Two million urine specimens can provide substantial information about a laboratory’s tested population without becoming a representative survey of everyone in the United States. A historical clinical-testing study illustrates both the value and the limits of large toxicology datasets.
What was the denominator?
LaRue and colleagues examined clinical specimens collected from January 2014 through December 2019. They selected a convenience sample of two million adult patient specimens with definitive methamphetamine test orders and used one specimen per person, based on the earliest collection date. Samples came from healthcare practices in all 50 states and the District of Columbia.
Testing used liquid chromatography–tandem mass spectrometry at one commercial laboratory. The investigators excluded results consistent with prescribed medications as reported by the ordering healthcare professional. That process depended on prescription information supplied with the order; it was not an independent verification of every person’s entire medication history.
Which positivity numbers are comparable?
Across the six-year dataset, 106,405 of two million specimens were methamphetamine-positive: 5.32% raw positivity. Separately, the regression model estimated marginal positivity probabilities of 1.0% in 2014 and 4.4% in 2019 after accounting for demographic, regional and practice-specialty variables. Their reported 95% confidence intervals were 1.0%–1.1% and 4.2%–4.6%, respectively.
The authors described that modeled change as a 340% relative increase. It is also a 3.4-percentage-point absolute increase. Neither is the same quantity as the pooled 5.32% raw figure, which covers all six years and reflects the actual mixture of sampled patients. Calling these numbers contradictory, or combining them into a single trend calculation, would mix estimands.
Why is this not national prevalence?
People were tested because healthcare professionals considered testing medically necessary. The dataset therefore selected for entry into care, laboratory use, a methamphetamine order and the participating practices. Random selection within that available dataset does not make the underlying clinical population nationally representative.
Adjusting for age, sex, census division and practice specialty helps address recorded differences. It cannot account fully for why a clinician ordered a test or for changes in unmeasured patient characteristics. The study is strong evidence about trends in this testing system, with a narrower basis for population-wide claims.
Why do copositivity percentages need another denominator?
Additional substances were not ordered for all two million specimens. Among methamphetamine-positive specimens, 78,510 had fentanyl testing, with 9.03% positivity. The relevant comparison is the subset tested for fentanyl, not all 106,405 methamphetamine-positive specimens. Heroin assessment used 6-monoacetylmorphine; alcohol assessment used ethyl glucuronide and ethyl sulfate. Those analytes are not interchangeable with blood parent-drug concentrations.
The study also modeled adjusted odds of other-drug detection in methamphetamine-positive versus methamphetamine-negative groups. An odds ratio is not a percentage of samples, an overdose probability or proof that one substance caused use of another.
Does a copositive urine establish a mixed dose?
No. Detection of two analytes in one urine specimen may reflect exposures at different times because analytes have different formation and elimination patterns. These data cannot distinguish intentional simultaneous administration, sequential use or unrecognized exposure in an individual case.
The findings also do not establish impairment at collection or at an earlier event. The authors’ commercial-laboratory affiliations and funding are relevant context: most authors were laboratory employees and one was a consultant. That does not negate the measurements, but transparency about data source and sampling is essential when presenting the results as evidence of broader trends.
Source
LaRue L et al. The Cyclical and Evolving Nature of Methamphetamine Use: An Evaluation of Near Real-time Urine Drug Test Results. Journal of Addiction Medicine. 2021;15:396–405; copyright/online record 2020. doi:10.1097/ADM.0000000000000778. Historical cross-sectional clinical laboratory evidence.
Leave a Reply