Effectiveness of problem gambling brief telephone interventions: an uncontrolled outcome study – 3 years later
1. EFFECTIVENESS OF PROBLEM
GAMBLING BRIEF TELEPHONE
INTERVENTIONS: AN
UNCONTROLLED OUTCOME STUDY
– 3 YEARS LATER
K Palmer du Preez, M Abbott, M Bellringer, J
Landon, N Garrett,& R Maheswaran
International Gambling Conference, Auckland, New Zealand,
10-12 February 2016.
2. Background
• GARC funded in 2009 to conduct a RCT of 4 interventions
delivered by the national Gambling Helpline:
• Three interventions were variants of MI
• One intervention was standard care: ‘counselling and options’.
• Outcome study reports on this standard care arm.
• Our results at 12 months have been presented at
previous forums 2013/14.
• Today: how clients were doing 3 years later.
3. Aims
• Examined at 3, 6, 12m:
• Outcomes for help-seekers calling the Gambling Helpline
• Differences in outcomes for those receiving telephone assistance
only vs. additional counselling services
• Client characteristics associated with treatment outcome.
• Purpose of additional 36m follow up:
• Assess durability of outcomes
• Identify and clarify predictors of treatment success.
4. Participants
• 150 participants.
• 57% female, 43% male.
• 43% Māori, 42% NZ European, 10% Pacific people and 5%
Asian/other.
• Majority 25-34 years (25%) or 35-44 years (27%).
• Primary problem gambling mode – 89% EGMs (76.2% pub,
8.8% club, 3.4% casino), 4.1% track, 3.4% casino tables.
Retention by assessment point
Baseline 3 months 6 months 12 months 36 months
Number 150 129 119 99 60
% Baseline 100% 86% 79% 66% 40%
• No differential attrition
5. Self-report
• Days gambled, money lost, treatment goal success
• Control over gambling, gambling impacts, problem gambling
severity (PGSI)
• Psychiatric co-morbidity
• Psychological distress
• Tobacco/drug/alcohol use
Baseline and Follow-Up Assessments
(3, 6, 12, and 36 months)
10. PGSI items
1. How often have you bet more that you could really afford to lose?
2. How often have you needed to gamble with larger amounts of money
to get the same feeling of excitement?
3. How often have you gone back another day to try and win back money
you lost?
4. How often have you borrowed money or sold anything to get money to
gamble?
5. How often have you felt that you might have a problem with gambling?
6. How often have people criticised your betting or told you that you had
a gambling problem, regardless of whether or not you thought it was
true?
7. How often have you felt guilty about the way you gamble or what
happens when you gamble?
8. How often has gambling caused you any health problems including
stress and anxiety?
9. How often has your gambling caused any financial problems for you or
your household?
12. Alcohol and substance use
Alcohol abuse (AUDIT – C)
& smoking by assessment
point
Drug abuse (DAST) by
assessment point
13. Accessing additional assistance
• Receiving additional assistance by 3, 6, 12 or 36m was not associated
with 36m PGSI 12 scores, nor any of our other outcome measures (days
gambled, money lost, control and treatment success).
14. Predictors of successful outcomes - PGSI
Multivariate analyses:
• Better outcomes for people who were partnered, had no previous treatment for mental
health issues or gambling.
#Adjusted for all other effects in the model
Change in PGSI-12 adjusted values Adjusted values#
Variable
Category
Est. least squares mean
diff.
