinical trials, including medication management [42], the BRENDA model [43, 44], behavioral family counseling [45], and contingency management [46]. In this trial, tailored psychosocial support aimed to improve medication adherence, but its specific effect is unclear since all participants received support. Participant receptiveness may influence adherence, as those in high-adherence groups might have been more motivated and responsive. While enhanced interventions could improve treatment outcomes, further research is needed to discern whether results depend on the support provided or participant motivation.

Depressive symptoms were associated with higher baseline scores for gambling urge and severity, but they were also linked to a slight reduction in gambling urge, while a smaller reduction in gambling severity was observed. Although statistically significant differences in gambling urge reduction existed, the absolute differences were minimal. Previous studies have found depressive symptoms associated with more severe gambling [16, 18], but the effects of lower depressive symptoms on treatment outcomes have been uncertain [17, 47]. Interestingly, participants with higher depressive symptom scores showed greater increases in QoL, consistent with prior findings [48], but it is important to note that those with severe depression were excluded from the trial. Future clinical trials should assess and address depressive symptoms, as their reduction may improve gambling outcomes and QoL.

In this study, higher self-perceived readiness to change correlated with greater decreases in gambling urge and severity, as well as increased self-efficacy and QoL. Previous research indicates that internal motivation is more effective for addiction treatment than external motivation [22]. Future studies should assess participant motivation more thoroughly at baseline, as those with low internal motivation may benefit from tailored approach (e.g., motivational interview or motivational enhancement therapy) and booster treatment to enhance participant motivation [49]. These findings suggest that readiness for change is a crucial factor in gambling disorder treatment and strengthening it could improve treatment outcomes.

No statistically significant differences in outcomes were found based on primary game types. Participants using chance games (EGMs) showed the highest urge to gamble at baseline and week 12. Research suggests that game types like EGMs and sports betting can affect treatment dropout rates [50]. While popular in Finland, the uneven representation of game types in our sample limits generalizability due to varying gambling cultures. Further research is needed in diverse contexts to evaluate how primary game type impacts treatment outcomes.

Strengths and Limitations

This study examined the associations of background factors on treatment outcomes for gambling disorder and explored the associations of medication adherence, which may be under-researched. Few participants were excluded due to missing data, and e-diary data may yield more accurate adherence information than recollections during visits. The study lacked standardized adherence recording methods, and traditional compliance monitoring was hindered by the characteristics of the IMP regimen. While e-diary data risk self-report bias [51], differential attrition was noted, with a higher non-completion rate in the placebo group, although no significant differences in IMP effects or withdrawal reasons were found. The naloxone group experienced slightly more adverse events. The authors suggest that study blinding likely remained intact and that the completion rate differences may result from chance, though the possibility of unblinding exists.

The therapist effect may have obscured pharmacological treatment impacts, as discussed previously [13]. Additionally, the clinical trial environment differs from real-world settings where treatment contact frequency is lower, which has been shown to enhance outcomes [52]. The follow-up duration may also be insufficient to evaluate effect longevity, and readiness to change was assessed with a single question; a full-scale measurement could provide better insights. Participants selected their treatment goals (abstinence or reduction), which should be considered in result interpretation.

Conclusion

This study found that sociodemographic factors and gambling game types had no impact on gambling symptoms, severity, self-efficacy, QoL, or depressive symptoms. Medication adherence was associated with positive treatment outcomes, regardless of treatment allocation, with participants showing higher adherence experiencing better results, likely reflecting greater motivation. While participants with more severe depressive symptoms exhibited a modest decrease in gambling symptoms and an increase in QoL, their gambling severity did not significantly change. Readiness to change was also associated with positive treatment outcomes. However, including treatment allocation in the analysis rendered most results statistically insignificant, suggesting that motivation and commitment to treatment play critical roles in improving outcomes for gambling disorder. Clinically, this underlines the importance of incorporating strategies that actively foster motivation such as motivational interviewing, structured feedback, or digital self-monitoring tools to strengthen adherence and engagement, thereby enhancing treatment effectiveness.

Statement of Ethics

This trial followed the ethical principles of the Helsinki Declaration, Good Clinical Practice standards, and local regulations. Approval was obtained from the National Committee on Medical Research Ethics (Registration No. 148/06.000.01/2017). The trial was registered on EudraCT (2017-001946-93) and ClinicalTrials.gov (NCT03430180). Written informed consent was acquired from all participants before any trial-specific procedures.

Conflict of Interest Statement

The authors declare that they have no competing interests.

Funding Sources

The trial was funded by the Finnish Institute for Health and Welfare. The investigational medicinal product was donated by Opiant Pharmaceuticals in Santa Monica, CA, USA, without terms or conditions. These parties were not involved in study design, execution and analysis, manuscript conception, planning, writing, or decision to publish.

Author Contributions

Niklas Mäkelä: conceptualization, and writing – original draft, review, and editing. Jarkko Isotalo: formal analysis, data curation, and writing – review and editing. Hannu Alho: writing – review and editing. Sari Castrén: writing – original draft, review, and editing.

Funding Statement

The trial was funded by the Finnish Institute for Health and Welfare. The investigational medicinal product was donated by Opiant Pharmaceuticals in Santa Monica, CA, USA, without terms or conditions. These parties were not involved in study design, execution and analysis, manuscript conception, planning, writing, or decision to publish.

Data Availability Statement

The authors declare that there is no preregistration in relation to this study. Due to legislative reasons, the dataset or supporting data are not available. Further inquiries can be directed to the corresponding author.