Reduced Use vs Abstinence: A Better Way to Read Stimulant Use Disorder Trials?

Why Abstinence May Be Too Narrow for Stimulant Trials

Stimulant use disorder remains one of the clearest treatment gaps in addiction medicine. Cocaine and methamphetamine use are associated with cardiovascular events, psychiatric crises, infectious risks, overdose, sleep disruption, violence exposure, and social instability. Yet unlike opioid use disorder, there is still no approved pharmacotherapy that reliably changes the standard of care. Against that background, medication trials often face a harsh binary judgment: did the patient become abstinent or not?

Complete abstinence is a clean endpoint. It is clinically meaningful, easy to communicate, and can be verified through urine toxicology. For regulators and trialists, it offers an apparently rigorous standard.

However, abstinence may be too narrow as the only way to read efficacy. Addiction treatment is not always a straight line from uncontrolled use to total abstinence. Many patients move through intermediate states: fewer days of use, shorter binges, less compulsive use, fewer risky episodes, better sleep, more treatment engagement, and greater stability. Those changes may not satisfy an abstinence endpoint, but they can still alter risk.

The JAMA Psychiatry meta-analysis by Amin-Esmaeili and colleagues enters exactly this debate.¹ It does not present a new medication breakthrough. Instead, it asks whether stimulant pharmacotherapy trials may be missing clinically relevant signals when they focus too heavily on complete abstinence. The value of the paper lies in that shift: it asks not only whether a drug produces abstinence, but whether it reduces use in a way that could matter for patients and public health.

What the Meta-Analysis Actually Did

The authors conducted a one-stage individual participant data meta-analysis of 12 US National Institute on Drug Abuse–sponsored multisite randomized clinical trials. The original trials were conducted between 2001 and 2011 and tested medications for cocaine or methamphetamine dependence.

The pharmacotherapies included topiramate, bupropion, modafinil, ondansetron, tiagabine, cabergoline, reserpine, selegiline, and baclofen. Each was compared with placebo in double-blind trials. All participants received cognitive behavioral therapy regardless of medication assignment, which is important because the medication signal was evaluated within a structured behavioral treatment context.

1. Pool the evidenceIndividual-level data from 12 randomized trials
2. Compare groupsActive medication versus placebo in double-blind designs
3. Measure two outcomesReduced use as primary; abstinence as secondary
4. Estimate effectsPooled and medication-specific statistical analyses

Study-design summary based on the published abstract. NIDA: National Institute on Drug Abuse.

The pooled dataset included 2,000 participants: 1,134 in cocaine trials and 866 in methamphetamine trials. The mean age was 39.7 years, and 29.4% of participants were women.

How the two outcomes were defined

That ordering matters. Reduced use was not treated as a consolation prize after abstinence failed. It was intentionally evaluated as an outcome in its own right. This is the paper’s most important conceptual move.

What Changes When Reduced Use Counts

The results were cautious rather than dramatic. Overall, 31.2% of participants achieved the reduced-use outcome at the trial endpoint, while 13.3% achieved abstinence. No statistically significant differences were found between the pooled active-treatment and placebo groups for either outcome, overall or after stratification by stimulant type. Effect-size estimates were nevertheless consistently larger for reduced use than for abstinence. In medication-specific analyses, cabergoline showed a positive signal for reduced cocaine use.

The endpoint chosen in a clinical trial does more than measure a result. It also shapes the field’s working definition of success. If complete abstinence is the only meaningful outcome, stimulant medication trials become brutally binary: a patient either stops using entirely or the treatment is counted as ineffective.

That standard has clarity and clinical appeal because abstinence usually implies lower exposure to stimulant-related harm. Yet stimulant use disorder is often episodic, relapsing, and entangled with psychiatric symptoms, housing insecurity, trauma, pain, employment instability, and polysubstance use. In that setting, a reduction from frequent to occasional use may not equal recovery, but it may still represent movement toward safety and control.

This is the study’s outcome framework, not a universal clinical recovery scale. Frequency alone cannot describe severity, risk, or recovery for an individual patient.

