Evaluation of a Danish meta-analysis of antidepressants

The Danish Medicines Agency has evaluated a Danish meta-analysis made by the Copenhagen Trial Unit on the effects of antidepressants (SSRIs). The analysis attracted extensive media coverage.

In connection with a consultation of the Danish Parliament’s Health and Senior Citizen's Committee, the Danish Medicines Agency has made a thorough evaluation of a Danish meta-analysis of antidepressants prepared by the Copenhagen Trial Unit (CTU). The CTU’s meta-analysis attracted considerable media coverage earlier this year, the key focus of which was that antidepressants have no effect. The purpose of the Danish Medicines Agency’s evaluation was to assess whether the CTU’s meta-analysis presented any new evidence and whether there may be any reasons for changing the recommendations for use of antidepressants.

In the opinion of the Danish Medicines Agency, the CTU’s analysis does not give rise to any changes to package leaflets, summaries of product characteristics or the recommendations for use of antidepressants by adult patients suffering from depression.

The CTU’s analysis is not better or more thorough than previous analyses and does not contribute any new evidence of the effect or risk of adverse reactions.

The Danish Medicines Agency’s evaluation of the meta-analysis was made by Eskild Colding-Jørgensen, Chief Medical Officer, Doris Stenver, Chief Medical Officer, and Anders Frøkjær Thomsen, external medical assessor and specialist in psychiatry, PhD.

Eskild Colding-Jørgensen comments on the conclusion: ”The research team behind the meta-analysis delivered a very strong message across the media that antidepressants have no effect. We believe that this message may cause uncertainty among patients who use antidepressant medicine; consequently, we are publishing our thorough evaluation of the analysis and we do not support the conclusions of the research team.”

About the meta-analysis

The specific publication: "Selective serotonin reuptake inhibitors versus placebo in patients with major depressive disorder. A systematic review with meta-analysis and Trial Sequential Analysis" published by the international journal BMC Psychiatry in February 2017.

This is a meta-analysis summarising all available and relevant trial results in this field. The trial results included in the meta-analysis were typically published previously and separately as scientific publications. Thus, the meta-analysis is not based on new data but presents a new comparison of old data.

About the methodology

The research team searched for eligible randomised clinical trials in a number of databases. A total of 131 trials were included in the overall meta-analysis. The primary focus of the analysis is on 49 trials that the team estimated would give detailed information on the effect of antidepressants.

The Danish Medicines Agency would like to express the following reservations about the CTU’s analysis:

• The trials included are generally very different with different purposes, different severity of depression, different duration of treatment, and different doses, which affect the overall conclusions. To draw good information from a meta-analysis, the set-up of the trials included must be reasonably comparable.
• Some of the 49 trials are not suitable for drawing broad conclusions, for example because of the very short duration of treatment – down to one week. This is too short a period of time to measure the effect of antidepressant medicines.
• Some of the 49 trials included patients who were given low doses of medicine. If the dose is too low, the patients in the trial may not have achieved the right effect.
• Some of the 49 trials included atypical patient groups, for example patients who are addicted to painkillers and are also given antidepressant medicine. The simultaneous use of several types of medicine may have an impact on the effect.

The CTU research team concludes that antidepressant medicines (SSRIs) have statistically significant effects. But the team also mentions that the effect measured was less than 2 HDRS points, which is below the threshold for clinical significance, according to the research team. The effect estimate predefined by the research team should be a minimum of 3 HDRS points, which is primarily based on other scientific articles.

The Danish Medicines Agency has identified several problems related to the CTU’s conclusions. First, the trials included in the analysis are very different and they were hardly all designed to demonstrate differences of 3 HDRS points.

Secondly, assessment of clinical significance requires considerations that are more varied and the involvement of relevant medical expertise.

It is important to underline that the CTU research team also found statistically significant effect when they applied a different effect estimate: the probability of remission of depression. This points to the clinical significance of the effect of antidepressants. Moreover, a few trials measured the effect on patients’ quality of life – and again, a statistically significant effect was found.

The Danish Medicines Agency’s evaluation of the meta-analysis was made in connection with a consultation of the Danish Parliament’s Health and Senior Citizen's Committee.