On 2017 May 30, Doug Berger commented:

Methodologic problems in this meta-analysis:

(1) This meta-analysis noted whether blinding was incorporated into the trial but did not (and probably cannot) evaluate if blinding was maintained properly.Trials with subjective endpoints (like major depression) that are also unable to maintain double-blinding throughout the trial should be invalidated. These trials cannot say anything about effectiveness of the drug being tested and an unblinded non-inferiority trial of poor quality comparing placebo to unblinded active drug is the result. An exit-analysis of the proportion of subjects and treaters who correctly guessed the arm they were assigned to is needed.

(2) Indications with objective as well as subjective endpoints in differing indications and therapeutic areas were included in this analysis which make interpretation impossible. Any indication with subjective endpoints and inadequate blinding should be excluded from a meta-analysis, however the quality of blinding would be difficult to confirm by this meta-analysis that did not specifically require proof with exit-analysis data.

(3) Clinical drug trials are carried out to confirm if drugs work, the drugs are not yet confirmed to be effective. Comparing placebo to a drug that does not work becomes a [placebo vs drug as placebo] comparison and says little about efficacy of placebo vs approved drug. Most drugs on the market with subjective endpoints (psychiatry) have previously been tested vs blind placebo, however, this meta-analysis did not limit itself only to drugs that went on to approval.

Conclusion: Placebo may function in a clinical setting as a useful tool for certain patients with certain conditions in assuaging some symptoms: pain, worry, hopelessness, etc, however, this meta-analysis had a number of methodologic problems as noted above that limit the conclusions that can be made on the actual effect of placebo vs drugs that have market approval.

On 2017 May 30, Doug Berger commented:

Methodologic problems in this meta-analysis:

(1) This meta-analysis noted whether blinding was incorporated into the trial but did not (and probably cannot) evaluate if blinding was maintained properly.Trials with subjective endpoints (like major depression) that are also unable to maintain double-blinding throughout the trial should be invalidated. These trials cannot say anything about effectiveness of the drug being tested and an unblinded non-inferiority trial of poor quality comparing placebo to unblinded active drug is the result. An exit-analysis of the proportion of subjects and treaters who correctly guessed the arm they were assigned to is needed.

(2) Indications with objective as well as subjective endpoints in differing indications and therapeutic areas were included in this analysis which make interpretation impossible. Any indication with subjective endpoints and inadequate blinding should be excluded from a meta-analysis, however the quality of blinding would be difficult to confirm by this meta-analysis that did not specifically require proof with exit-analysis data.

(3) Clinical drug trials are carried out to confirm if drugs work, the drugs are not yet confirmed to be effective. Comparing placebo to a drug that does not work becomes a [placebo vs drug as placebo] comparison and says little about efficacy of placebo vs approved drug. Most drugs on the market with subjective endpoints (psychiatry) have previously been tested vs blind placebo, however, this meta-analysis did not limit itself only to drugs that went on to approval.

Conclusion: Placebo may function in a clinical setting as a useful tool for certain patients with certain conditions in assuaging some symptoms: pain, worry, hopelessness, etc, however, this meta-analysis had a number of methodologic problems as noted above that limit the conclusions that can be made on the actual effect of placebo vs drugs that have market approval.