| Study |
Bias |
| Randomisation process |
Deviations from intended interventions |
Missing outcome data |
Measurement of the outcome |
Selection of the reported results |
Overall |
| Authors' judgement |
Support for judgement |
Authors' judgement |
Support for judgement |
Authors' judgement |
Support for judgement |
Authors' judgement |
Support for judgement |
Authors' judgement |
Support for judgement |
Authors' judgement |
Support for judgement |
| Subgroup 1.1.1 SSRIs |
| EUCTR2008‐002159‐25‐FR |
Some concerns |
Study described as randomized, however no detail is provided concerning the generation of the allocation sequence nor whether it was concealed or not until participants were enrolled (unpublished study). No major imbalances between intervention groups according to the results posted on clinicaltrials.gov |
Low risk of bias |
Study described as double blind. No further detail is provided concerning the blinding procedure (unpublished study). A mITT approach has been used to measure the outcome, excluding from the analysis those patients who were unable to follow up. Completers were analyzed as randomized |
Some concerns |
8/20 patients in the escitalopram group (40%) and 6/18 in the placebo group (33%) were unable to follow up, hence their outcome wasn't available. No evidence is provided to justify the fact that the result was not biased by missing outcome data. Missingness in the outcome could depend on its true value, however specific reasons are provided for those drop‐outs that occurred and seem balanced between the two intervention groups, making it unlikely that this missingness of the outcome might depend on its true value |
Some concerns |
The Montgomery‐Asberg Depression Rating Scale (MADRS) is a validated tool to measure the outcome in question. No information is available concerning the frequency and way the outcome was measured in the two intervention groups. Concomitant medications for insomnia and nausea were allowed in both groups according to the study protocol. No detail is provided concerning the blinding of the assessors, however it seems implied in the double‐blind design of the study |
High risk of bias |
On the study protocol HADS scale was listed as the main tool to evaluate efficacy. However only results of the assessment with the MADRS and CES‐D scales are available. |
High risk of bias |
The overall risk of bias is high. The study is unpublished. Important information on the way the outcome was measured are indeed missing (no detail on randomization/ concealment of the allocation sequence, no clear detail on how the outcome was measured in the two study groups). Moreover the attrition rates are high in both group and the way the main outcome was measured was not coherent with the study protocol. |
| Fisch 2003 |
Low risk of bias |
Quote: "[...] randomly assigned in a double‐blind manner to receive either fluoxetine (20‐mg tablets) or an identical placebo tablet. The randomisation was performed centrally through a preprinted randomisation table, and the study drug was sent by overnight mail directly to the patient". It is safe to assume that this method of randomization might have ensured concealment of the allocation sequence. No major imbalances between intervention groups |
Low risk of bias |
Study described as double blind. Fluoxetine and placebo pills were identical in appearance. No further detail is provided concerning the blinding procedure. A mITT approach has been used to measure the outcome, excluding from the analysis those patients who were unable to follow up. The other patients were analyzed as randomized |
High risk of bias |
26/64 patients in the fluoxetine group (40%) and 21/65 in the placebo group (32,3%) were unable to follow up, hence their outcome wasn't available. No evidence is provided to justify the fact that the result was not biased by missing outcome data. Missingness in the outcome could depend on its true value, drop out to adverse events are reported in the paper constitute only a small percentage of the total drop‐out rates and are well distributed among the two intervention groups. However other reasons for drop‐outs are not described in enough detail to estimate how likely it is that the missingness in the outcome might have depended on its true value |
Low risk of bias |
The BZSDS (Brief Zung Self‐Rating Depression Scale) is a validated tool to measure the outcome in question. According to study authors "The mean interval between visits (in weeks) was similar for patients in each treatment arm" and "Longitudinal assessments were performed at baseline and every 3 to 6 weeks thereafter and included the measurement of QOL and depression. The visit interval varied among patients and often depended on the schedule for anticancer therapy. Patients were assessed for 12 weeks, and complete assessment involved three to five sessions of data collection (depending on the individual patient’s visit intervals)." No detail is provided concerning the blinding of the assessors, however it seems implied in the double‐blind design of the study |
