Did Amphetamines Improve Adult ADHD Symptoms Without Improving Treatment Retention? A Historical Cochrane Review

Improvement in ADHD symptom scores and remaining in treatment are different outcomes. A 2018 Cochrane review found short-term symptom benefits from amphetamines in adults with ADHD, but no clear improvement in overall treatment retention and more withdrawals attributed to adverse events. Those findings can coexist without contradiction.

The review also rated the evidence as low or very low quality. That qualification belongs beside the effect estimates, rather than disappearing when a result crosses a statistical significance threshold.

What evidence was available to this review?

Castells, Blanco-Silvente and Cunill searched the literature in August 2017 and included 19 studies enrolling 2,521 adults. All studies compared an amphetamine with placebo; three also included an active comparator. Mean study duration was 5.3 weeks, with studies ranging from one to 20 weeks.

The drugs were dexamphetamine, lisdexamfetamine and mixed amphetamine salts. Lisdexamfetamine is a dexamphetamine prodrug. The review did not study methamphetamine administration. Its therapeutic adult-ADHD findings cannot be generalized to illicit methamphetamine, repeated high-dose exposure or people without ADHD.

The reported study dose ranges were 10.2–21.8 mg/day for dexamphetamine, 30–70 mg/day for lisdexamfetamine and 12.5–80 mg/day for mixed amphetamine salts. These are pharmaceutical dosing descriptions, not measured biological concentrations or dose recommendations.

Why is 2,521 not the denominator for every result?

Trials did not all report the same endpoints or contribute usable data to every analysis. Each pooled comparison therefore has its own study and participant count.

The same number of studies in the final two rows does not mean identical participants or identical denominators. Reusing the review’s total enrollment for every outcome would misdescribe the analyses.

What does symptom improvement mean here?

Negative standardized mean differences favored amphetamines on the symptom-rating scales. A standardized effect expresses the difference in standard-deviation units so that results from different scales can be combined. It is not a 90% symptom reduction, a count of symptoms eliminated or a probability that one individual will respond.

Clinician and patient ratings also came from different evidence sets. The difference between their pooled estimates is not a within-person test proving that clinicians saw more improvement than patients. Neither rating directly measures driving, work productivity or every aspect of daily functioning.

Why was retention not clearly improved?

The pooled retention ratio was 1.06, but its confidence interval included one, the no-difference value. The review therefore did not establish improved retention with amphetamines. This is different from demonstrating exact equivalence or proving that no participant benefited.

Retention integrates many reasons for continuing or discontinuing. A symptom improvement can coexist with side effects, inconvenience or other reasons for leaving a trial. It should not be assumed that better symptom scores necessarily lead to more people completing treatment.

How should the adverse-event withdrawal ratio be read?

A risk ratio of 2.69 describes a relative comparison with placebo for withdrawals attributed to adverse events. It is not a 269% absolute withdrawal probability. The corresponding absolute difference depends on the baseline withdrawal risk in the comparison population.

This endpoint is narrower than experiencing any adverse event. It counts leaving because of an adverse event, not every reported symptom, every serious event or all causes of discontinuation. It also does not mean that the overall chance of staying in treatment must change by the inverse of 2.69: all-cause retention and adverse-event withdrawal are different outcomes.

Why did evidence quality remain limited?

No included study had low risk of bias across every domain assessed. The review identified concerns about blinding because stimulants can produce noticeable subjective effects, attrition, and possible carry-over in crossover studies. Sixteen studies were industry funded, one publicly funded and two did not report funding sources.

Funding alone does not demonstrate that a result is false. The review’s judgments concern confidence in the evidence, incorporating design limitations, precision and other factors. A precise-looking effect estimate should not erase those assessments.

The selected participants and short follow-up also limit generalization. A mean duration of 5.3 weeks cannot establish that benefits and harms remain unchanged for years. Conversely, the absence of long-term evidence in this historical review is not a claim that no later research exists.

What is the appropriate practical conclusion?

This review supported short-term symptom improvement in its adult ADHD populations, alongside uncertain retention benefit and increased adverse-event withdrawals. It did not establish an individual treatment outcome, a diagnostic response test or a drug-concentration impairment rule.

For a scientific or forensic explanation, preserve the specific outcome, denominator and uncertainty. A measured amphetamine result cannot establish that someone experienced the average therapeutic benefit, suffered an adverse event or was functionally impaired at a particular time.

Scientific source

Castells X, Blanco-Silvente L, Cunill R. Amphetamines for attention deficit hyperactivity disorder (ADHD) in adults. Cochrane Database of Systematic Reviews. 2018;(8):CD007813. doi:10.1002/14651858.CD007813.pub3. Full-text systematic review. Historical evidence synthesis; literature search August 2017.