A person who uses cocaine or methamphetamine on fewer days may experience fewer intoxication episodes, less sleep deprivation, lower cumulative cardiovascular strain, fewer high-risk sexual or injection-related events, and more days available for work, treatment, family responsibilities, or medical care. Continued stimulant use can still be dangerous, including because of possible exposure to fentanyl-contaminated supplies. From a harm-reduction and treatment-development perspective, however, the difference between frequent and low-frequency use is not trivial.

This is where the meta-analysis changes the interpretive frame. More than twice as many participants achieved reduced use as achieved abstinence at the endpoint. That does not prove reduced use is sufficient. It suggests that abstinence captures only one subset of improvement.

For medication development, that distinction can matter. A compound that modestly reduces use may appear useless if judged only by abstinence. If reduced use is prespecified and clinically validated, the same compound might warrant further study, especially in a field with no approved pharmacotherapies. The point is not to lower standards so weak medications look better. It is to align trial endpoints with clinically meaningful change.

Why measurement still needs to be rigorous

Reduced use cannot become a vague category. It needs a clear threshold, reliable measurement, and evidence that the threshold predicts outcomes that matter. Ideally, reduced-use endpoints should be linked to fewer overdoses, fewer emergency visits, lower psychiatric crisis rates, improved cardiovascular outcomes, better functioning, and greater treatment retention.

The distinction between self-report and toxicology also matters. In this meta-analysis, reduced use was based on self-report, while abstinence was verified through urine toxicology. Self-report can be clinically useful, but it is vulnerable to recall error and social-desirability bias. Future trials could strengthen interpretation by combining self-report with toxicology, digital or ecological assessments where feasible, and clinically grounded outcomes.

The most balanced reading is therefore neither pessimistic nor permissive. The study found no significant pooled medication effect, which limits any claim of efficacy. It also showed why reduced use may reveal information that abstinence alone does not capture. In a difficult therapeutic area, that matters for how the next generation of trials is designed.

How Future Trials Should Read Efficacy More Carefully

Stimulant use disorder trials should not abandon abstinence. It remains important for patients who want to stop using, for clinicians trying to reduce risk, and for regulators seeking clear evidence. It is also a useful safety-oriented endpoint when stimulant use is associated with psychosis, severe cardiovascular disease, pregnancy, injection-related risk, or repeated overdose exposure.

Nevertheless, abstinence should not be the only lens. A trial that fails to produce abstinence is not automatically meaningless. A medication that reduces use is not automatically a cure. Both statements can be true.

• Prespecify both outcomes. Define abstinence and reduced use before the trial starts.
• Use transparent thresholds. Do not invent a more favorable cutoff after seeing the results.
• Validate clinical meaning. Link use reductions to health, safety, functioning, and quality of life.
• Combine measurements. Use credible self-report alongside toxicology and other feasible measures.
• Report treatment context. Make clear what behavioral care participants received.
• Keep claims proportional. A signal for further study is not proof of an effective treatment.

Reduced-use thresholds should be clinically justified and validated against outcomes patients and health systems care about: functioning, overdose, emergency care, psychiatric symptoms, sleep, retention, quality of life, and reduced exposure to contaminated drug supplies.

The field also needs to avoid a rhetorical trap. Reduced use should not become a way to rescue disappointing trials by changing the rules after the fact. Its value depends on discipline: the threshold must be transparent, the measurement strategy credible, and the clinical meaning demonstrated rather than assumed.

The JAMA Psychiatry meta-analysis does not announce a pharmacological breakthrough for stimulant use disorder. Its contribution is quieter and more methodological. It shows that abstinence-only interpretation may be too blunt for a disorder in which partial reductions in use can still carry clinical and public-health significance.

A better reading of stimulant trials would treat outcomes as a continuum. Complete abstinence remains a major goal; reduced use may be an intermediate but meaningful signal. In a field where medication development has repeatedly disappointed, that broader outcome logic could help distinguish truly ineffective drugs from interventions that produce incomplete but potentially useful benefit.

Questions Readers Often Ask