Skip to main content

Table 2 Randomized study quality indicators

From: Efficacy of depression treatments for immigrant patients: results from a systematic review

Author (year) Randomization Allocation masking Attrition Missing data handling Limitations
Dwight-Johnson et al. (2011) [54] Yes: Stratified permuted-block randomization Participants: No Intervention: 16% Intent-to-treat analysis (ITT) employed No power calculation
ITT can increase chance of false positive
  Researchers: No Control: 30%
Outcome Assessor: Yes
Ueberlacker et al. (2011) [62] Yes: method not described Participants: No Intervention: 26% Not described in detail, but it appears that missing data points have been excluded. Small sample, risk of attrition bias
Researchers: No Control: 42%
No power calculation
Outcome Assessor: Yes
Yeung et al. (2010) [64] Yes: computer-generated table Participants: No Not reported Not reported No power calculation
Researchers: No
Outcome Assessor: Yes
Yeung et al. (2012) [63] Yes: randomized using computer-generated numbers Participants: No Intervention: 27% Used data from week 6 if no data available at week 12. If neither data point available participant was excluded from analysis Power calculation suggest much larger sample is required
Researchers: No Control: 15%
Outcome Assessor: Yes
Choi et al. (2012) [53] Yes: randomization process by independent person Participants: No Intervention: 34% Baseline carried forward The missing data approach may introduce false positives. No power calculation Small sample
Researchers: No Control: 10%
Outcome Assessor: No
Beeber et al. (2010) [51] Yes: block randomization Participants: No Intervention: 13% Power calculation completed and extra participants included to compensate for possible attrition Small sample
Researchers: No Outcome Assessor: No Control: 10%
Renner et al. (2011) [59] Yes: method not described Participants: Not reported Intervention CBT: 52% Non-completers excluded Small sample high risk of attrition bias
Researchers: Not reported Intervention SHG: 28%
Potential risk of selection bias
Outcome Assessor: Not reported Control: 45%
No power calculation