In January 2025, a research group at the University of Dundee published a careful evaluation of a market-leading nitazene test strip in Harm Reduction Journal. The study, by Marland, Nisbet, and Nic Daéid (doi:10.1186/s12954-025-01287-9), tested the strip against 36 nitazene analogs and 93 other compounds. It found that caffeine, present in heroin samples at concentrations of 21 to 25 percent by weight in three Scottish prison seizures, triggers false positives at concentrations above approximately 300 micrograms per milliliter.
That is a problem. Not because caffeine is particularly hard to identify, but because it is one of the most common adulterants in the unregulated heroin supply globally, and the false positive cannot be diluted out without simultaneously diluting the nitazene below the strip's detection threshold.
The Dundee paper got significant attention in the harm-reduction press. The deeper question it raised is how strip specificity is being evaluated at all.
Specificity is not a footnote
For most rapid lateral-flow immunoassays, the published spec sheet leads with limit of detection (the lowest analyte concentration that produces a reliable positive) and includes a brief statement about cross-reactivity, often expressed as "no cross-reactivity at X mg/mL" for a list of structurally adjacent compounds. The list is rarely complete and the test concentrations are rarely chosen to reflect what is actually in the unregulated supply.
The Notre Dame work on fentanyl test strips, also in Harm Reduction Journal, makes the same point about a different analyte. Lockwood, Vervoordt, and Lieberman in 2021 (doi:10.1186/s12954-021-00478-4) showed that methamphetamine, MDMA, and diphenhydramine trigger false positives on a widely-distributed fentanyl strip at concentrations above 1 mg/mL. The 2024 lot testing protocol from Marya Lieberman's group (Fernando et al., doi:10.1186/s12954-024-01058-y) formalized a seven-compound interference panel as part of routine lot QA.
What a useful interference panel looks like
The shape of a useful interference panel for any rapid drug-checking strip should follow three principles:
- Test the things actually in the supply. Caffeine, lactose, mannitol, levamisole, lidocaine, procaine, diphenhydramine, methadone, methamphetamine, MDMA, and other commonly co-present compounds belong on the panel. Their absence from a manufacturer's claim is itself information.
- Test at concentrations that match field reality. A "no cross-reactivity at 100 ng/mL" claim is meaningless if the cutting agent is present at 25 percent by weight. Test up to and beyond the upper bound of what gets seen in seized samples.
- Publish the failure modes, not just the successes. A buyer is better served by knowing "false positive on diphenhydramine above 0.2 mg/mL" than by being told "high specificity." Specificity is a curve, not a label.
How DSG applies this
DSG's drug-checking strips are evaluated against an interference panel modeled on the Fernando et al. seven-compound set, extended with adulterants specific to the analyte class (for example, detomidine for medetomidine strips, levamisole and caffeine for nitazene strips, designer-benzodiazepine analogs for benzodiazepine strips). The false-positive thresholds we observe are documented and provided to institutional buyers as part of the technical brief that accompanies a quote.
No lateral-flow immunoassay is free of interference. Buyers should know the threshold above which a co-present compound will produce a confounding result, so frontline staff can be trained to interpret accordingly and the strip can be paired with confirmatory mass spectrometry where stakes warrant it.
An interference profile published with the strip is worth more than a sensitivity number on its own. The first tells you when the strip lies. The second tells you when it tells the truth.
The cumulative shift
Over the past three years, the academic literature on drug-checking strip evaluation has shifted from a sensitivity-only frame to a specificity-and-sensitivity frame. The Notre Dame group, the University of Dundee group, the Ghent group (de Vrieze et al., 2024, doi:10.1186/s12954-024-01078-8) on nitazene analog detection, and NIST (Sisco et al., 2024) on multi-analyte cross-reactivity have all converged on the same point. A strip that detects an analyte but throws confounding positives in the field has not solved the problem.
Procurement should follow the literature. The next time a supplier walks into your office with a sensitivity claim and no specificity data, ask why.
Further reading
- Marland S, Nisbet LA, Nic Daéid N. (2025). Limitations of nitazene rapid test strips for harm reduction drug checking. Harm Reduction Journal 22:5. doi:10.1186/s12954-025-01287-9
- Lockwood TLE, Vervoordt A, Lieberman M. (2021). High concentrations of illicit stimulants and cutting agents cause false positives on fentanyl test strips. Harm Reduction Journal 18:30. doi:10.1186/s12954-021-00478-4
- de Vrieze J, Stove C, Vandeputte M. (2024). Detection of nitazene analogues with a commercially available lateral-flow assay. Harm Reduction Journal 21:118. doi:10.1186/s12954-024-01078-8