Martin Plöderl, PhD
Clinical Psychologist, Psychotherapist, Salzburg, Austria
Despite decades of research and more than 100,000 patients in clinical trials, it is still debated if the efficacy of antidepressants is clinically meaningful and if prescriptions are justified, given the small benefits and the harms (Jakobsen et al., 2019; Munkholm et al., 2019). Nonetheless, there is an ever increasing prescription of antidepressants. For example, in Germany, antidepressant use almost tripled since the new millennium (Plöderl & Padberg, in press).
There is less debate that antidepressants are often overused, supported by the finding that many who were prescribed antidepressants did not even have a formal psychiatric diagnosis (Mojtabai & Olfson, 2011). Furthermore, for mild depression, most guidelines do not recommend antidepressants as first line treatment (DGPPN, 2015; Malhi et al., 2015; National Institute for Health and Clinical Excellence, 2010). For example, the German guidelines conclude that, for mild depression, antidepressants are not superior to placebo, resulting in an unfavourable risk-benefit ratio because of the harms of AD. The NICE guidelines includes very similar arguments: “Do not use antidepressants routinely to treat persistent subthreshold depressive symptoms or mild depression because the risk-benefit ratio is poor (p. 327)”
In contrast, for moderate to severe depression, many guidelines recommend antidepressants as first-line treatment. This is based on reviews of clinical trials reporting statistically and/or clinically significant antidepressant-placebo differences for moderate to severe depression (Kirsch et al., 2008; National Institute for Health and Clinical Excellence, 2010; Plöderl & Hengartner, 2019). However, the problem with this line of research is that depression severity was mostly based on the mean baseline-level of depression of all patients in a specific trial.
The methodologically superior approach to investigate the efficacy in relation to depression severity is using individual patient data instead of the average depression severity in trials. Such analyses were published only quite recently. We systematically reviewed these studies (Plöderl & Hengartner, 2019) and found that the drug-placebo differences did not or only slightly increased with baseline-levels of depression (see Table 2 in our study). Importantly, even for severe depression, the drug-placebo difference was generally below the threshold of clinical importance. There is no agreement what is a clinical important difference, but a review of different approaches concludes that the lower-bound is about 3-5 points on the typical 17-item Hamilton Depression Rating Scale (Hengartner & Plöderl, 2021). Similar conclusions follow from using a less controversial assessment of depression, the 6-item Bech scale, which only covers core depression symptoms. The reason for the discrepancy between the results from individual patient data and the older group-mean analyses is caused by the so called ecological bias, as demonstrated by Rabinowitz et al. (2016). It should be noted that the small efficacy of antidepressants discussed so far is most likely overestimated due to the methodological biases (Gøtzsche, 2015; Hengartner, 2017; Munkholm et al., 2019). Biases, for example due to unblinding, may depend on baseline severity, so the increasing drug-placebo difference could be an artefact (Gøtzsche & Gøtzsche, 2017).
Taking the novel findings from patient-level analyses for granted raises a problem for the treatment recommendations in guidelines and clinical practice, because the drug-placebo difference fails to be clinically significant not only for mild depression, but also for moderate to severe depression. According to the logic of the guidelines, due to the lack of clinically meaningful efficacy and the harms of antidepressants, this leads to a problematic risk-benefit ratio of antidepressants across the whole spectrum of depression severity. Thus, the recommendation to use antidepressants as a first-line treatment for moderate to severe depression needs to be downgraded or at least questioned. Not surprisingly, our findings were reviewed very critically (we made this public via the OSF https://osf.io/4kh2a/). It will be interesting to see how guidelines-committees will incorporate the inconvenient findings from patient-level data. In their current forms, the guidelines are at odds with the available evidence.
Based on discussions I had so far, my guess is that the inconvenient findings from patient-level studies will not change the treatment recommendations. It seems that changing a prescription practice after decades causes a great deal of dissonance and further threatens the reputation of psychiatry (“they were wrong all the time”). Instead, these findings will likely be challenged with more or less scientific arguments, as I will describe in the following.
