When Nightmares Improve, Reassess the Rest of PTSD

The Psychiatric Record | Clinical Evidence | Tuesday, October 6, 2026

A useful response to treatment can also be easy to overinterpret. If nightmares become less frequent or less distressing, that improvement deserves recognition. The next question is what has happened to the rest of the person's PTSD.

“Doing better on prazosin” is too broad to answer both questions. It could mean fewer nightmare awakenings, a more tolerable night, or a wider improvement in daily life. Those accounts are not interchangeable, and the treatment plan should not depend on guessing which one the patient meant.

Prazosin makes this distinction unusually concrete. The same guideline takes different positions depending on the outcome being treated.

The 2023 VA/DoD guideline weakly suggests prazosin for PTSD-associated nightmares in recommendation 32. Recommendation 18 weakly suggests against prazosin for PTSD overall; its discussion specifies monotherapy. A recommendation about the whole disorder should not erase the narrower nightmare recommendation, or vice versa.

The nightmare evidence is mixed. The guideline reports that systematic reviews found benefit even when they included a large VA Cooperative trial in which prazosin did not improve nightmares or global PTSD symptoms compared with placebo. It rates confidence as low because of small samples, heterogeneity and bias. It does not promise that an individual will respond.

Recommendation 33 finds insufficient evidence to recommend for or against imagery rehearsal therapy, exposure relaxation and rescripting therapy, imaging rescripting and reprocessing therapy, or NightWare for PTSD-associated nightmares. That is a statement about the evidence, not a finding that these treatments do not work.

For broader PTSD treatment, the guideline's position is stronger. Recommendations 7 and 8 are strong recommendations for specified individual trauma-focused psychotherapies (cognitive processing therapy, EMDR and prolonged exposure), preferred over medication when both are available and feasible. Patient preference and access remain part of the choice.

These distinctions permit a more accurate conversation than either “prazosin works for PTSD” or “the guideline is against prazosin.” The clinician needs to name the treatment target before applying either conclusion.

A follow-up can begin with the problem the patient wanted to change. Are nightmares occurring less often? Are they less distressing when they occur? What does the patient now do differently the following day?

Those questions are ordinary clinical examples, not a proposed new rating scale. They help separate a change in the target symptom from an assumption about overall recovery.

The patient's account may contain both improvement and substantial remaining difficulty. A quieter night does not make daytime concerns unimportant. Conversely, persistent PTSD symptoms do not make the nighttime improvement worthless. The point is to preserve both findings long enough to make a proportionate decision.

Consider the difference between documenting “PTSD improved” and documenting the particular sleep-related change the patient reports, then separately assessing the remaining symptoms and functioning. The latter leaves less room for an isolated response to become an unintended statement of remission.

This is especially important when the medication is the most visible part of the plan. A prescription has a name, a dose and a refill date. The broader treatment needs equally explicit goals, even when access is difficult or the patient is still considering the options.

Prazosin's US prescribing information lists hypertension as its indication. Using it for PTSD-associated nightmares is off-label. That status is compatible with a guideline suggestion, but it should remain visible rather than being mistaken for an FDA-approved PTSD indication.

The label warns about syncope, including after the initial dose, rapid dose increases or introduction of another antihypertensive. It also describes dizziness and lightheadedness, and additive blood-pressure lowering with other antihypertensives or PDE-5 inhibitors. These considerations require an individualized medication and safety review; this article does not provide a titration regimen.

A report of fewer nightmares and a report of troublesome dizziness can both be true. Neither should disappear into a single “better” or “worse” judgment. The decision needs to account for the benefit the patient values and the cost of obtaining it.

The same discipline applies when there is little improvement. Clarify what was tried and what changed before converting an unsatisfactory result into a conclusion about every available treatment for PTSD.

The practical task is to maintain two lines of follow-up: the nightmare target and the remaining PTSD treatment needs. They can inform each other without being collapsed into one outcome.

The VA/DoD document is a 2023 guideline, based on evidence reviewed through May 2022, not a new 2026 trial. Its recommendations guide a discussion; they do not determine an individual's response or replace reassessment. A new symptom, adverse effect or change in preference can alter what is appropriate.

The editorial implication is straightforward: welcome a meaningful improvement, describe it precisely, and keep asking what remains untreated. A successful symptom-targeted intervention should make the next conversation more specific. It should not end it prematurely.

Educational Disclaimer: The Psychiatric Record provides general educational information for psychiatric and mental-health professionals. Content does not constitute medical, legal, regulatory, compliance, billing, or other professional advice; does not establish a standard of care; and is not a substitute for independent professional judgment. Appropriate assessment and treatment depend on the individual, setting, and current authoritative guidance. This article is not a diagnostic instrument or medication regimen. Patients should not change medication on the basis of this article.