**Guidelines for the pharmacological acute treatment of major depression: Conflicts with current evidence as demonstrated with the German S3 Guidelines**
This project takes a critical view on the German guideline for the acute pharmacological treatment of major depression (see here https://www.leitlinien.de/nvl/html/depression/kapitel-3#section-23). Those familiar with the literature easily will detect problems in the current guideline: outdated studies, missing of several important meta-analyses or almost no discussion of method-biases. Moreover, the current evidence from individual-level meta-analyses is in conflict with the main argument of the guideline (antidepressants work well for severe depression). Thus, our findings are of clinical importance and should be known as soon and widely as possible. Furthermore, The findings are of international relevance, since guidelines of other countries are largely comparable and have the same conflict with the current evidence.
Because our manuscript was already rejected in two major German psychiatric journals, in our opinion for rather dubious reasons, and because of the importance of our findings for clinical practice, we decided to make the manuscripts available via the OSF. Furthermore, we want to make the reviews available in an anonymous way, so readers can follow how critical evidence is treated from within academic psychiatry.
(from the current submission, see https://osf.io/sru38/download):
Several international guidelines for the acute treatment of moderate to severe major depression recommend a first-line treatment with antidepressants (AD). This is based on the assumption that AD obviously outperform placebo, at least in the case of severe depression. However, the efficacy of AD for severe depression can only be definitely clarified with individual patient data, but related studies have been available only recently. In this paper, we point out discrepancies between content of guidelines and the scientific evidence by taking a closer look at the German S3 guidelines for the treatment of major depression. Based on recent studies and a systematic review of studies using individual patient data, it turns out that AD are marginally superior to placebo in moderate as well as in severe depression. The clinical significance of this small drug-placebo-difference is questionable, even in the most severe forms of depression. In addition, the minor efficacy is likely an overestimation of the true effect size due to systematic method biases. In the S3 guideline, a related discussion is lacking despite substantial empirical evidence confirming these biases. In light of recent data and underlying biases, the current recommendations are in contrast to the evidence. The risk-benefit ratio of AD for severe depression may be similar to the one estimated for mild depression and thus could be unfavorable. Downgrading of the related grade of recommendation would be a logical consequence.
**Submission to first German psychiatric journal**
We submitted a manuscript (see here https://osf.io/p2nqz/) early in autumn 2018, and it was rejected after having been reviewed from four peers (reviews see here https://osf.io/5udr3/download).
The first reviewer acknowledged that the paper was clearly written, but that it is still not known which patients benefit from antidepressants and that we do no provide a solution for clinical practice.
The second reviewer acknowledged that the guidelines are outdated and that an update with current evidence is a good thing, but that the selection of our evidence is severely biased and guided by certain interests, not transparent, and perhaps not representative. The biases that we discuss in the paper seem very speculative, and we have missed to discuss other biases that reduce the efficacy of antidepressants. For example, we should have discussed the paper by Fountoulakis and Möller (2010) that refuted the findings by Kirsch et al. (2008), that the findings of RCTs cannot be translated into clinical practice, and that RCTs may include patients who over-exaggerate their symptoms in order to participate in the trials to obtain good medical support.
The third reviewer stated that the new evidence was presented rather one-sided and does not add something substantially new to the discussion. For example, the reviewer discussed the new Cipriani et al. (2018) meta-analysis which found that all antidepressants are superior to placebo. According to the reviewer, we also should have discussed all pro/con aspects of the evidence, and not just repeat old, well-known arguments.
The fourth reviewer made the argument that we do not seem to be aware that guidelines such as the one we criticize involved more than 30 organizations and a complex consensus process. Thus, it is not possible to change guidelines just so, based on the sights/opinions of single persons.
Interestingly, none of the four reviewers discussed the main-argument of our paper and we consider this as an acceptance. Here are some responses to the reviewers claims:
The first reviewer requested a solution for clinical practice. This is an awkward argument. To use an analogue: should those who show the negative effects of carbon-burning to global climate change also present solutions to the problem in order to get their arguments published?
