A new study forces us to ask what antidepressants are actually doing.

Every year in the US, more than 40 million adults receive a prescription for an anti-depressant. That’s 15% of the adult population. In the pandemic era, that number is even higher.

Despite a small, and vocal, group of physicians who argue that anti-depressants are not effective, the empiric data is fairly consistent – the drugs do have a significant, albeit modest effect on a multitude of depression outcomes. Randomized trials show that drugs like SSRIs reduce scores on depression inventories and increase the remission rate above placebo by about 30%. Today – we’ll examine whether that translates into improved quality of life.

Much-publicized concerns about the risk of suicide in those taking SSRIs have been supported in some populations of adolescents, but not in adults.   Indeed, people tend to initiate anti-depressants when they are in their worst mental state – which may explain increased suicide risk in the same time period as well as the general improvement in symptoms even among those exposed to placebo – a regression to the mean effect. A recent Swedish study of more than 500,000 individuals, for example, found the highest risk of suicidal behavior occurred in the 30 days prior to initiation of an SSRI, and was reduced thereafter.

I appreciate that the data is a bit murky here. Overall, these seem like decent, if not game-changing drugs – and have a particular role in the acute treatment of major depression. But this week a new study appearing in PLOS One asks an important question. Do anti-depressants actually improve quality of life? That’s not quite the same thing as asking if they treat depression.

The study leverages the Medical Expenditures Panel Survey – this is a nationally-representative longitudinal survey that tracks a variety of health conditions and, importantly, quality of life as measured by the standard SF-12 instrument. The output of the SF-12 gives a score for both physical and mental quality of life.

The authors looked at a ten-year period from 2005 to 2016.  Each year, around 20 million individuals were diagnosed with depression and about 60% were treated with anti-depressants. We don’t know which anti-depressants, which is a limitation of this study – but given prescribing trends we can be pretty sure the majority were SSRIs.

There were some differences between people who got treated and those who did not.  Treated individuals were more likely to be female, white, married, and of higher socio-economic status. Importantly, those who were treated also had lower quality of life scores at baseline.

And, as multiple other studies have shown, mental health improved over time, regardless of treatment status. That is the way of regression to the mean. Depression is a waxing and waning disease – one tends to get the diagnosis when it is waxing, meaning the natural history (for most, but certainly not all) is improvement.

The big question, of course, is whether quality of life scores improved more in the group who received anti-depressants.

And the answer is – not really.  Even after adjustment for all those baseline differences, mental quality of life scores go up by about 1 point in both groups, and physical quality of life declines a bit.

So – what’s going on here?

There will be some who will use this data to argue that anti-depressants are not effective in the long term – that any use should be time limited, and ideally coupled with psychotherapy. Of course, a recent randomized trial appearing in the New England Journal found a significantly increased risk of relapse when SSRIs were stopped compared to continued among individuals doing well on treatment. So we might need to be careful here.

There are other explanations for the observed data though. Most important is residual confounding. We know there are a variety of differences between those being treated for depression and those not being treated – some of these were measured and adjusted for – and some were not. An important confounder not measured?  Depression severity.

If the treated group had more severe depression, regardless of quality of life, we might not expect them to improve as dramatically as the untreated group. In fact – we might be happy if quality of life stood still if the counterfactual was further decline.

The other consideration is that we may be using anti-depressants too broadly. It is likely that some individuals benefit significantly from the treatment, and others don’t. By prescribing broadly, we dilute the observed effect sizes at the population level. Not every drug needs to be a blockbuster.

And of course, the last thing to remember is that quality of life is a difficult thing to measure – and, as venerated as the SF-12 is, it is still a rather blunt tool to probe the well-being of an individual or a population. What is going on in our heads and in our hearts is quite a bit more profound than what can be captured with 12 questions and a Likert scale.

A version of this commentary first appeared on Medscape.com.