Standard
Error
p-value
Assessment point 12 months -6.06 0.69
36 months -9.06 0.82 <0.01
Marital status Partnered -9.30 0.81
Not partnered -5.83 0.69 <0.001
Previous treatment for mental
health
No -9.26 0.59
Yes -5.87 0.95 <0.01
Previously treatment for problem
gambling
No -8.88 0.64
Yes -6.25 0.88 0.01
15. Predictors of Control over gambling.
• More socioeconomic deprivation, lower quality of life - less
improvement in control
• Belief in treatment success associated with improvement in
control
Change in control over gambling at 36
months
Adjusted values#
Est. least squares
mean diff. Standard Error p-value
NZDI (quartiles) 0 - 0.58 4.85 0.35
0.59 - 1.23 4.42 0.38
1.24 - 2.37 3.31 0.38
2.38+ 3.66 0.45 0.03
WHOQoL-8 (quartiles) 0 - 20 2.56 0.43
21 - 25 4.52 0.38
26 - 29 4.50 0.38
30+ 4.66 0.39 0.001
Belief in treatment
success
Lower than
median
3.60 0.26
Higher than
median
4.52 0.28 0.02
#Adjusted for all other effects in the model
16. Additional treatment
• Two thirds received some additional assistance beyond
helpline telephone session (1/3 formal)
• Mostly accessed in first 3m
• Encouragement to do so built into standard care approach
• NZ PG services are varied & widely available – impact??
• Receiving additional assistance not associated with
outcomes
• Can’t infer that additional treatment not beneficial.
• Participant choice not random allocation.
17. Comorbidity
• Improvement in distress & depression comparable to
outcomes of specialised therapies
• Relatively low improvement in tobacco use and alcohol
abuse
• Implications for treatment:
• Suggests additional referral/treatment warranted
• Co-morbid addictive behaviours may provide ‘relapse triggers’ for
gambling and limit long term treatment effectiveness (Nower et al,
2013).
• Acceptability to client and impact on gambling treatment goals?
18. Factors associated with worse outcomes
• Helps us identify client groups who may benefit from more
intensive support
• PGSI-12 = key clinical measure
• Illness, sick leave, widowed; Without partner
• Previously received treatment for mental health or gambling issues.
• Other outcome measures (days, dollars, control)
• High PGSI-12, low quality of life, low self-efficacy at baseline
• Past year MH treatment.
• Previous formal help seeking may indicate comorbidity
and more chronic gambling problems?
19. Factors associated with worse outcomes
• All associated with a worse outcomes on a single
outcome measure only.
• At both the 12m and 36m follow up points: clinically and
statistically significant improvements were seen
irrespective of:
• age,
• gender,
• ethnicity,
• almost all sociodemographic and baseline client characteristics we
considered.
20. Concluding comments
• Significant improvement in gambling and related problems
36m after brief telephone intervention.
• Additional improvements in:
• Days gambled, median PGSI, proportion of PGs.
• Similar outcomes to trials and outcome studies of face-to-
face therapy
• Additional treatment not associated with better outcomes.
• Suggest: Most gamblers do not require intensive long
duration face-to-face interventions.
• Challenge is to identify those who do from the outset.
21. Further research
• Uncontrolled outcome studies have limitations: no random
allocation, many confounding factors.
• Large sample RCTs important to establish which clients
respond better to which treatments
• Assessment of costs of treatment programmes relative to
outcome and benefits is important:
• Incorporated into our new Trial of face-to-face services in NZ.
22. Thanks to…
• Our participants.
• The Gambling Helpline (now part of Homecare Medical)
• The diligent students who undertook much of our follow-
up work.
• The New Zealand Ministry of Health.
• Full report will be available New Zealand Ministry of
Health (www.health.govt.nz) and Gambling and Addictions
Research Centre (www.aut-grc.ac.nz) websites shortly.
Notas do Editor
The GARC at AUT was funded by MOH to conduct a RCT of 4 clearly defined interventions delivered by the national Gambling Helpline.
The RCT involved 462 Helpline callers who met the eligibility criteria and were randomly allocated to the 4 treatment groups.
One arm of the RCT was a manualised version of the Helpline’s standard care. Others were variations of motivational interviewing.
This study reports on the ‘treatment as usual’ arm of the RCT supplemented by additional clients recruited after the RCT recruitment was completed.
This was a prospective (36 months) cohort study with aims to:
Report on outcomes for help-seekers calling the Gambling Helpline.
Determine whether there are any differences in participant outcomes between those who only access telephone assistance, and those who also access professional counselling services.
Identify client characteristics that are associated with treatment outcome.
Similar to HL usual client base.
Results at 12 months have been presented at previous forums 2013/14.
For our gambling outcome measures days gambling in the past month, dollars lost per day and control over gambling we saw substantial improvements from baseline to 3 months, that were maintained at 6 and 12m and also held at 36 months post intervention.
10 point scale.
