Methodological Flaws, Conflicts of Interest, and Scientific Fallacies: Implications for the Evaluation of Antidepressants’ Efficacy and Harm

In this 2017 narrative review, Michael Hengartner discusses the evidence assessing the effectiveness and harms of antidepressants. He finds the evidence of benefits insufficient and the harms sufficiently great to argue treatment should not be recommended for the majority of patients on an evidentiary basis.

Significant biases confound the perceived benefits of antidepressants in acute therapy. Industry-funded research consistently shows greater treatment efficacy than research funded and conducted entirely by individuals without conflicts of interest. This is at least partially due to publication bias, as industry-funded antidepressant trials with less positive results often either remain unpublished or are distorted to make the results seem positive. In sum, this creates a published literature base that demonstrates greater drug efficacy than a complete view of the literature would support.

Some consistent methodological choices further distort results. One key area is trial selection criteria, which generally exclude patients with major depression and comorbidities or suicidal ideation; such selective practices separate antidepressants in a trial setting from their real-world clinical use, potentially exaggerating their average benefits (by excluding the hardest-to-treat patients) and downplaying their risks (by excluding the patients most susceptible to increased suicide risk and other side effects).

Despite these biases, all of which would inflate the mean perceived benefit attributed to antidepressant treatment, no meta-analysis has shown an effect size for antidepressants that reaches clinical significance (Hengartner cites evidence that an improvement of fewer than 7 points on the HAMD scale would not be recognized as an improvement in depression symptoms). Some researchers have argued that when antidepressants are compared to active placebos (which intentionally introduce side effects to reduce patients’ opportunity to discover they are in the placebo group), the benefits disappear entirely.

Research testing the impact of antidepressants in long-term therapy is consistently distorted by the fact that long-term antidepressant use can cause durable neurochemical changes in the brain and dependence. Depression relapses in discontinuation trials, then, may often be discontinuation or withdrawal symptoms. Long-term observational studies (which provide the only data on the impact of antidepressants over many years) suggest outcomes for antidepressant users receiving continuous maintenance therapy are no better than those who never received antidepressants in the first place.

Finally, multiple meta-analyses have found antidepressants increase risk of suicide attempts, self-harm, and suicidal ideation (with a risk ratio of between 2.5 and 4.3). As noted above, this effect may in fact be understated, as patients at the greatest risk of suicide are generally excluded from antidepressant trials. Similarly, multiple observational studies have shown antidepressant use is associated with increased mortality, particularly in older patients.

Hengartner concludes:

My reading of the literature is that some patients may benefit from acute pharmacotherapy, but on average clinical benefits are debatable and should be weighed against adverse side effects. Continuation and maintenance therapy is not recommended due to an apparent lack of clear clinical benefits, coupled with a possibly increased vulnerability to chronic depression, increased suicide risk, and, in particular in older adults, higher all-cause mortality.