A new major study finds that antidepressants are effective for the acute treatment of major depression. The study is considered a definitive systematic review and meta-analysis including 522 trials comprising 116,477 participants. This includes unpublished data from pharmaceutical companies, to address the concern that some negative data was being hidden.

In the study all 21 antidepressants studied were more effective than placebo, ranging from 1.37 times as effective to 2.13 times as effective.

This is not surprising to anyone familiar with the evidence – it is, in fact, just a review of that evidence. However, many in the public might be confused because these results seem to contradict previously reported studies that purport to show that for many patient antidepressants are no more effective than placebo.

This confusion, however, is largely due to poor reporting. The key factor that is often missed is the severity of the depression. Often I hear people claim that, “Antidepressants have been shown not to work for depression.” But that is not a meaningful statement because you cannot scientifically refer to the evidence regarding “depression” without qualifying “mild to moderate” vs “severe” depression. That detail often gets dropped in mainstream reporting, and therefore the public consciousness.

This current meta-analysis was looking at the acute treatment for major (severe) depression. It is now overwhelmingly established by the evidence that current antidepressants are indeed effective in treating major depression. This study focused on acute treatment, but they are also effective long term.

Clinical questions like this, however, always have lots of variables to consider. We can ask about subsets of patients, disease variants, various outcomes, different treatment strategies, other treatments, side effects, treatment cost, and treatment duration.

But getting back to disease severity – clearly in the research there is a difference between the effectiveness of antidepressant for major depression (very effective) vs mild to moderate depression (of uncertain efficacy). Is this difference real, and if so what are the cause(s)?

One possibility is that it is simply harder to reach statistical significance with an effect as the effect size gets smaller. You need more and more powerful (larger) studies to detect the smaller effect. With diseases and disorders, effect size is partly determined by the severity of the condition. The milder the symptoms the less room there is for improvement, and therefore the smaller the potential effect size.

So we often see with any clinical research that the milder the symptom you are treating, or the rarer the negative outcome you are trying to prevent, the harder the effect is to detect, and the more powerful studies need to be. It is therefore also usually true that as symptoms become milder eventually you drop below the threshold of detection, at least to statistical significance, and at least as a practical matter.

It is therefore entirely unsurprising that studies of mild to moderate depression have less statistical significance than studies of severe depression. Of course they do. We have to be cautious in interpreting these results. Such results are likely just an artifact of effect size.

However, we also have to consider the alternate possibilities that the mechanisms of mild to moderate depression are different than severe or major depression. This is not implausible. Perhaps mild to moderate depression is most often environmental, and major depression is most often biological.

But this also raises the further question – what are the antidepressants treating? Are they treating the cause(s) of depression, or should they be considered just symptomatic treatment, in which case the cause may not really matter?

These are complex questions that we don’t have definitive answers to. Also, patients with depression (like most types of mental illness) are likely a very heterogeneous group. The entire spectrum likely exists out there – every combination of various biological and environmental contributions, interacting with personality types, life situation, and other medical conditions. Patients therefore need an equally complex approach to their depression, with some combination of counseling, addressing their life situation, and medication when necessary.

At this point what we can say is that the currently available antidepressants do what they are supposed to do – they reduce the symptoms of depression. They don’t work in everyone, and this effect is clearly present for major depression, with progressively smaller benefits that are harder to prove in milder depression.

Antidepressants can be considered, without stigma, in anyone who is having significant difficulty due to depression. And of course should be considered along with the full spectrum of other established treatments, such as the various types of counseling. Interventions also have to be monitored for their effect, and individualized to the patient.

There is no reason, however, to single out antidepressants and to stigmatize them as a treatment option. Also we need to counter the misinformation that “antidepressants don’t work.” They clearly do, in the right context.

This new study does also raise another tangential but important issue – the behavior of the pharmaceutical companies. Often I also see conflation of mistrust of “Big pharma” with the relevant science. It is possible for antidepressants (or any drug) to be an effective and reasonable treatment option, while specific pharmaceutical companies behave badly. We should not throw the science out with with bathwater.

Rather, we need to properly regulate this complex and vital industry. This is happening, although it is an admittedly moving target. The design and conduct of clinical trials for pharmaceuticals are carefully regulated. Now we also have mandated registration for clinical trials, so the results cannot be buried. Further, there is pressure to include any unpublished data in systematic reviews, so the results cannot be skewed by avoiding publication of negative data.

That is what this study did. It is a good trend I would like to see more. It does address a real issue and raises confidence in the results. In the end, that’s what it is all about – doing proper science and being a watchdog on industry so that we can be highly confident in the data and draw reliable conclusions.

It is ironic that the most vocal critics of this system are often those who are simultaneously arguing for reduced regulation of industry and lower standards of science (i.e., alternative medicine).