Journal Club
Alcohol, Other Drugs, and Health: Current Evidence
May–June 2012
Featured Article
A Randomized Controlled Trial of a Brief Intervention for Illicit Drugs Linked to the Alcohol, Smoking
and Substance Involvement Screening Test (ASSIST) in Clients
Recruited from Primary Health-Care Settings in Four Countries
Humeniuk R, et al. Addiction. 2012;107(5):957-66.
www.aodhealth.org 3
Study Objective
• To evaluate the effectiveness of a brief intervention (BI) for illicit drugs (cannabis, cocaine, amphetamine-type stimulants and opioids) linked to the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST).
4
Study Design• Prospective randomized controlled trial
conducted in primary health-care settings* in Australia, Brazil, India, and the United States.
• Patients screened with the ASSIST who scored in the moderate-risk range for cannabis, cocaine, amphetamine-type stimulants, or opioids (N=731) were assigned to either waitlist (control group) or to A brief intervention (BI) for the drug receiving the highest score.
• ASSIST-specific scores for cannabis, stimulants, or opioids as well as ASSIST total illicit substance involvement scores at baseline and 3 months were compared.
*Sexually-transmitted disease, dental, primary-care, and other outpatient clinics.
www.aodhealth.org 5
Assessing Validity of an Article about Therapy
• Are the results valid?
• What are the results?
• How can I apply the results to patient care?
www.aodhealth.org 6
Are the Results Valid?
• Were patients randomized?
• Was randomization concealed?
• Were patients analyzed in the groups to which they were randomized?
• Were patients in the treatment and control groups similar with respect to known prognostic variables?
www.aodhealth.org 7
Are the Results Valid? (cont‘d)
• Were patients aware of group allocation?
• Were clinicians aware of group allocation?
• Were outcome assessors aware of group allocation?
• Was follow-up complete?
www.aodhealth.org 8
Were patients randomized?
• Yes.
– Patients were randomized to BI or waitlist immediately following the baseline interview.
– Randomization was stratified by gender, substance, and level of use (high/low).
www.aodhealth.org 9
Was randomization concealed?
• Yes.
– Randomization lists for each drug category and country were prepared by the study coordinating center in Australia using a web-based randomization program.
www.aodhealth.org 10
Were patients analyzed in the groups to which they were
randomized?
• Yes (intention-to-treat analysis).
www.aodhealth.org 11
Were the patients in the treatment
and control groups similar?• Unknown.
– A demographic profile questionnaire was administered at baseline, however, demographic data broken down by group assignment was not provided.
– More than two-thirds of the total sample were men; the mean age was 31.4 years (SD=9.3), and the average years of education was 9.5 (SD=5.2). Just over half had never been married, and roughly one-third were either married or cohabiting. Most identified themselves as Caucasian (59.6%), followed by Indian (24.4%) or African (7.3%). Fifteen per cent (15%) of participants had received previous treatment for drug or alcohol problems.
– Groups did not differ significantly at baseline with respect to their total illicit substance involvement scores or specific substance involvement scores.
www.aodhealth.org 12
Were patients aware of group allocation?
• Yes.
– It was not possible to blind the patients as to whether they were receiving the BI or not.
www.aodhealth.org 13
Were clinicians aware of group allocation?
• Yes.
– Clinical research staff were not blind to group allocation as they were responsible for administering the intervention at baseline.
www.aodhealth.org 14
Were outcome assessors aware of group allocation?
• Yes.
– In the majority of cases, the same clinical researcher performed both the baseline and follow-up interviews.
www.aodhealth.org 15
Was follow-up complete?
• Forty-nine of 372 participants in the intervention group were lost to follow-up (13%) compared with 51 of 359 participants in the control group (14%).
www.aodhealth.org 16
What Are the Results?
• How large was the treatment effect?
• How precise was the estimate of the treatment effect?
www.aodhealth.org 17
How large was the treatment effect?
