Why 98.7% Fingerprint THC Agreement Needs Its Underlying Counts
A fingerprint THC screen achieved 98.7% agreement with laboratory fingerprint testing in 75 postmortem cases. The underlying counts were 73 true negatives, one true positive and one false positive. Those numbers convey much more about the evidence than the headline percentage alone.
What “accuracy” referred to
Hudson and colleagues evaluated a fluorescence-based lateral-flow device using sweat remaining in fingerprints collected from deceased individuals. A second fingerprint collected in parallel was analyzed by liquid chromatography with tandem mass spectrometry.
For THC, the screen used a cutoff of 190 picograms per fingerprint. Its 98.7% accuracy was agreement with the laboratory classification of the comparison fingerprint. It did not measure agreement with observed intoxication, driving ability or the person’s condition at a known earlier time.
One confirmed positive cannot support a broad sensitivity claim
The study’s table reported 100% THC sensitivity because the single laboratory-positive case was detected. It also reported a 50% positive predictive value: only one of the two positive screens was confirmed in the comparison fingerprint.
These measures are mathematically consistent. High overall accuracy was largely driven by correctly classifying negative samples. With only one confirmed THC-positive specimen, the sensitivity estimate is especially uncertain. It should not be advertised as proof that the device detects every cannabis user.
Matrix and contamination remain relevant
The authors identified accidental contact as a potential issue when detecting parent THC on fingers. A handwashing strategy possible in living volunteers was unavailable for these postmortem cases. Confirming THC chemically does not automatically resolve how it reached the fingerprint.
Blood and urine comparisons also involve different matrices and available sample counts. Their agreement figures should not be pooled with the 75 fingerprint comparisons as though they measure one identical outcome. Several authors were affiliated with the device manufacturer, a relevant context for evaluating product-development research.
The study supports further examination of a screening method, with confirmation of positives. As with other cannabis evidence, chemical presence, route of exposure and impairment are separate questions.
Primary research
Drug screening using the sweat of a fingerprint: lateral flow detection of Δ9-tetrahydrocannabinol, cocaine, opiates and amphetamine. DOI: 10.1093/jat/bky068.
