Morgan T. Sammons, PhD, ABPP

On February 24, to minimal fanfare but in the teeth of a fierce email storm on various division listservs, APA’s Council of Representatives voted to approve a Clinical Practice Guideline for the treatment of PTSD. News that this guideline, several years in the making, would be on Council’s agenda awakened multiple listserv dragons from their fitful slumbers. Millions of electrons were expended in an indecisive back and forth as to whether the guideline should immediately be exposed on a hillside and left to perish or whether it represented the ne plus ultra in definitive treatment of PTSD. Leaving aside the fact that many of the aforementioned sleepy-eyed dragons appear not to have read the guidelines before leaping into combat, the order of battle fell out somewhere along these lines:

The anti-guideline forces were arrayed in several camps, one under a banner reading “clinical guidelines are the devil’s handiwork and will be the death of psychological practice,” and another reading “clinical practice guidelines do not reflect the real world, besides, I’m perfectly capable of treating my patients without them” (it was a very large banner). A noisy claque chanted the war cry “we don’t need no stinkin’ guidelines!” On the other side of the valley, the somewhat less numerous pro-guideline forces were huddled behind various redoubts, one bearing a standard reading “any practice that is not solidly evidence-based is unethical” and another saying “Freud is dead, and you’re not so Jung yourself.” Suddenly, on the field separating the two armies, a solitary, somewhat scrawny equerry appeared, carrying a small banneret that read “keep calm, nobody pays much attention to guidelines anyway.” Both forces took immediate aim at the hapless messenger, and the din of battle resounded. At last report, the opponents were locked in remorseless struggle with no apparent victor.

One of the problems with guidelines in the field of mental health is that, regardless of the treatment modality employed, outcomes aren’t terribly good to begin with. Even for the best, most carefully studied treatments, effect sizes are quite modest. Interestingly, effect sizes for pharmacological and psychological treatments, at least for depression, tend to be about the same. This is one of the big problems in combined treatment research, because the multiplied effect of combining two only modestly potent interventions tends, of course, to be rather small. So if you recognize the fact that even your best treatments don’t work terribly well, and then throw in the high degree of nonspecificity of psychological treatments (pharmacological interventions are also largely nonspecific, but that is a topic for another time)—and on top of that add the knowledge that most mental disorders are chronic and relapsing—you come up with a situation where definitive treatment recommendations are elusive at best. Under these circumstances, the value of specific guidelines is reduced.

Another problem with guidelines is that they, like meta-analyses, are only as good as the evidence upon which they are based. Often, this evidence is not very compelling. Studies in any particular area may be few and are likely based on patients not representative of “real-world” patients. They often describe manualized, highly monitored treatments that are hard to implement with fidelity in real-world practice. In modern medicine, particularly modern psychiatry, we also must deal with the issue of fraud. We have decades of published data based on studies that have been funded and heavily influenced by pharmaceutical manufacturers. Study enrollment, design, statistical analysis, and reporting of outcomes have been deliberately skewed to show favorable results for the funder’s drug. Many scholarly articles were ghostwritten–the psychiatric historian David Healy once estimated that 50% of all published articles in psychiatry were ghostwritten by pharmaceutical companies, with an eminent researcher “lending” his or her name as first author. Successive scandals led major medical publishers to adopt more rigorous conflict of interest policies and mandate increased transparency in research design. Beginning in 2007, the registration site helped keep researchers honest by ensuring that study outcomes were established before, not after, patients had been enrolled in a trial. But only trials of drugs or devices that might be regulated by the FDA are required to enroll in Studies of behavioral interventions need not apply. While it can be argued that the stakes are smaller for behavioral interventions, the urgency to publish combined with the bias against publishing negative findings creates incentives to fiddle with results that not all behavioral scientists can resist. In other words, a healthy degree of skepticism all round is a mandatory presupposition for interpreting much of the research that underlies clinical practice guidelines.

I have read the guideline, so you don’t have to—although you should, whether or not you treat patients with PTSD. You can download it at The bottom line up front: It’s not so bad, and clinicians shouldn’t be overly worried about it impinging on their practice. Having said that, we must also recall the wise observation that “nothing is quite so bad as something that is ‘not so bad.’” In that regard, a few things make this particular guideline a bit difficult to deal with. To begin with, as the guideline’s authors hastened to point out, the strength of the evidence supporting various treatments is not overly compelling. As I’ve said, this is a failing common to many guidelines, no different from what we encounter in reviewing guidelines for other disorders. Definitive evidence is often lacking. Clinical recommendations and the yardsticks by which we measure them change as time passes and an evidence base accrues. Nothing is as sure or predictable as we would like it to be. But what makes this particular guideline a bit unusual is that it is not really an original guideline—it is essentially an extrapolation of an earlier guideline published by AHRQ in 2013 ( The authors are candid in acknowledging their debt to the previous guideline, but they have not pushed the evidence base much further than the AHRQ guidance did. So to view this as an original guideline is not quite accurate. The anti-guideline forces can relax a bit—APA is not foisting upon us a new set of standards for treatment of PTSD, they are simply updating and adopting a guideline that already existed.

