Discontinuing treatment for anxiety disorders??

A recent meta-analysis found that in a variety of anxiety disorders, the risk of relapse was high if antidepressants were discontinued within one year of therapy (see doi.org/10.1136/bmj.j3927.

Details:

--28 studies were included in the analysis (including 4 unpublished one) from 1995 to 2012, with 5233 patients who had anxiety disorder, obsessive-compulsive disorder, or PTSD; and summarized relapse rates in patients who had responded clinically to antidepressant use for their anxiety disorders and then were randomized to continuing the antidepressant vs switching to placebo, assessing:

    --the prevalence of relapse per treatment group

    --the risk of relapse or time to relapse between these groups

    --and whether relapse risk was related to the type of anxiety disorder, type of antidepressant, mode of discontinuation (taper vs abrupt discontinuation), duration of previous treatment, duration of follow-up, whether concurrent psychotherapy was allowed, or whether studies excluded comorbidities

--essentially all of the studies were drug-company sponsored

Results: 

--relapse prevalence (28 studies): 36.4% in the placebo group and 16.4% with continued meds

--overall relapse rate ratio was  3.11 for patients on placebo relative to those who kept on the meds, for studies with duration of treatment of 8-52 weeks

--subgroup analyses (none statistically significant, though several studies had small numbers of patients), comparing stopping vs continuing meds:

    --anxiety type:

        --GAD (6 studies): odds ratio (OR) 4.20

        --OCD (7 studies): OR 2.43

        --Panic d/o (6 studies): OR 2.88

        --PTSD (4 studies): OR 2.43

        --social phobia (5 studies): OR 3.19

    --antidepressant used:

        --SSRI (21 studies): OR 2.86

        --SNRI (3 studies): OR 5.03

    --discontinuation:

        --abrupt (11 studies): OR 2.52

        --taper (17 studies): OR 3.61

    --concurrent psychotherapy used:

        --no (15 studies): OR 2.64

        --yes (13 studies): OR 3.86

    --comorbidity mostly excluded:

        ​--no (10 studies): OR 2.82

        --yes (18 studies): OR 3.20

    --also, no difference related to year of publication, quality of the study per Cochrane criteria, duration of treatment (again, 52 weeks was the max), or duration of followup(only 24-28 weeks).

--and time to relapse was shorter when antidepressants were discontinued (HR 3.63)

--tolerability, dropout rates, and withdrawal symptoms: limited and non-systematic in the studies (!!!). most reported that antidepressants were "well-tolerated" (but, notably, these patients all were taking them and had good responses to them). dropout rates were slightly higher in the placebo group (21.9% vs 17.2%) and perhaps related to withdrawal symptoms (??, though the 4 studies that reported this specifically found no difference, but not sure what med they were using: eg paroxetine and venlafaxine can have very severe withdrawal symptoms)

Commentary:

--most patients with anxiety disorders (57%, per 2008 treatment guidelines) are on meds, with more recent studies finding that antidepressants are the first line drugs. Cognitive behavioral therapy also helps a lot

--they never defined “comorbidity” in their meta-analysis (and I suspect there were somewhat differing definitions in the different individual studies), but from my review of a few of their references, it seems that they mean psychiatric comorbidities, where for example one study (Bruce SE. Am J Psychiatry. 2005; 162:1179) found that recovery was less likely in patients with an array of anxiety disorders but who had  comorbid major depressive disorder, comorbid panic disorder with agoraphobia plus generalized anxiety disorder, comorbid GAD with social phobia, comorbid alcohol or other substance use disorder, etc.

--some notable findings (at least to me): the increase in relapses off meds didn't seem to matter if the patient had panic vs PTSD vs OCD etc (GAD nonsignificantly more), the type of antidepressant used (the increase in those taken off SNRIs may be because not so many people were treated with them; also hard to interpret definitively the relative benefit of SSRI vs SNRI without head-to-head comparisons), whether there was abrupt discontinuation (they did consider that stopping fluoxetine abruptly was really a taper, given its long half-life, but still i would have thought tapering would be better....there were also 3 studies having abrupt cessation of paroxetine, which usually is considered unsafe), whether there was psychotherapy (especially since other studies have found CBT in particular to be so helpful with anxiety disorders; though perhaps they are clumping here other less-effective psych therapies??), or whether there were comorbidities (despite the long-held assumption that this would matter, and did so in the quite good Am J Psych study). also pretty limited data on whether 1 year of meds was better than less, though one small study did find that 1 year was better than 6 months and another did not

--it should also be noted that almost all the studies were drug company sponsored. and of the 4 unpublished studies, 3 found no significant effect of continuing therapy (just goes to show you...).  does make one wonder about the validity of this analysis, but the authors did seem to do a pretty extensive search through the usual sites, including clinicaltrials.gov, and it is probably unlikely that there are enough unfound unpublished studies which would alter the overall results

So,

--it is important to put this study into context: it only applies to patients who had responded to antidepressants, it mostly excluded patients with comorbidities (who typically have more likelihood of relapse off antidepressants), and it does not have very extensive treatment or follow-up intervals (ie, would antidepressants for 18 months or 2 years be even more effective?; would those who stopped earlier have a higher rate of relapse if followed longer?)

--that being said, it was impressive that the rates of relapse off antidepressants was relatively similar independent of the anxiety disorder (GAD vs PTSD etc)

Although the authors' conclusion is "the advice is to continue antidepressants for at least a year​", they did not actually find that stopping after 6 months was any worse.  it does seem reasonable to discuss the option of medication discontinuation with patients at some point after they have been stable (probably at least 6 months from the meager literature available), noting that there is a pretty high relapse rate in those who discontinue meds within the first year. But, on the other hand,  2/3 of them did fine in this study (again, with limited follow-up time).

for prior blogs on anxiety:

benzos-may-not-increase-mortality-risk

early-life-stress-in-mice-changes-in-microbiome-and-later-anxiety