The review by Kalfas on antidepressant withdrawal in JAMA Psychiatry will mislead as it is based largely on people who have taken antidepressants for 8-12 weeks in 11 RCTs-this will under-estimate the risks for millions of long-term users. Thread (1/n)

"However, psychiatrists, psychologists and patients have widely criticised the study due to its reliance on short-term clinical trials lasting only 8-12 weeks. The reality is that people most often use antidepressants for years.".

– for everyone to relapse at the same moment would be a wild co-incidence! In contrast Kalfas study relied on 5 studies using spontaneous adverse effect reporting (known to under-estimate and miss symptoms) versus this systematic simultaneous measurement in Rosenbaum 5/n.

but this is implausible for depression conceived as a relapsing, remitting condition with each person having a typical time between episodes….(4/n).

This is entirely consistent with withdrawal effects including depressive symptoms. Some will argue this means that withdrawal and depression both happening at same time….(3/n).

It showed a large increase in depressive symptoms, driven by a rise in withdrawal symptoms, which completely resolves, along with the w/d symptoms, when antidepressants are restarted after a week. Picture – b=sertraline, c=paroxetine. DESS=withdrawal, MADRS/HRSD = depression(2/n)

Another prob with Kalfas study is the idea that any worsening in mood must be relapse not withdrawal.This is countered by much better evidence in the Rosenbaum study which usefully measured withdrawal effects AND depression scores whilst drugs were stopped and then re-started1/n

Rosenbaum 1998:

Sertraline withdrawal caused overall functional impairment. In other words, clear evidence of much greater effects of withdrawal after longer term treatment (6/n)

For sertraline, 42.4% of people worsened when stopping, 6.1% when continuing their drug. In this study paroxetine withdrawal caused work impairment, relationship problems, decreased overall function and social activities – already in 4 days. (5/n).

When stopping paroxetine 57.1% of patients worsened on a modified DESS scale, 5.7% of people who continued. Measured after just 4 days. Symptoms still increasing (4/n).

Other studies have found near identical results. Michelson (2000) – ‘included’ in Kalfas study, but did not contribute data to the main analysis. 13.5 -15.9 months of drug use before stopping. (3/n).

It shows that 1 week after stopping paroxetine the average increase in DESS was 7.7. For the group who continued the increase was 0. The SMD is 1.5 (5-fold more than what Kalfas found in short-term studies). (2/n).

People have been reasonably asking are there any high-quality studies of wirthdrawal from longer term use. Rosenbaum (1998) is very well conducted. It was ‘included’ in the Kalfas review but not in the main analysis. It meets criteria to do so.(1/n).

There are no randomised trials showing a relationship between cigarettes smoked and lung cancer - these are based on surveys. A survey is a reasonable way to see a duration-response effect and it was strong and backed up by other studies. odd to argue against really.

Thinking of agomelatine: only antidepressant without a withdrawal syndrome. Only antidepressants to show no 'relapse prevention' properties in discontinuation trials (where withdrawal and relapse are not distinguished). Co-incidence?

RT @PsychotropicDC: Presenting a WEBINAR:. Hyperbolic tapering tips: Barriers and enablers.with Mark Horowitz, MBBS, PhD.lead author of the….

Now argues there is no relationship between duration of use and withdrawal from ADs. Means taking them for 1 day or 10 years makes no diff + can be stopped in the same way. OK then. Also 60%+ withdrawal rates in this study did not make it into his paper.

Appropriately skeptical reception of the JAMA study down under: "A large systematic review of research into the process of withdrawing from the drugs was published this week, but it quickly faced backlash from some psychiatrists and experts."

Entirely predictable. I have heard similar from another friend's parent. If the risks of withdrawal are unjustifiably downplayed for long term users and guidance to taper carefully undermined by this paper and its press release then people will be exposed to unnecessary harm.

I don't understand what Dr Lewis means here by 'similar'? In ANTLER they found an SMD of 0.98 for sertraline, 0.52 for citalopram, 0.51 for fluoxetine at 12 weeks, and SMD of 0.41 at 7 months for all. Much more common and long-lasting than Kalfas. Quite different after l-t use.