What does the guideline actually say? I quote:

The panel strongly recommends the use of the following psychotherapies/interventions (all interventions that follow listed in alphabetical order) for adult patients with PTSD: cognitive behavioral therapy (CBT), cognitive processing therapy (CPT), cognitive therapy (CT), and prolonged exposure therapy (PE). The panel suggests the use of brief eclectic psychotherapy (BEP), eye movement desensitization and reprocessing (EMDR), and narrative exposure therapy (NET). There is insufficient evidence to recommend for or against offering Seeking Safety (SS) or relaxation (RLX). For medications, the panel suggests offering the following (in alphabetical order): fluoxetine, paroxetine, sertraline, and venlafaxine. There is insufficient evidence to recommend for or against offering risperidone and topiramate. Based on the updated search, the panel concluded that all of its treatment recommendations, except those for EMDR and NET, were unlikely to change.

Insofar as psychotherapeutic interventions are concerned, the panel has given us a fairly wide range of alternatives, which although somewhat skewed towards the cognitive, represent a rather global, atheoretical set of recommendations. An element of exposure, whether imaginal or in vivo, is generally regarded as an important component of effective treatment of PTSD, but any guideline that recommends “brief eclectic psychotherapy” perforce reflects the nonspecific nature of many psychological treatments.

For pharmacotherapy, the guideline is slightly more problematic. As the authors correctly stated, only the SSRIs paroxetine (Paxil, generic) and sertraline (Zoloft, generic) are FDA indicated for treating PTSD. The evidence base for medications is less robust than that for psychological interventions, but the authors found sufficient data to suggest that fluoxetine (Prozac, generic) paroxetine, sertraline, and the mixed function antidepressant (MFA) venlafaxine (Effexor and generic) be considered. The authors deviated from the AHRQ guidelines in that AHRQ also found limited evidence supporting the use of the anticonvulsant topiramate (Topamax, generic) and the second-generation antipsychotic risperidone (Risperdal, generic). The APA panel did not find this sufficiently compelling to recommend their routine use. This is sensible advice, as the use of second generation antipsychotics for PTSD is most definitely off-label, and the risks of exposure to any antipsychotic must be carefully weighed against any hoped-for benefit. Much off-label use of newer antipsychotics has been for secondary symptoms of anxiety disorders, such as insomnia. Indeed, the use of quetiapine (Seroquel, generic) for PTSD-related insomnia in military populations had for a period of time reached near-epidemic proportions. Prescribing an antipsychotic agent for sedation or insomnia, when less-dangerous, equally effective drug and non-drug alternatives exist, does not generally represent good practice.

On the other hand, the list of pharmacological agents is not optimally presented. Psychologists familiar with these medications know that paroxetine may be challenging for many patients, that the half-life of fluoxetine may limit its use, and that patients with certain medical conditions like hypertension should approach the use of venlafaxine with caution. These nuances are not addressed in the APA guideline. While it could reasonably be argued that specific prescribing information is beyond the guideline’s scope, omitting such commentary does two things. First, it creates a misperception that these agents are more effective than other SSRIs or MFAs, which is not the case. These drugs are simply better studied. Second, it increases the risk that a less-well-informed clinician might simply recommend to a primary care prescriber the use of fluoxetine, paroxetine, sertraline or venlafaxine without further elaboration, and that a prescriber not familiar with these agents would treat them all as clinically equivalent. Perhaps the easiest solution would have been to mimic the position taken by the British National Institute for Health and Care Excellence (NICE) in its guideline for depression. I paraphrase the NICE recommendations:

‘Normally choose a generic SSRI. Consider that SSRIs are associated with an increased risk of bleeding. Consider prescribing a gastroprotective drug in older people who are taking nonsteroidal anti-inflammatory agents or aspirin. Fluoxetine, fluvoxamine and paroxetine have more drug interactions. For people who also have a chronic physical health problem, consider using citalopram or sertraline as these have fewer interactions. Paroxetine is associated with a higher incidence of discontinuation symptoms.’

As can be seen, the NICE guideline, unlike the APA guideline, succinctly provides important information to prescribers and non-prescribers alike. At a minimum, it puts the onus on the prescriber to investigate the appropriateness of any one SSRI or MFA, without instilling the erroneous impression that only the four drugs mentioned are effective treatments. This would have been a wise choice for the APA guideline developers to follow, although their reasoning underlying drug treatment recommendations is worth reading (pp. 67-73 of the guideline).

Now to the fundamental question: Does this guideline advance our understanding of the treatment of PTSD? The honest answer has to be “a little.” It is an objective, fairly thorough analysis of extant guidelines, it incorporates some newer evidence, and its method is rigorous. It is honest in its appraisal of the strength of the evidence underlying its recommendations. Does it provide sure-fire, evidence-based intervention strategies for PTSD? No, but as we qualified earlier, this is not reasonable to expect. More importantly for those fearful that the guideline might mandate inflexible, prescriptive treatments, it does not do this, and the authors’ recommendations are carefully grounded in a recognition of the limitations of the treatments they studied. If we employ Andre Gide’s famous rubric, “Trust those who seek the truth, beware those who have found it,” I think the PTSD guideline will pass muster. It is neither as good as hoped nor as bad as feared. It is a step in a process towards elucidating more effective treatments in a field where certainty is rare and ambiguity abounds. So let’s lay down our arms, break bread with our erstwhile opponents, and understand that at least in this instance, there are no victors and there are no vanquished.