Some concerns |
No protocol or pre‐specified analysis plan is available for the study in question. However relevant data for the pre‐specified (methods) outcomes are well‐reported (results). The way the outcome was measured seems coherent with the prespecified plan (methods). It is however possible that the study authors measured other outcomes and/or conducted multiple analyses, but did not report them. |
High risk of bias |
We considered the measurement of the outcome to be at high risk of bias for the high attrition rates of study participants in both groups (40% in the fluoxetine group, 32,3% in the fluoxetine group). Furthermore reasons for drop‐outs are not described in enough detail. |
| Musselman 2006 |
Some concerns |
Study described as randomized, however no detail is provided concerning the randomization of the allocation sequence nor whether it was concealed or not until participants were enrolled. No major imbalances between intervention groups (with the exception of stage, being less advanced in the placebo‐treated group, and previous chemotherapy, being less frequent in the placebo‐treated group) |
Low risk of bias |
Study described as double blind. No further detail is provided concerning the blinding procedure. Patients were analyzed as randomized. |
High risk of bias |
5 out of 13 patients in the paroxetine group (40%), 4 out of 11 patients in the desipramine group (36%) and 5 out of 11 patients in the placebo group (45%) were unable to follow up, hence their outcome wasn't available. No evidence is provided to justify the fact that the result was not biased by missing outcome data. Reason for leaving the study are well described and overall balanced between groups, however dropout rates are relevant. Moreover, a relevant portion of missing data are possibly related to the true outcome (2 patients in the paroxetine group versus 2 in the desipramine group versus 0 patients in the placebo group dropped due to inefficacy) |
Low risk of bias |
the Hamilton Depression Rating Scale (HAM‐D) is a validated tool to measure the outcome in question. Patients in both groups were assessed for both efficacy and adverse events at baseline, weekly during weeks 1 through 6 and on a monthly basis thereafter for a total of 5 months of double‐blind, randomized treatment. No detail is provided concerning the blinding of the assessors, however it seems implied in the double‐blind design of the study |
Some concerns |
No protocol or pre‐specified analysos plan is available for the study in question. However prespecified outcomes are reported for the endpoint assessment (week 6) in the methods section. The way the outcome was measured seems coherent with the prespecified plan (methods). It is however possible that the study authors measured other outcomes and/or conducted multiple analyses, but did not report them. |
High risk of bias |
Although the overall quality of the study was acceptable, we considered the way the outcome was measured to be at high risk of bias for the high attrition rates. Moreover the reasons for the dropouts among the three groups are not well balanced enough and they may be related to the true outcome |
| Razavi 1996 |
Some concerns |
Study described as randomized, however no detail is provided concerning the randomization of the allocation sequence nor whether it was concealed or not until participants were enrolled. No major imbalances between intervention groups |
Low risk of bias |
Study described as double blind. No further detail is provided concerning the blinding procedure. A mITT approach has been used to measure the outcome, excluding from the analysis those patients who were unable to follow up. The other patients were analyzed as randomized |
High risk of bias |
15/45 patients in the tluoxetine group (33%) and 7/46 patients in the placebo group (15,2%) were unable to follow up, hence their outcome wasn't available. No evidence is provided to justify the fact that the result was not biased by missing outcome data. Missingness in the outcome it's likely to depend on its true value. Indeed, there is imbalance between the two groups, even though reasons for dropping out from the study are well described. |
Low risk of bias |
The Hospital Anxiety and Depression scale is a validated tool to measure the outcome in question. The outcome was measured at regular follow up visits in both intervention groups. No detail is provided concerning the blinding of the assessors, however it seems implied in the double‐blind design of the study |
Some concerns |
