A treatment can show a benefit on its primary endpoint without demonstrating a benefit on every other outcome. The challenge is to describe the result precisely enough that readers can see both its value and its limits. A 2026 phase 3 mirtazapine trial provides an example involving methamphetamine use days, biological testing and tolerability.
What was randomized, and what changed?
McKetin and colleagues randomized 344 adults at six Australian outpatient clinics; 339 received study medication, comprising 172 assigned mirtazapine and 167 placebo. Oral mirtazapine was 30 mg/day for 12 weeks. At week 12, modeled self-reported use days in the preceding 28 days declined by 7.0 with mirtazapine and 4.8 with placebo: a between-group difference of −2.2 days (95% CI, −4.2 to −0.2; P = .02).
Methamphetamine-negative oral-fluid samples were 13.2% versus 12.0%, an estimated 1.3-percentage-point difference (95% CI, −4.7 to 7.2; P = .68). Other overall secondary endpoints were not significant. Drowsiness occurred in 80/172 versus 55/167, and reported weight gain in 17/172 versus 5/167. Medication discontinuation for adverse reactions was 40/172 versus 25/167, without a statistically significant difference. Limitations included self-reported use, imperfect adherence and discontinuation. The authors’ June correction concerned a surname in the visual abstract.
Why is the treatment effect different from the total improvement?
The comparison of interest is the difference between assigned groups, rather than the entire change in one group. Reporting the full decline in the medication group as though all of it were attributable to medication would omit the improvement observed with placebo.
Arithmetic also matters when describing the outcome. A difference expressed in days per 28-day period is not a number of people achieving abstinence. It cannot be turned into an abstinence success rate merely by attaching a percentage sign.
Does the oral-fluid result invalidate the primary finding?
A careful appraisal should first identify what each endpoint measures and how it was analyzed. An interview covering a period of time and specimens collected at particular visits are different observation schemes. Disagreement between their statistical results calls for examination of the methods; it does not, by itself, prove dishonesty, establish that one measurement is error-free, or justify deleting the unfavorable endpoint from a summary.
The appropriate description preserves both results. A statistically significant primary comparison should not be dismissed simply because another comparison is nonsignificant. Equally, the primary result should not be expanded into an assertion that all corroborating tests improved.
How does this fit the supplied research collection?
The supplied 2025 McKetin review considers the limitations of maintenance pharmacotherapies and argues for attention to withdrawal, longer-term function and relapse. That is a narrative research perspective, not the new trial itself. Its emphasis helps frame a useful question: which aspect of recovery did a particular experiment measure?
For a case assessment or research summary, separate changes in use frequency, biological tests, daily function and adverse effects. Do not presume that improvement in one establishes improvement in the others. Nor does assignment to a treatment establish an individual’s impairment status at a separate event.
This article interprets study design and reporting. It does not prescribe mirtazapine or recommend changing medication.
Scientific sources
McKetin R, Shoptaw S, Saunders L, et al. Mirtazapine for Methamphetamine Use Disorder: A Randomized Clinical Trial. JAMA Psychiatry. 2026;83(6):581–589. Primary randomized human treatment trial; published online April 1, 2026. DOI: 10.1001/jamapsychiatry.2026.0159. Publisher full text. Visual-abstract correction.
McKetin R, Kim JH, Turner A, Berk M. Methamphetamine Pharmacotherapy: A Need to Re-focus on the Complex Neurobiological Changes that Occur Both During and After Methamphetamine Use Disorder. CNS Drugs. 2025;39:1061–1070. Supplied narrative review and current-opinion article, published September 14, 2025. DOI: 10.1007/s40263-025-01214-3.