Intention-to-treat analysis—ANOVA total illicit substance involvement scores
n Baseline score (SD)
Follow-up score (SD)
Mean effect size
(% decrease)
Interaction effect*, p, power
Interaction by country effect, p
Australia:
F=6.5, p<0.001
BI 86 46.8 (19.3)
39.0 (17.6) 16.7% F=14.9, p<0.001, power=97%
Control 84 43.7 (18.4)
42.7 (20.0) 2.3%
Brazil:
BI 94 29.2 (14.4)
21.8 (13.9) 25.3% F=9.5, p<0.005, power=86%
Control 71 24.7 (11.9)
22.6 (11.8) 8.5%
India:
BI 89 34.7 (14.0)
26.5 (13.1) 23.6% F=9.4, p<0.005, power=86%
Control 88 34.8 (14.7)
31.2 (13.5) 10.3%
USA:
BI 103 34.9 (22.3)
31.1 (19.7) 10.9% F=2.5, p=0.11, power=35%
Control 115 39.0 (24.6)
31.3 (18.7) 19.7%
Pooled:
BI 372 36.1 (18.9)
29.5 (17.5) 18.3% F=7.4, p< 0.01, power=77%
Control 359 36.2 (19.9)
32.2 (17.9) 11.0%
*Interaction of time and experimental condition in predicting total illicit substance involvement score. BI: brief intervention; SD: standard deviation.
www.aodhealth.org 18
How large was the treatment effect? (cont’d)
Intention-to-treat analysis—cannabis scores
nBaseline
score (SD)
Follow-up score (SD)
Mean effect size (%
decrease)
Interaction effect*, p, power
Interaction by country effect, p
Australia:
F=5.9, p<0.001
BI 17 20.2 (5.3) 17.2 (6.1) 14.9% F=2.6, p=0.12, power=34%
Control 14 19.4 (7.6) 19.0 (7.6) 2.1%
Brazil:
BI 67 13.3 (6.5) 9.3 (8.2) 30.0% F=9.5, p<0.005, power=86%
Control 45 12.0 (6.0) 12.0 (7.1) 0.0%
India:
BI 54 22.8 (2.0) 18.9 (6.1) 17.1% F=10.8, p<0.001, power=90% Control 52 22.3 (2.5) 21.8 (4.9) 2.2%
USA:
BI 74 16.8 (7.7) 15.1 (9.5) 10.1% F=3.0, p=0.08, power=41%
Control 72 16.2 (6.7) 12.3 (7.0) 24.1%
Pooled:
BI 212 17.5 (7.1) 14.4 (8.9) 17.7% F=4.0, p<0.05, power=52%
Control 183 17.1 (6.8) 15.4 (7.9) 9.9%*Interaction of time and experimental condition in predicting cannabis-specific substance involvement score.BI: brief intervention; SD: standard deviation.
www.aodhealth.org 19
How large was the treatment effect? (cont’d)
Intention-to-treat analysis—stimulant scores
n Baseline score (SD)
Follow-up score (SD)
Mean effect size
(% decrease)
Interaction effect*, p, power
Interaction by country effect, p
Australia:
F=2.8, p=0.06
BI 68 16.8 (7.1) 11.9 (7.3) 29.2% F=8.5, p<0.005, power=83%
Control 70 15.5 (6.8) 13.7 (7.7) 11.6%
Brazil:
BI 27 15.7 (6.9) 6.5 (5.7) 58.6% F=7.0, p<0.01, power=74%
Control 26 11.1 (6.0) 7.7 (6.1) 30.6%
USA:
BI 23 20.9 (7.9) 16.2 (11.8) 22.5% F=0.08, p=0.8, power=6%
Control 33 18.5 (7.6) 13.2 (10.5) 28.6%
Pooled:
BI 118 17.3 (7.4) 11.5 (8.6) 33.5% F=9.4, p<0.005, power=86%
Control 129 15.4 (7.2) 12.4 (8.5) 19.5%
Intention-to-treat analysis—opioid scores
India:
BI 35 22.7 (2.6) 13.0 (8.6) 42.7% F=7.6, p<0.01, power=78%
Control 36 22.5 (2.2) 18.2 (7.8) 19.1%*Interaction of time and experimental condition in predicting stimulant- and opioid-specific substance involvement scores.BI: brief intervention; SD: standard deviation.
www.aodhealth.org 20
How Can I Apply the Results to Patient Care?
• Were the study patients similar to the patients in my practice?
• Were all clinically important outcomes considered?
• Are the likely treatment benefits worth the potential harm and costs?
www.aodhealth.org 21
Were the study patients similar to those in my practice?
• Participants were adult men and women from 4 countries. The majority were not from the United States. Diverse study sites were selected to represent a broad range of cultural, political and economic systems in which substance-related problems occur.
www.aodhealth.org 22
Were all clinically important outcomes considered?
• No.
−Drug use was indirectly assessed using ASSIST scores, limiting the clinical relevance and interpretation.
www.aodhealth.org 23
Are the likely treatment benefits worth the potential harm and
costs?
• No harms or costs were presented. The benefits were modest, of questionable clinical relevance, likely biased due to lack of blinding, and not statistically significant in the US sample.
• Additional research is needed prior to widespread dissemination.