A common counter-argument is that we should not base judgments of depression severity on levels of depression assessed in a typical clinical interview (most often the Hamilton Depression Rating Scale). However, accepting this argument means that one invalidates established guidelines which fully rely on depression severity judged by clinical interviews. Interestingly, it seems that clinicians and researchers are often not aware of this serious implication.
A related and frequent counterargument is that depressed patients in clinical trials do not represent patients with true/real/severe depression we see in actual clinical practice. Again, this argument implies a brute attack on the current guidelines which are nearly exclusively based on clinical trials with the usual selection of study patients. Furthermore, studies with patients reflecting clinical practice more realistically did not find that antidepressants work better than in typical clinical trials. On the contrary. For example, the well-conducted PANDA trial did not find a statistically significant drug (sertraline)-placebo difference, no matter how depression was assessed and independent of depression severity (Hengartner et al., 2020; Lewis et al., 2019).1 Smaller efficacy was also observed in the famous STAR*D trial which was designed to be more generalizable to real clinical situations by using minimal exclusion criteria (Gøtzsche, 2015; Kirsch et al., 2018; Pigott, 2015). Finally, in clinical practice, many depressed patients have comorbidities, or are suicidal, but it was found that antidepressants work less well here, compared to depressed patients without comorbidities or suicidality (Perlman et al., 2019). However, this is likely also true for treatment with placebo, but unfortunately, related placebo-controlled randomized trials are lacking, to my knowledge.
Some also claim that typical clinical research ignores different types of depression, and that antidepressants do work for melancholic/biological/endogenous depression. This was indeed found in older clinical trials with first-generation antidepressants. I doubt if this would replicate in more rigorous, pre-registered trials, and it could not be found in more recent studies with common current antidepressants (Arnow et al., 2015; Cuijpers et al., 2017; Maj et al., 2020). In a very recent patient-level analysis, no differential response for depression with and without melancholic features was found (Imai et al., 2021). The authors concluded that “the advantage of the antidepressants over placebo is the same for those with or without such features; their judgement as to the benefits of the antidepressant treatment should, therefore, not be influenced by the presence or absence of melancholic features.”
There is also the frequent claim that the positive effects of antidepressants for severely depressed patients are obvious for those who work in clinical practice. I have respect for clinicians with rich practical experience. After working 16 years in a psychiatric department with many severely depressed patients, I can say to have substantial experience myself. However, I rarely had the impression that antidepressants are effective, long before I became aware of the inconvenient scientific results about the small efficacy. I see patients several times a week, therefore I usually have more and richer „data“ than psychiatrists who see patients less often and only for short check-ups.
However, arguing with clinical experience alone means leaving evidence based medicine. Do we really want to go down this line? Psychiatry is full of horrible treatments believed to be effective (lobotomy, malarial therapy etc). We clinicians have our biases when judging the efficacy of our favourite treatments, and this also applies for antidepressants. For example, it is often claimed that antidepressants take time until efficacy unfolds (2 weeks, 6 weeks, 12 weeks, you chose). This makes it easy to attribute improvements occurring at any time to the drug. Similarly, regression to the mean makes us believe that it was the treatment that led to improvement, when in fact patients would have got better with or without treatment, simply because patients are selected above a certain threshold of depression severity (those who will then be admitted to the hospital, for example). See David Colquhoun’s blog for further explanation and examples of regression to the mean http://www.dcscience.net/2015/12/11/placebo-effects-are-weak-regression-to-the-mean-is-the-main-reason-ineffective-treatments-appear-to-work/. It is also known that we selectively remember evidence confirming our beliefs and ignore falsifying evidence. For these reasons, we should be sceptical (but not dismissive) of clinical experience, and we need evidence based medicine2.
Unfortunately, there are, to my knowledge, hardly any clinical trials including severely depressed patients from psychiatric clinics. However, the assumption that these patients differ in ways that may be crucial for efficacy of antidepressants is unproven. It is unlikely that antidepressants work especially well here, given the disappointing findings from severely depressed patients in typical clinical trials, studies on more representative patient samples, and patients with comorbidities.
Consequently, it is scientifically problematic to assume that antidepressants work for the truly/severely depressed patients based on (potentially biased) clinical experience and in opposition to what the evidence tells, or in the absence of solid evidence. As I have outlined, there are good reasons to assume that antidepressant do not work well here, too. The least one can do is to acknowledge that we don‘t really know how antidepressants work for severely depressed patients. Everything else is not evidence based medicine.
