A reported intention to use methamphetamine and a clinically classified opioid overdose can coexist without establishing exactly what product was consumed. A 2024 study of San Francisco overdose-response records identified 32 events in which methamphetamine was the reported intended substance and no opioid use was reported as intentional.
The finding raises an important exposure question. It does not, by itself, prove that 32 methamphetamine samples were contaminated with fentanyl, estimate the contamination rate of the local supply, or determine an individual person’s intent.
What kind of study was this?
Bazazi and colleagues retrospectively reviewed consecutive suspected-overdose records from the San Francisco Street Overdose Response Team, covering June through September 2022. The team included community paramedics and peer counselors responding to suspected overdoses, often with other emergency services.
This was a review of medically attended, nonfatal events. It was not a controlled drug-administration experiment, a survey of all methamphetamine users, a fatal-overdose investigation or a systematic chemical analysis of purchased drugs.
The publication date was April 2024, but the observations concerned four months in 2022. They do not establish current drug-market conditions in San Francisco or elsewhere.
How did the denominator change?
| Stage | Reported count | Interpretation |
|---|---|---|
| Suspected-overdose response records reviewed | 448 | Starting record set |
| Records meeting the study’s opioid-overdose definition | 281 | Clinical classification from available documentation |
| Classified events with intended-substance information | 171 | Complete cases for the main intentionality analysis |
| No reported intentional opioid use | 73 | 42.7% of the 171 complete cases |
| Methamphetamine reported as intended | 32 | 43.8% of the 73; 18.7% of the 171 |
Those percentages answer different questions. Neither 43.8% nor 18.7% is the probability that a methamphetamine product contained fentanyl or that a person using methamphetamine would overdose.
The unit of analysis was a response record or event. The reported totals should not automatically be described as a cohort of that many unique people with independently verified exposures.
How was an opioid overdose classified?
The study’s adapted case definition required a narrative describing suspected overdose or unresponsiveness, naloxone administration by a bystander or emergency personnel, and subsequent improvement in responsiveness or respiratory rate.
The researchers adapted an existing emergency-record classification approach because their clinical-impression information was unstructured. They acknowledged imperfect sensitivity and specificity. Records not meeting the definition could represent another type of event or insufficient documentation, rather than a confirmed absence of overdose.
The case definition did not require a fentanyl-specific laboratory result. It therefore identified a documented clinical pattern, not the exact opioid, dose, product composition or route of unintended exposure.
How was intention determined?
Intended substance use came from patient self-report recorded in paramedic free-text notes. Two researchers independently reviewed batches until achieving more than 95% concordance, after which one reviewer conducted abstraction and the principal investigator checked a random 5% subset.
Concordance described agreement in interpreting the records. It did not independently verify that every recorded account was complete or accurate.
Among the 73 events without reported intentional opioid use, 53 involved an intended stimulant. Methamphetamine accounted for 32, crack cocaine for 11, powder cocaine for eight and an unspecified stimulant for two. These were reported intended substances, not confirmed analytical identifications.
Did the study identify the exposure mechanism?
No. The authors discussed several possibilities: substitution or mistaken identification, contamination of a non-opioid product, storage together, and reuse of equipment previously used for fentanyl. The record review did not distinguish those mechanisms.
Intentional opioid use might also have been underreported. Conversely, uncertainty in self-report does not justify assuming that every report of unintended use was false. The evidence supports preserving the uncertainty rather than selecting one explanation without additional information.
A forensic investigation would need case-specific evidence, such as retained-product testing, biological toxicology, packaging, timing and a fuller exposure history, to evaluate competing explanations. This study did not supply those measurements for each event.
How much information was missing?
Of the 281 events classified as opioid overdoses, 110 lacked intended-substance information. The complete-case percentage of 42.7% therefore cannot simply be applied to all 281, much less all overdoses in the city.
The paper’s discussion contains an arithmetic inconsistency: it gives a hypothetical missing-data estimate as 28% with the fraction 78/281, whereas the main results and table report 73 events without intentional opioid use. Using the reported main-results numerator, 73/281 is approximately 26.0%. That calculation is a hypothetical proportion if every missing-intent event involved intentional opioid use, not a corrected observed prevalence.
The alternative clinical definition added three classified overdoses, but none had intended-substance information, so it did not change the complete-case analysis. Missingness itself remained unresolved.
What should be concluded?
The study documents a meaningful mismatch between reported intended substances and clinical event classification in a selected historical record set. It supports further investigation of unintended opioid exposure among people intending to use stimulants.
It does not establish the frequency of fentanyl contamination in methamphetamine, prove the mechanism of an individual event, or turn a clinical classification into a chemically confirmed dose. Separating intention, clinical response and analytical identification makes the finding more useful and more defensible.
Scientific source
Bazazi AR, Low P, Gomez BO, et al. Overdose from Unintentional Fentanyl Use when Intending to Use a Non-opioid Substance: An Analysis of Medically Attended Opioid Overdose Events. Journal of Urban Health. 2024;101(2):245–251. doi:10.1007/s11524-024-00852-0. Full text. Historical observational review of 2022 emergency-response records; the main-results and discussion numerator discrepancy is identified above.
Leave a Reply