Strangely enough, there are only two medications FDA-approved for PTSD–Zoloft (sertraline) and Paxil (paroxetine). This is despite the fact that PTSD is a common problem, with a lifetime prevalence of between 5 and 10%, and can be extremely debilitating.
Contrary to popular opinion, Zoloft’s PTSD indication is not limited to women alone. It is, in fact, FDA-approved for PTSD in both sexes. There was pre-release buzz about the FDA considering limiting the approval to women, because the efficacy data was more compelling for women. Ultimately, the FDA decided that this was likely just an artifact of the relatively few men who were enrolled in the clinical trials, so the more broad approval was granted.
How effective is Zoloft for PTSD, and what do the data look like? Pfizer submitted the results of four clinical trials as part of its FDA filing; Zoloft bested placebo in only two of them (according to the package insert), but this was enough for the FDA, presumably because at that time there was no medication approved for PTSD whatsoever.
Of the two Zoloft studies showing efficacy (JAMA 2000; 283:1837-1844, and Arch Gen Psychiatry 2001; 58:485- 492), we’ll describe one in detail, because this will give you a good feel for the type of data upon which you are making your decisions to prescribe a given drug for a PTSD patient. The Paxil data comes from studies with very similar design, but with some intriguing differences which we’ll mention later in this article.
In one study (Arch Gen Psych 2001; 58:485-492), researchers randomized 100 PTSD patients to Zoloft (flexible daily doses from 50-200 mg QD, mean dose 146 mg QD) and 108 patients to placebo. The trial lasted 12 weeks, a good length that has become the standard in PTSD trials.
As in any clinical trial, it’s important for you as the reader to ask how similar the patients in the trial are to patients you are likely to see in your practice. The answer usually comes in a section titled “demographic and clinical characteristics of subjects” and while it may seem terrifically boring to read, it’s important information. In this study, most of the patients were women (84% of the Zoloft group), most were white (83%), the original traumatic event was typically a physical or sexual assault (64%) occurring a long time ago (over 18 years ago on average). These patients had been suffering PTSD for an average of 12 years. Many (40%) were suffering major depression at entry, about 20% had an anxiety disorder, and 25% had a remote history of drug or alcohol use. Crucially, patients with active drug or alcohol problems were excluded, limiting the real world generalizability of the findings.
Patients were allowed to be in therapy during the study, but only if they had started it at least three months prior to entry. However, patients were not allowed to receive cognitive behavioral therapy at all, presumably because this treatment is so effective for PTSD (see the related articles in this issue) that it would have confounded the efficacy data.
The minutiae to come in the paragraph you are now beginning to read is presented not merely to convince you that TCR has actually read these articles (it has), but rather to give you a clearer idea of what constitutes “efficacy” in PTSD trials. You are probably familiar with the Hamilton Depression Scale, the gold standard outcome measure for most antidepressant trials. The world of PTSD has its own “Hamilton,” which is the CAPS-2 (Clinician Administered PTSD Scale Part 2), a 17-item subsection of which is the usual main outcome measure for PTSD clinical trials. The maximum possible score is 136, and a score of at least 50 is generally required for study entry. In this study, the average CAPS- 2 score before treatment was 73.9 in the Zoloft group, which translates to “severe” PTSD symptomatology. The efficacy measure was the reduction in CAPS-2 score from baseline to endpoint. After 12 weeks, the Zoloft group went from 73.9 to 40.9 (a 33 point, or 45% improvement); the placebo group went from 73.5 to 47.3 (a 26.2 point, or 36% improvement).
How significant are these numbers? Well, in the December 2003 issue of TCR, we went through this kind of exercise in looking at the controversial issue of the robust placebo response rate in antidepressant trials. The bottom line was that 75-80% of antidepressant response rates in clinical trials are actually due to placebo response. The numbers here are similar. Divide the 36% improvement due to placebo by the 45% improvement due to Zoloft, and you arrive at 0.8, indicating that 80% of the effect of Zoloft in these patients may be accounted for by the placebo effect. Is that enough of an extra therapeutic effect to justify prescribing a low side effect medication? Many psychiatrists would say “yes,” but you’ll have to be the judge.
The CAPS-2 wasn’t the only efficacy measure used in this study; the other primary measures also beat placebo, but not by much–in each case, the placebo response accounted for at least 70% of the drug response. A disquieting result was that Zoloft, usually an effective antidepressant, did not do any better than placebo in terms of improving scores on the Hamilton Depression Scale, an outcome the authors were at a loss to explain.
How about Paxil (paroxetine)? GlaxoSmithKline, the manufacturer, submitted three studies to the FDA; all three beat placebo on the CAPS-2, but only the two more impressive of the three were actually published (J Clin Psychiatry 2001; 62:860-868, and Am J Psychiatry 2001; 158:1982-1988). The Paxil PTSD studies were similar in methodology to the Zoloft studies, with one potentially critical difference: the Paxil studies explicitly excluded patients who were receiving disability or were involved in litigation relevant to PTSD. Most clinicians would agree that such patients are very hard to treat because of the secondary gain attached to their symptoms. And indeed, Paxil appears to shine a bit more brightly than Zoloft, with a more impressive separation from placebo and a robust effectiveness for depressive symptoms. One of these studies was a welcomed rarity in research: a fixed dose study comparing Paxil 20 mg QD, Paxil 40 mg QD, and placebo. Both doses of Paxil performed equivalently, which is good news because SSRI side effects are dose dependent. Note that these
Note that these were not headtohead studies, of course, so we can’t conclude anything about the relative efficacy of the two contenders. In fact, no large headtoheads have been published comparing SSRIs for PTSD. Most likely, they all work for this condition, as there are controlled data endorsing fluoxetine (J Clin Psychiatry 2002; 63:199-206), and open data for Celexa (Int J Neuropsychopharmacol 2000; 3:135-140), and fluvoxamine (Drugs 2000; 60:925-54).
TCR VERDICT: Though not spectacular, SSRIs work