There is an unethical implication from the untested assumption that antidepressants work for the truly depressed patients and the associated perceived necessity to prescribe. This is because antidepressants clearly have harms (e.g., sexual dysfunction, problems with sleep and digestions, nausea, suicidality, etc.), adding extra burden to those already suffering from severe depression. Perhaps there is a lowered ability to tolerate the harmful effects of the drugs, compared to less severely depressed patients? We don’t know. Additionally, it is common that, after an initial nonresponse to the antidepressant, there is an increase of the dose or the addition of another antidepressant. This is not supported by evidence and increases the risk for additional or more severe side effects, leading to more drop-outs in trials (Davies et al., 2019; Dold et al., 2017; Rink et al., 2018). Antidepressants may sometimes cause worsening or chronification of depression, as postulated by psychiatrist Govianni Fava in his theory of iatrogenic comorbidity (Fava, 2020). Some may say that this is purely speculative, but it is no more speculative than the assumption that antidepressants work in real practice. Fava provides good arguments deserving serious consideration. However, in my experience, iatrogenic comorbidity is hardly considered in clinical practice at all and I never experienced that antidepressants were given up as a treatment option after nonresponse – instead, the rule seems to be to always use more medication.
Reactions to the arguments I have outlined here are sometimes harsh, cynical, or “eminence based”. When I provide the evidence, the discussion usually ends, as there are no convincing counterarguments. One psychiatrist closed the discussion with “no matter what the studies say, we will always prescribe drugs“. Sometimes I experience the undertone of “a psychologist cannot know about these issues or judge the evidence”, or accusations such as “you say this because you have conflicting interests as a psychologist”, or even arguments such as “you do harm to patients by arguing like this”. Just recently I received a response from a prominent psychiatrist who said the following: “I suggest you be humble in your statements. This is not a theoretical battle between you and guideline makers. Real patient lives are at stake. I know fully the limitations of meds, but the patients I am thinking of definitely need them [medication] or ECT because they are placebo non responders“. I agree that real life is at stake and that humbleness is much needed.
Where does this leave us in clinical practice? In my experience, treating severe depression is a challenge, and THE therapy is not existing. Classic psychotherapeutic approaches are sometimes inappropriate or impossible. Rather, it is a collaborative effort from all disciplines (psychiatrists, psychologists, nurses, working therapy, etc.), relatives, and of course and foremost, the patient him/herself. Ideally, a bio-psycho-social idiographic working-model of the individual depression will be established and key therapeutic mechanisms can be inferred from that. This needs humble, open minds who are able to dismiss/attune/reconsider hypothesis related to the working-model and treatment efforts.
The use of antidepressants in the sense of a drug-centered model (Moncrieff & Cohen, 2005) could be beneficial, but it seems we are far away from such a guiding theory in clinical practice which follows a rather “blind” routine prescription practice.
To sum up, current evidence does not support the assumption that antidepressants work well even for severe depression, and proponents of antidepressants should at least admit that there is uncertainty. This needs to be acknowledged in the guidelines and in clinical practice.
