A prescription-amphetamine dose association can be clinically important without establishing a universal safe dose or predicting a particular patient’s outcome. A 2024 case-control study is especially instructive because its headline result depends on three details: the selected psychiatric population, a combined psychosis-or-mania outcome and a standardized dose scale.
What was actually compared?
Moran and colleagues compared 1,374 patients aged 16–35 hospitalized for incident psychosis or mania with 2,748 patients undergoing an initial psychiatric hospitalization for other reasons at McLean Hospital, using 2005–2019 records. Past-month prescription amphetamine exposure was associated with an adjusted odds ratio of 2.68 (95% CI 1.90–3.77) relative to no exposure. Above 30 mg/day dextroamphetamine equivalents, the estimate was 5.28 (3.06–9.13). Moran et al., 2024.
The investigators multiplied mixed amphetamine salts doses by 0.75 and lisdexamfetamine doses by 0.3. Thus, “30 mg” referred to equivalents, not every product’s labeled milligrams. The high-dose category contained 50 cases and 35 controls. Dose was missing for 53 of 415 exposed patients. Methylphenidate’s estimate was 0.91 (0.54–1.55), without a statistically clear association. Hospitalized control selection, incomplete information and residual confounding constrain causal inference. Methods, Table 2 and limitations.
Why cannot 50 cases out of 85 establish incidence?
The investigator selected people according to whether they belonged to the case or control group. The resulting fraction of cases among high-dose subjects therefore reflects that sampling process. It does not describe what happened to a population of 85 initially unaffected high-dose patients followed forward in time.
This is also why “5.28-fold odds” must not become “a 5.28-fold probability” or “five in every hundred patients.” An absolute probability needs a suitable starting risk and a study design that supports that calculation. Applying an arbitrary baseline would introduce information the paper did not establish for the patient under review.
Why does the dose scale matter?
Using the study’s arithmetic, 40 labeled milligrams of mixed amphetamine salts corresponds to 30 mg equivalents, whereas 30 labeled milligrams corresponds to 22.5 mg equivalents. These are illustrations of the published classification method, not instructions for switching medicines or assertions that formulations have identical effects. A report that compares a prescription directly with the study’s boundary should show the product, daily dose and conversion.
The boundary also separates categories used for analysis. A category boundary does not demonstrate a biological cliff: crossing it does not establish that a new effect begins at that exact milligram, and being below it does not establish an absence of risk. The confidence interval describes uncertainty around a group estimate, not the range of possible outcomes for one patient.
What must remain separate in an individual case?
Psychosis and mania were combined in the main outcome, so the headline number should retain both terms. Neither outcome is interchangeable with a toxicology finding, a general impression of agitation or a conclusion about driving performance.
A useful individual assessment separates the prescription record from evidence of ingestion, then aligns reported symptoms with treatment changes, other exposures and the clinical course. The association provides a reason to examine that chronology carefully. It does not by itself establish that the medication caused the episode, that the person misused it or that a detected concentration proves impairment.
This article interprets historical observational evidence published in 2024. The numerical boundary is a research classification, not a current prescribing guideline or a stand-alone rule for medical or legal decisions.
Source
Moran LV, Skinner JP, Shinn AK, et al. Risk of Incident Psychosis and Mania With Prescription Amphetamines. American Journal of Psychiatry. 2024;181:901–909. doi:10.1176/appi.ajp.20230329. Full-text manuscript.