Vast improvement in self reported areas of life affected by gambling at 3 months. Slight improvement continued to be noted at each of the six, 12 and 36 month assessments.
PGSI median scores using 12m and 3m timeframe decreased over time:
PGSI-12 median score at 12 months was half baseline, and reduced even further at 36m to 5.
PGSI-3: we can see the substantial improvement that occurred in the first 3 months that was maintained at 6 and 12 months.
There was a further substantial decrease at 36m, median = 2, low risk gambler category.
Risk/problem Category – almost all PGs at baseline, reduced to 58% at 12 months, even fewer at 36m 38%.
Although 2/3 of participants remained either problem gamblers or moderate risk gamblers at 36m (albeit at lower severity), one-third had improved significantly to a low-risk or non-problem level.
When the PGSI items are examined we can contextualise the 2/3 PG or moderate risk further.
A36m, the items participants rated as the most problematic on average were: feeling that one has a problem with gambling (5) and feeling guilty about what happens when one gambles (7).
It could be that participants are still being classified as problem gamblers due to long lasting effects of gambling, even when they may have stopped gambling completely: for example with item number:
5. A perception that one always has a gambling problem (akin to being an alcoholic),
7. Lasting feeling of guilt over what happened when one gambled.
9. Lasting financial impacts of gambling?
Further analysis is required to assess changes in responses to individual questions over time. It is likely that responses to some questions will change markedly within the first three months and that others will change later.
Kessler 10 categorisation of distress shifted dramatically in the first 3 months, at 36m just 3% classified as experiencing high levels of distress compared to over 50% at baseline.
In terms of affective disorder, we saw significant improvements in major and minor depressive disorders, though not dysthymia probably due to the more chronic nature of this condition.
Slight reductions in smoking across the four assessment points, and alcohol abuse from baseline to 6 months, however at 12 and 36 months alcohol abuse rose again to over 60%.
This points to some specificity of gambling intervention in that other issues were not necessarily improved.
While most of our participants did not report other drug misuse, those experiencing problems reduced from 12.3% at baseline to 2% at 12 months, rising slightly to 5% at 36m.
Uptake of additional professional treatment services sat at around 1/3 during the first 3 months post intervention and declined to 13% at 36 months.
Although many people accessed additional support over and above the initial helpline conversation, those who accessed this support did no better than those who didn’t.
Males were more likely to access these further services in the first 3 months regardless of whether they had received treatment in the past. No other sociodemographic variables predicted uptake.
Those who had received treatment in the past 12 months for either gambling or mental health were also more likely to engage in counselling or therapy.
Males had significantly (p = .02) greater odds (adjusted OR 2.85) for uptake of formal services (adjusted for previous gambling and mental health treatment)
Those who had received previous gambling or mental health treatment were also more likely to take up formal services.
At 12m:There were no significant differences between those who accessed additional treatment and those who didn’t in relation to PGSI scores from baseline to 12 months.
At 12m: There were smaller improvements however for those outside the paid workforce. Only socio ecc variable that held.
At 12m: Those gambling on pub egms showed less improvement when confounders adjusted for.
At 12m: An unexpected finding was that callers who gambled on pub pokies improved much less than those who gambled at casinos whether they had additional counselling or not. In the past pokies in pubs, clubs and casinos have been lumped together.
At 12m…Belief in treatment success was also significantly related to an increase in control over gambling, additionally - a lower perceived difficulty in overcoming the gambling problem.
Though it was not those with more severe problems that sought out that extra support.
Additional gambling treatment was not associated with better outcomes.
Telephone interventions were monitored carefully, there is substantial variability in the approach to face-to-face service delivery – we did not describe or account for this.
Cannot be certain what this means:
Could be additional treatment contributes little above the initial session.
Could be that the unaccounted for variability in face-to-face approaches was an issue.
Could be that clients are relatively accurate in determining what they need – i.e., without the additional treatment those clients might have fared worse.
Additional service access was not predicted by more serious gambling problems, mental health issues nor other relevant baseline issue.
This information helps us with trying to identify those client groups who may need additional more intensive support.
Overall: Minimal subgroup differences.
Gambling trajectories: changes in treatment goals, influencing factors and consequences.