The second reviewer mentioned that the selection of evidence seems highly biased. We agree that our review was not systematic and we changed that in a revision. However we were quite sure that we did not miss relevant evidence, as we follow the literature quite closely (and the systematic review indeed did not change our arguments). A main motive for our paper was that we found the guidelines to be based on a rather selective and outdated selection on the evidence. To prove this, no systematic review was necessary. It was sufficient to point out important studies and meta-analyses that were available at the time of guideline construction/update and likely known to those familiar with the literature, but that were nonetheless missing in the guidelines (see Table 1 in the manuscript that lists the missing but available studies). As reviewer #4 stated, more than 30 organizations were involved in the guidelines. Thus, it is hard to imagine that these important studies were missed. When we discussed the method-biases, we did it briefly, assuming that reviewers are familiar with the studies on the issue. Reviewer #2 stated that these biases are highly speculative. We disagree, several of these biases are based on sound empirical studies (and we outlined this in more detail in the revision). Interestingly, reviewer #2 wanted to have the paper by Pace et al. (2003) discussed (https://doi.org/10.1097/01.CCM.0000054907.33928.48), because, according to the reviewer, it showed that uninsured patients want to be included in trials to obtain treatment and tend to give biased reports of symptoms. However, Pace et al. never discussed this bias in their paper. Instead, they defend the inclusion of uninsured patients in trials, so this is no empirical study about this bias at all. So, it seems that the reviewers comment itself is highly speculative and not based on evidence. Reviewer #2 stated that we were highly one-sided in our selection of critical papers, for example because we did not mention a study by Fountoulakis and Möller (2010) https://doi.org/10.1017/S1461145710000957. This study re-analyzed the meta-analysis of Kirsch et al. (2008), but used a different weighing procedure, and concluded that the efficacy of antidepressants is higher than reported in Kirsch et al. (2008) and even clinically significant for some antidepressants. Bringing this study into the discussion is problematic because it was severely flawed and soon debunked by Huedo-Medina et al. (2011) https://doi.org/10.1017/S1461145711001878. It is irritating that a reviewer is pointing out an outdated and debunked study and we believe that this would not happen in a transparent review system where reviewers are identified and the reviews are published too. Moreover, it seems that reviewer #2 did not get the point of our main-argument: the Kirsch et al. 2008 meta-analysis is not appropriate to answer the question of efficacy of antidepressants in relation to baseline severity of depression. For this, individual-level data is needed, and we present several recent meta-analysis using such data.
The third reviewer argued that we did not raise new arguments and that we discussed the evidence one-sided. For example, the reviewer points out the Cipriani et al. (2018) meta-analysis which found that all antidepressants are superior to placebo. This is striking, as we explicitly mentioned this meta-analysis several times in the manuscript, discussing the lack of clinical significance of all antidepressants in this analysis. Thus, either the reviewer did not read our manuscript well enough or he/she did not get the point of statistical significance versus clinical significance. Moreover, we disagree that we did not present novel evidence, because we presented results from recent patient-level meta-analysis with important conclusions not having been discussed so far, at least not in the guidelines.
The fourth reviewer criticized us for our request to change the guidelines immediately. We did not request that the guidelines should be immediately changed. We only highlighted that, with the current evidence and the guidelines own logic, a downgrading of the recommendations for the treatment of moderate to severe depression with antidepressants is necessary. We concluded that these findings should be acknowledged as soon as possible for the sake of the safety of patients.
Reviewer #4 stated that guidelines involved a consensus of over 30 organizations and they cannot simply be changed just by arguments from a few. Of course we know that development of guidelines is a challenge. On the other hand, this new evidence cannot be ignored.
Rising awareness that guidelines are not in accordance with the current evidence is an important and necessary next step, even more so as it takes much time until guidelines are updated.
But it seems that the reviewers and editors do not think that such a step is important enough. Even if the reviewers were right in some of their criticism of our manuscript, we could have easily defended our points in more detail, or correct some of the issues. None of the four reviewers went into the main argument of our paper.
**Submission to second German psychiatric journal**
We acknowledged several criticism of the reviewers and revised our paper (see here https://osf.io/chmp5/). We included a *systematic* review of the individual level studies (see here https://osf.io/4kh2a/) and a more detailed description of the method-biases. The systematic review only supported our main argument.
We submitted the manuscript on November 28, 2018. The paper never underwent a peer-review, but it took nearly two months (January 17, 2019) until we were informed about the rejection of the paper (see https://osf.io/9u7pm/download). The argument for rejection was that the criteria for a review paper were not fulfilled. According to these criteria "the research question of the review must be clearly defined and the search strategy with their conclusions must be presented in a replicable manner" (translation ours).
We expected that it would be hard to get our inconvenient findings published, but we were still surprised to receive such reactions and reviews. Therefore, we decided to chose an international journal with an open review system as a next step.
**Informing the ÄZQ (Ärztliches Zentrum für Qualität in der Medizin)**
On April 13, 2019, we informed the ÄZQ via eMail about our critical findings, the rejections in the German journals, and that we published our findings on the OSF. We got a reply on April 15, 2019:
"Vielen Dank für Ihre Mail und den Hinweis auf Ihre Publikation. Für die NVL Depression starten wir zeitnah mit der Aktualisierung. In diesem Kontext werden wir systematisch die neue Evidenz zu allen relevanten Fragestellungen recherchieren und alle identifizierten Studien kritisch nach methodischen und klinischen Gesichtspunkten bewerten. In der Zusammenschau aller Ergebnisse prüft die Leitliniengruppe im Konsens, ob sich daraus Änderungsbedarf für die jeweiligen Empfehlungen ergibt. Ihren Hinweis nehmen wir gern auf und berücksichtigen ihn bei der Recherche."
On September 7, 2019, after the publication of our paper in BMC Psychiatry, we informed the ÄZQ again, asking to forward our paper to the panel responsible for updating the guidelines.
**Submission to first international psychiatric journal BMC Psychiatry**
Draft of submitted manuscript here https://osf.io/sru38/, see here for a version with some typos corrected https://osf.io/w9vpx/download
The two reviewers only had minor suggestions and the paper is now published.
Plöderl, M., & Hengartner, M. P. (2019). Guidelines for the pharmacological acute treatment of major depression: Conflicts with current evidence as demonstrated with the German S3 Guidelines. BMC Psychiatry (open access). https://doi.org/10.1186/s12888-019-2230-4