Arnow, B. A., Blasey, C., Williams, L. M., Palmer, D. M., Rekshan, W., Schatzberg, A. F., Etkin, A., Kulkarni, J., Luther, J. F., & Rush, A. J. (2015). Depression subtypes in predicting antidepressant response: A report from the iSPOT-D trial. American Journal of Psychiatry, 172(8), 743–750. https://doi.org/10.1176/appi.ajp.2015.14020181
Cuijpers, P., Weitz, E., Lamers, F., Penninx, B. W., Twisk, J., DeRubeis, R. J., Dimidjian, S., Dunlop, B. W., Jarrett, R. B., Segal, Z. V., & Hollon, S. D. (2017). Melancholic and atypical depression as predictor and moderator of outcome in cognitive behavior therapy and pharmacotherapy for adult depression. Depression and Anxiety, 34(3), 246–256. https://doi.org/10.1002/da.22580
Davies, P., Ijaz, S., Williams, C. J., Kessler, D., Lewis, G., & Wiles, N. (2019). Pharmacological interventions for treatment-resistant depression in adults. Cochrane Database of Systematic Reviews. https://doi.org/10.1002/14651858.CD010557.pub2
DGPPN. (2015). S3-Leitlinie/Nationale Versorgungsleitlinie Unipolare Depression—Langfassung, 2. Auflage, 5. Version. https://doi.org/10.6101/AZQ/000364
Dold, M., Bartova, L., Rupprecht, R., & Kasper, S. (2017). Dose Escalation of Antidepressants in Unipolar Depression: A Meta-Analysis of Double-Blind, Randomized Controlled Trials. Psychotherapy and Psychosomatics, 86(5), 283–291. https://doi.org/10.1159/000477770
Fava, G. A. (2020). May antidepressant drugs worsen the conditions they are supposed to treat? The clinical foundations of the oppositional model of tolerance. Therapeutic Advances in Psychopharmacology, 10, 204512532097032. https://doi.org/10.1177/2045125320970325
Gøtzsche, P. C. (2015). Deadly psychiatry and organised denial. People’s Press.
Gøtzsche, P. C., & Gøtzsche, P. K. (2017). Cognitive behavioural therapy halves the risk of repeated suicide attempts: Systematic review. Journal of the Royal Society of Medicine, 110(10), 404–410. https://doi.org/10.1177/0141076817731904
Hengartner, M. P. (2017). Methodological flaws, conflicts of interest, and scientific fallacies: Implications for the evaluation of antidepressants’ efficacy and harm. Frontiers in Psychiatry, 8. https://doi.org/10.3389/fpsyt.2017.00275
Hengartner, M. P., & Plöderl, M. (2021). Estimates of the minimal important difference to evaluate the clinical significance of antidepressants in the acute treatment of moderate-to-severe depression. BMJ Evidence-Based Medicine, 2021. https://doi.org/10.1136/bmjebm-2020-111600
Hengartner, M. P., Plöderl, M., Braillon, A., Jakobsen, J. C., & Gluud, C. (2020). Sertraline in primary care: Comments on the PANDA trial. The Lancet Psychiatry, 7(1), 17. https://doi.org/10.1016/S2215-0366(19)30381-5
Hieronymus, F., Lisinski, A., Nilsson, S., & Eriksson, E. (2019). Influence of baseline severity on the effects of SSRIs in depression: An item-based, patient-level post-hoc analysis. The Lancet Psychiatry, 6(9), 745–752.
Imai, H., Noma, H., & Furukawa, T. A. (2021). Melancholic features (DSM-IV) predict but do not moderate response to antidepressants in major depression: An individual participant data meta-analysis of 1219 patients. European Archives of Psychiatry and Clinical Neuroscience, 271(3), 521–526. https://doi.org/10.1007/s00406-020-01173-4
Jakobsen, J. C., Gluud, C., & Kirsch, I. (2019). Should antidepressants be used for major depressive disorder? BMJ Evidence-Based Medicine, bmjebm-2019-111238. https://doi.org/10.1136/bmjebm-2019-111238
Kirsch, I., Deacon, B. J., Huedo-Medina, T. B., Scoboria, A., Moore, T. J., & Johnson, B. T. (2008). Initial severity and antidepressant benefits: A meta-analysis of data submitted to the Food and Drug Administration. PLoS Medicine, 5(2), e45. https://doi.org/10.1371/journal.pmed.0050045
Kirsch, I., Huedo-Medina, T. B., Pigott, H. E., & Johnson, B. T. (2018). Do outcomes of clinical trials resemble those “real world” patients? A reanalysis of the STAR* D antidepressant data set. Psychology of Consciousness: Theory, Research, and Practice, 5(4), 339–345. http://dx.doi.org/10.1037/cns0000164
Lewis, G., Duffy, L., Ades, A., Amos, R., Araya, R., Brabyn, S., Button, K. S., Churchill, R., Derrick, C., Dowrick, C., Gilbody, S., Fawsitt, C., Hollingworth, W., Jones, V., Kendrick, T., Kessler, D., Kounali, D., Khan, N., Lanham, P., … Lewis, G. (2019). The clinical effectiveness of sertraline in primary care and the role of depression severity and duration (PANDA): A pragmatic, double-blind, placebo-controlled randomised trial. The Lancet Psychiatry, 6(11), 903–914. https://doi.org/10.1016/S2215-0366(19)30366-9
Maj, M., Stein, D. J., Parker, G., Zimmerman, M., Fava, G. A., De Hert, M., Demyttenaere, K., McIntyre, R. S., Widiger, T., & Wittchen, H. (2020). The clinical characterization of the adult patient with depression aimed at personalization of management. World Psychiatry, 19(3), 269–293. https://doi.org/10.1002/wps.20771
Malhi, G. S., Bassett, D., Boyce, P., Bryant, R., Fitzgerald, P. B., Fritz, K., Hopwood, M., Lyndon, B., Mulder, R., & Murray, G. (2015). Royal Australian and New Zealand College of Psychiatrists clinical practice guidelines for mood disorders. Australian & New Zealand Journal of Psychiatry, 49(12), 1087–1206.
