checkedPsychopharmacology and Psychiatry Updates2026-11-02last verified 2026-11-03

did prazosin beat placebo 64% to 27% in the 2013 soldier trial, with a number needed to treat of 2.7?

verdict: unverifiable
no method, no file, or the data cannot answer it
“The score is showed 64% responders on Prazoson, 27% on placebo, number needed to treat 2.7,”

a guest, Dr. David Osser, psychiatrist treating PTSD in the VA system said on Psychopharmacology and Psychiatry Updates, 2026-11-02, at 06:31. hear it in context → the speaker cited: the 2013 Raskin study.

the episode: Prazosin for PTSD Nightmares (at 06:31). we link the show's own page; we do not re-host audio.

first, the part that holds

The trial is real and the speaker describes it correctly: a 2013 randomized, placebo-controlled trial in 67 active-duty soldiers back from Iraq and Afghanistan, which found prazosin beat placebo on global function measured by the clinical global impression scale. The number needed to treat is also right if the rates are right: 64% minus 27% leaves a gap of 37 points, and 1 divided by 0.37 is 2.7.

what we found when we went and looked

The 2013 trial's PubMed abstract confirms 67 soldiers, 15 weeks, and a placebo comparison, with clinical global impression as one of three primary outcomes. It says prazosin "was effective for … global function" but gives no percentages. We could not open the full paper, which has the responder rates, in this session: the journal's site and every copy we tried blocked us. So we could not read the 64% vs 27% figures ourselves. The number needed to treat of 2.7 is the speaker's own arithmetic from those two rates, and it is correct. A larger 2018 trial of 304 veterans with chronic PTSD found no difference between prazosin and placebo on the same global-change measure. A 2020 meta-analysis of six trials still found a small pooled benefit.

assertion by assertion

assertionverdictwhat the record says
the trial was published in 2013 and enrolled 67 participantsconfirmedPubMed lists it in the American Journal of Psychiatry, 2013, volume 170(9), pages 1003-10: a 15-week randomized controlled trial with 67 active-duty soldiers with combat PTSD, randomly assigned to prazosin or placebo. A PMC systematic review describes it the same way. [1][2]
64% responded on prazosin versus 27% on placebo by clinical global impressionunverifiableThe abstract we read names global function (clinical global impression) as a primary outcome and says prazosin was effective for it, but it gives no responder percentages, confidence intervals or definition of a responder. The full text is the only place those appear, and it returned 403 on every route we tried. PMC copies were behind a captcha. Search snippets attributed to the journal page showed 64% vs 27%, but we did not read that page and cannot cite it. The direction of the result is confirmed; the size is not checked. [1]
the number needed to treat was 2.7confirmedAs arithmetic, it checks out: 1 / (0.64 − 0.27) = 1 / 0.37 = 2.70. Nothing we read shows the paper reporting a number needed to treat, so this looks like the speaker's own calculation. It is only as solid as the two rates it comes from, which we could not read in the record. [1]
a later, larger trial found the same global-change benefit (context, not said in the quote)falseThe 2018 multisite trial randomized 304 veterans with chronic PTSD (152 to prazosin, 152 to placebo). At 10 weeks the between-group difference on clinical global impression of change was 0 (95% CI −0.3 to 0.3; P=0.96), with no differences on nightmares or sleep quality either. A 2020 meta-analysis of six trials (429 patients) still found a modest pooled benefit on overall PTSD symptoms (SMD −0.31, 95% CI −0.62 to −0.01). Its authors put the large trial's null result down partly to a large placebo response. [3][4]

both directions

for the speaker: The PubMed abstract confirms the speaker's description: 2013, 67 active-duty soldiers, a placebo-controlled design, and global function by clinical global impression as a primary outcome on which prazosin was "effective". A PMC systematic review describes the trial the same way and reports significant improvement on all three primary outcomes, including overall clinical status. The number needed to treat of 2.7 follows correctly from the rates the speaker gave. The speaker also got the direction right: prazosin, not placebo, came out ahead on global improvement.