No protocol or pre‐specified analysis plan is available for the study in question. Outcomes are not clearly pre‐specified in the method section (quote: "[...] evaluate, in a double‐blind placebo‐controlled design, the effectiveness of fluoxetine to treat and/or to control anxiety and depression [...]"). For relevant outcomes mean scores on rating scales are reported for 'visit 1' (but it is not clearly explained if it matches with the baseline point) and for 'visit 5'. |
High risk of bias |
We judged the measurement of this outcome to be at high risk of bias. The randomization process is not described in enough detail and the attrition rates are quite high and not well balanced between the two groups (although the reasons for the drop‐outs are described ). The way the outcomes were measured and analyzed were not pre‐specified in enough detail in the method section |
| Tavakoli Ardakani 2019 |
Low risk of bias |
Study described as randomized. Generation of the allocation sequence was random and it was concealed until participants were enrolled and assigned to interventions. Quote: "Both placebo and sertraline tablets were prepared and packaged by the same manufacturer (SobhanTM Pharmaceutical Company, Rasht, Iran), were visually identical and labelled with a code number from the random list. At inclusion, general practitioners gave the code number from the drug unit to the patient and all patient related data." No major imbalances between intervention groups |
Low risk of bias |
Study described as double blind. No further detail is provided concerning the blinding procedure; A mITT approach has been used to measure the outcome, excluding from the analysis those patients who were unable to follow up. The other patients were analyzed as randomized |
Low risk of bias |
9 patients out of 56 (16%) were lost to follow‐up and their outcome wasn't available. Of the 9 patients who were lost to follow up 5 belonged to the sertraline group and 4 to the placebo group. The number of patients with missing outcome data is inferior to 20% and this can be considered acceptable considered the type of population and the trial intervention |
Low risk of bias |
The Hospital Anxiety and Depression scale is a validated tool to measure the outcome in question; The outcome was measured at 3 and 6 weeks in both intervention groups; No detail is provided concerning the blinding of the assessors, however it seems implied in the double‐blind design of the study |
Some concerns |
No protocol or pre‐specified analysis plan is available for the study in question. Relevant data for the pre‐specified (methods) outcomes are well‐reported (results). It is however possible that the study authors measured other outcomes and/or conducted multiple analyses, but did not report them. |
Some concerns |
The overall assessment of the outcome had some concerns for a risk of bias, mainly due to the lack of a registered protocol or a pre specified analysis plan |
| Subgroup 1.1.2 TCAs |
| Musselman 2006 |
Some concerns |
Study described as randomized, however no detail is provided concerning the randomization of the allocation sequence nor whether it was concealed or not until participants were enrolled. No major imbalances between intervention groups (with the exception of stage, being less advanced in the placebo‐treated group, and previous chemotherapy, being less frequent in the placebo‐treated group) |
Low risk of bias |
Study described as double blind. No further detail is provided concerning the blinding procedure. Patients were analyzed as randomized. |
High risk of bias |
5 out of 13 patients in the paroxetine group (40%), 4 out of 11 patients in the desipramine group (36%) and 5 out of 11 patients in the placebo group (45%) were unable to follow up, hence their outcome wasn't available. No evidence is provided to justify the fact that the result was not biased by missing outcome data. Reason for leaving the study are well described and overall balanced between groups, however dropout rates are relevant. Moreover, a relevant portion of missing data are possibly related to the true outcome (2 patients in the paroxetine group versus 2 in the desipramine group versus 0 patients in the placebo group dropped due to inefficacy) |
Low risk of bias |
the Hamilton Depression Rating Scale (HAM‐D) is a validated tool to measure the outcome in question. Patients in both groups were assessed for both efficacy and adverse events at baseline, weekly during weeks 1 through 6 and on a monthly basis thereafter for a total of 5 months of double‐blind, randomized treatment. No detail is provided concerning the blinding of the assessors, however it seems implied in the double‐blind design of the study |
Some concerns |
No protocol or pre‐specified analysos plan is available for the study in question. However prespecified outcomes are reported for the endpoint assessment (week 6) in the methods section. The way the outcome was measured seems coherent with the prespecified plan (methods). It is however possible that the study authors measured other outcomes and/or conducted multiple analyses, but did not report them. |
