Outcome | No. of studies | Subjects, n | Pooled OR or SMD (95% CI) | I2, % | Summary of sex differences | GRADE quality of evidence | |
---|---|---|---|---|---|---|---|
M | F | ||||||
Illicit opioid use | |||||||
Cohort studies24,30,43 | 3 | 976 | 814 | 0.81 (0.50 to 1.31) | 82 | - | Very low*† |
RCTs32,40*40† | 3‡ | 75 | 38 | 1.39 (0.61 to 3.19) | 0 | - | Moderate§ |
Treatment retention27,30,35 | 3 | 1010 | 585 | 1.01 (0.62 to 1.63) | 77 | - | Low |
Polysubstance use | |||||||
Alcohol use24,30,39 | 3 | 809 | 701 | 0.52 (0.31 to 0.86) | 77 | Women less likely to use alcohol | Moderate*¶ |
Amphetamine use35,38 | 2 | 2691 | 462 | 1.47 (1.12 to 1.94) | 0 | Women more likely to use amphetamines | Low |
Benzodiazepine use35,38 | 2 | 2691 | 462 | 0.94 (0.70 to 1.27) | 44 | - | Low |
Cannabis use35,38 | 2 | 2691 | 462 | 0.85 (0.67 to 1.08) | 0 | - | Low |
Cocaine use35,38,39 | 3 | 2826 | 571 | 1.07 (0.64 to 1.78) | 76 | - | Very low*†§¶ |
Other | |||||||
Methadone dosage (maintenance)26,35 | 2 | 562 | 234 | -2.38 (−5.67 to 0.91) | 0 | - | Low |
Legal involvement24,30 | 2 | 674 | 592 | 0.63 (0.47 to 0.84) | 39 | Women less likely to report arrests or legal supervision | Moderate*†¶ |
Employment during treatment24,25,28,30,39 | 5 | 1291 | 1030 | 0.39 (0.21 to 0.73) | 91 | Women less likely to be employed | Moderate*§¶** |
Marital status24,25,30,39 | 4 | 1100 | 878 | 0.96 (0.75 to 1.21) | 0 | - | Low |
Death27,43 | 2 | 581 | 353 | 1.61 (0.60 to 4.33) | 83 | - | Low |
Note: CI = confidence interval, GRADE = Grading of Recommendations Assessment, Development and Evaluation,23, OR = odds ratio, RCT = randomized controlled trial, SMD = standardized mean difference. *Differences in outcome definition and measurement among studies. †Studies did not adjust for relevant treatment-related confounders (e.g., methadone dosage, opioid use, other medications). ‡This meta-analysis included 2 studies, but 1 study (Schottenfeld et al.,40) had 2 methadone intervention arms (65 mg and 20 mg), which were both included separately. §Small samples and wide confidence intervals across studies. ¶Significant association at p < 0.01. **Inadequate statistical measures and some missing data.