against: Nothing we read contains the 64% and 27% figures, so the size of the effect rests on a paper we could not open. Citing this one trial alone also leaves out what came next. The largest prazosin trial for PTSD, 304 veterans in 2018, found no difference at all on the same global-change measure (difference 0, P=0.96), or on nightmares or sleep. Pooled across six trials, the benefit on overall PTSD symptoms is small (SMD −0.31, with a confidence interval reaching nearly zero). An NNT of 2.7 from 67 soldiers is a single-trial figure with wide uncertainty. It is not a settled estimate of how often prazosin helps.

what this cannot say

Our UNVERIFIABLE verdict reflects what we could reach, not a gap in the evidence. The responder rates are in the full paper. The journal blocked automated access, and the PMC copies we tried were behind a captcha in this session. Someone with journal access could settle the 64%/27% question in a minute. We also could not check how the paper defined a "responder" (which CGI change scores counted), and that choice moves the percentages. On the data side, 67 participants gives wide confidence intervals around any responder rate, so an NNT from them is imprecise. The trial was in active-duty soldiers; the null 2018 trial was in older veterans with chronic PTSD who were clinically stable, so the two populations differ. The meta-analysis we cite pooled only six trials, with notable variation between them.

sources, each one fetched and read

  1. A trial of prazosin for combat trauma PTSD with nightmares in active-duty soldiers returned from Iraq and Afghanistan. American Journal of Psychiatry, 2013;170(9):1003-10. PMID 23846759 What we read there: 15-week randomized controlled trial; 67 active-duty soldiers; primary outcomes CAPS nightmare item, Pittsburgh Sleep Quality Index, and Clinical Global Impressions; "Prazosin was effective for trauma nightmares, sleep quality, global function, CAPS score, and the CAPS hyperarousal symptom cluster." The abstract gives no responder percentages. Fetched 2026-11-03 (HTTP 203). https://pubmed.ncbi.nlm.nih.gov/23846759/
  2. Using prazosin to treat posttraumatic stress disorder and associations: a systematic review (PMC8169333) What we read there: "A clinical trial published in 2013 ... monitored for 15 weeks active US Army Iraq and Afghanistan military service members diagnosed with PTSD. A total of 67 patients were administered the CAPS, PSQI, and CGI questionnaires. There was significant improvement over the study's three primary outcomes." No responder percentages or NNT given. Fetched 2026-11-03 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC8169333/
  3. Trial of prazosin for post-traumatic stress disorder in military veterans. New England Journal of Medicine, 2018. PMID 29414272 What we read there: 304 randomized (152 prazosin, 152 placebo); at 10 weeks no significant difference in CGIC score (between-group difference 0; 95% CI −0.3 to 0.3; P=0.96), nightmares or sleep quality; "prazosin did not alleviate distressing dreams or improve sleep quality." Fetched 2026-11-03 (HTTP 203). https://pubmed.ncbi.nlm.nih.gov/29414272/
  4. Prazosin for treatment of post-traumatic stress disorder: a systematic review and meta-analysis. PMID 32362287 What we read there: Six randomized placebo-controlled trials, 429 patients; overall PTSD SMD −0.31 (95% CI −0.62 to −0.01), nightmares SMD −0.75 (−1.24 to −0.27), sleep quality SMD −0.57 (−1.02 to −0.13); in the largest trial "a relatively large placebo effect size, particularly for nightmares, contributed to no treatment differences." Fetched 2026-11-03 (HTTP 200). https://pubmed.ncbi.nlm.nih.gov/32362287

speakers are described by role, not named. one sentence is quoted under fair use, with attribution and a link to the original at the timestamp; we do not re-host audio or video. nothing on this page is a reason to start, stop or change a medication. if you have the file behind anything marked unverifiable, the research board is where people can answer. adam