High risk of bias |
Although the overall quality of the study was acceptable, we considered the way the outcome was measured to be at high risk of bias for the high attrition rates. Moreover the reasons for the dropouts among the three groups are not well balanced enough and they may be related to the true outcome |
| Subgroup 1.1.3 Other antidepressants |
| Liu 2021 |
Low risk of bias |
Quote: "patients were randomized into different groups through a computer‐generated randomized list using the SPSS software (SPSS Inc., Chicago, USA)". No detail is provided concerning the concealment of the allocation sequence". No major imbalances between intervention groups. No further detail is provided concerning the concealment of the allocation sequence, however, although information on the sequence generation and the allocation concealment are poorly reported, we considered that, given the overall good methodological quality of the study, major issues of the randomization process are unlikely to have occurred. |
Low risk of bias |
Study described as double blind. No further detail is provided concerning the blinding procedure. Patients were analyzed as randomized |
Low risk of bias |
According to study authors "No patients were lost to follow up" |
Low risk of bias |
The Hospital Anxiety and Depression scale is a validated tool to measure the outcome in question. The outcome was measured at day 1,3,7, at 1 month and 3 months in both intervention groups. No detail is provided concerning the blinding of the assessors, however it seems implied in the double‐blind design of the study. |
Some concerns |
No protocol or pre‐specified analysis plan is available for the study in question. However relevant data for the pre‐specified (methods) outcomes are well‐reported (results). The way the outcome was measured seems coherent with the prespecified plan (methods). It is however possible that the study authors measured other outcomes and/or conducted multiple analyses, but did not report them. |
Some concerns |
We considered the measurement of the outcome to have some concerns of risk of bias. The overall methodology was good, all patients were randomized as analyzed and no drop‐out occurred. The way the outcome was measured is appropriate, however no protocol or pre‐specified analysis plan was available for the study in question |
| Van Heeringen 1996 |
Some concerns |
Study described as randomized, however no detail is provided concerning the randomization of the allocation sequence nor whether it was concealed or not until participants were enrolled. No major imbalances between intervention groups |
Low risk of bias |
Study described as double blind. Placebo and mianserin pills were identical in appearance. No further detail is provided concerning the blinding procedure. A mITT analysis has been conducted, including in the final analysis all patients who had at least 1 post‐baseline follow‐up assessment. |
High risk of bias |
6/28 patients in the tluoxetine group (33%) and 15/27 patients in the placebo group (55%) were unable to follow up, hence their outcome wasn't available. No evidence is provided to justify the fact that the result was not biased by missing outcome data. Quote: "Significantly more placebo‐treated patients (n = 15) than mianserin‐treated patients (n = 6) prematurely terminated the study (Fishers' exact test P,=0.014). When withdrawals because of lack of efficacy were analysed, the difference was again significant and in favour of mianserin (2 and 11 patients in the mianserin and placebo treatment group, respec tively; Fisher's exact test, P= 0.006)." Missing of the outcome can indeed be explained by lack of efficacy in the placebo group. |
Low risk of bias |
The Hospital Anxiety and Depression scale is a validated tool to measure the outcome in question. The outcome was measured at regular follow up visits in both intervention groups at week 2, 4 and 6. No other reason to suspect that this kind of bias might have occurred. No detail is provided concerning the blinding of the assessors, however it seems implied in the double‐blind design of the study |
Some concerns |
No protocol or pre‐specified analysos plan is available for the study in question. Outcomes are not clearly prespecified (quote: "The aim of our study was to evaluate the efficacy and safety of mianserin in patients with breast cancer [...]"). However, mean change scores on HDRS, response rates and rates of relevant adverse events are reported. |
High risk of bias |
We considered the measurement of the outcome to be at high risk of bias. The attrition rates are high in both groups, with a significant imbalance between the two arms for the droupouts due to lack of efficacy (2 in the mianserin group vs 11 in the placebo group). The potential bias due to missing of outcome data favors placebo over comparator |
