Risk of bias
Risk of bias ratings, assessed with the Cochrane Risk of Bias tool, for the eligible RCT are shown in Table 2. Risk of bias was rated “Low” for Random Sequence Generation and Allocation Concealment. Risk of bias was “Unclear” for Blinding of Participants and Personnel, since a physician with access to patient data at trial entry, including EPDS scores, made final treatment recommendations. Because the EPDS was only administered pre-treatment to women in the screening arm of the trial, this physician was likely not blind to whether women were in the screening or non-screening trial arms. Risk of bias was also “Unclear” for Blinding of Outcome Assessment since outcomes were based on self-report questionnaires, and women would have known whether they received depres- sion management services and that depressive symptoms were being assessed. Risk of bias was “High” for Selective Outcome Reporting. This was because in their 2005 trial reg- istration (NCT00251342, http://clinicaltrials.gov/ct2/show/NCT00251342), the authors indicated that there would be 2 primary outcome measures, the EPDS and GHQ-12. In the published article [39] reporting trial results, however, they described only one primary outcome measure (EPDS scores), which was statistically significant. GHQ-12 scores, which were not statistically significant, were described as a secondary outcome. Remaining risk of bias domains were rated “Low” or “Unclear.” Other Sources of Bias was rated “Unclear” because changing the status of outcome variables based on trial results raises general concerns about the fidelity of trial conduct and reporting, and the effect sizes reported appeared to be substantially larger than what could be reasonably expected from the low-intensity depression care intervention, which involved a psychiatric referral or nurse counselling (See Table 2 and Discussion).
Risk of biasRisk of bias ratings, assessed with the Cochrane Risk of Bias tool, for the eligible RCT are shown in Table 2. Risk of bias was rated “Low” for Random Sequence Generation and Allocation Concealment. Risk of bias was “Unclear” for Blinding of Participants and Personnel, since a physician with access to patient data at trial entry, including EPDS scores, made final treatment recommendations. Because the EPDS was only administered pre-treatment to women in the screening arm of the trial, this physician was likely not blind to whether women were in the screening or non-screening trial arms. Risk of bias was also “Unclear” for Blinding of Outcome Assessment since outcomes were based on self-report questionnaires, and women would have known whether they received depres- sion management services and that depressive symptoms were being assessed. Risk of bias was “High” for Selective Outcome Reporting. This was because in their 2005 trial reg- istration (NCT00251342, http://clinicaltrials.gov/ct2/show/NCT00251342), the authors indicated that there would be 2 primary outcome measures, the EPDS and GHQ-12. In the published article [39] reporting trial results, however, they described only one primary outcome measure (EPDS scores), which was statistically significant. GHQ-12 scores, which were not statistically significant, were described as a secondary outcome. Remaining risk of bias domains were rated “Low” or “Unclear.” Other Sources of Bias was rated “Unclear” because changing the status of outcome variables based on trial results raises general concerns about the fidelity of trial conduct and reporting, and the effect sizes reported appeared to be substantially larger than what could be reasonably expected from the low-intensity depression care intervention, which involved a psychiatric referral or nurse counselling (See Table 2 and Discussion).
การแปล กรุณารอสักครู่..