Mojtabai, R., & Olfson, M. (2011). Proportion Of Antidepressants Prescribed Without A Psychiatric Diagnosis Is Growing. Health Affairs, 30(8), 1434–1442. https://doi.org/10.1377/hlthaff.2010.1024
Moncrieff, J., & Cohen, D. (2005). Rethinking models of psychotropic drug action. Psychotherapy and psychosomatics, 74(3), 145–153.
Munkholm, K., Paludan-Müller, A. S., & Boesen, K. (2019). Considering the methodological limitations in the evidence base of antidepressants for depression: A reanalysis of a network meta-analysis. BMJ Open, 9(6), e024886. https://doi.org/10.1136/bmjopen-2018-024886
National Institute for Health and Clinical Excellence. (2010). Depression: The NICE guideline on the treatment and management of depression in adults. Updated edition 2018. British Psychological Society. https://www.ncbi.nlm.nih.gov/pubmed/22132433
Perlman, K., Benrimoh, D., Israel, S., Rollins, C., Brown, E., Tunteng, J.-F., You, R., You, E., Tanguay-Sela, M., Snook, E., Miresco, M., & Berlim, M. T. (2019). A systematic meta-review of predictors of antidepressant treatment outcome in major depressive disorder. Journal of Affective Disorders, 243, 503–515. https://doi.org/10.1016/j.jad.2018.09.067
Pigott, H. E. (2015). The STAR*D trial: It is time to reexamine the clinical beliefs that guide the treatment of major depression. The Canadian Journal of Psychiatry, 60(1), 9–13. https://doi.org/10.1177/070674371506000104
Plöderl, M., & Padberg, T. (in press). Antidepressiva—Wirksam und gut verträglich? Psychotherapie im Dialog.
Plöderl, Martin, & 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, 19(1), 265. https://doi.org/10.1186/s12888-019-2230-4
Rabinowitz, J., Werbeloff, N., Mandel, F. S., Menard, F., Marangell, L., & Kapur, S. (2016). Initial depression severity and response to antidepressants v. placebo: Patient-level data analysis from 34 randomised controlled trials. British Journal of Psychiatry, 209(05), 427–428. https://doi.org/10.1192/bjp.bp.115.173906
Rink, L., Braun, C., Bschor, T., Henssler, J., Franklin, J., & Baethge, C. (2018). Dose Increase Versus Unchanged Continuation of Antidepressants After Initial Antidepressant Treatment Failure in Patients With Major Depressive Disorder: A Systematic Review and Meta-Analysis of Randomized, Double-Blind Trials. The Journal of Clinical Psychiatry, 79(3). https://doi.org/10.4088/JCP.17r11693
Stone, M., Kalaria, S., Richardville, K., & Miller, B. (2018). Components and trends in treatment effects in randomized placebo-controlled trials in major depressive disorder from 1979-2016. American Society of Clinical Psychopharmacology, Miami. https://www.researchgate.net/publication/334806796_Components_and_trends_in_treatment_effects_in_randomized_placebo-controlled_trials_in_major_depressive_disorder_from_1979-2016