When Scientific Limitations Are Framed as Policy Failures: A Critical Look at the misuse of flawed University of Sydney Research in Roadside Drug Testing Debates

Subject Article

T. Arkell et al., 'Detection of Δ9 THC in oral fluid following vaporized cannabis with varied cannabidiol (CBD) content: An evaluation of two point-of-collection testing devices', Drug Testing and Analysis, vol. 11, no. 10, 2019, pp. 1486-1497

There is a difference between scientific inquiry and policy advocacy dressed as science. Blurring that line does a disservice to both.

A widely cited (primarily by the Greens and Cannabis Parties) 2019 study from the University of Sydney examining oral fluid THC detection is increasingly being used to argue that roadside drug testing is unreliable. That claim does not survive even a basic forensic reading of the paper.

The Fundamental Flaw: The Wrong Device

Let’s start with the issue that should immediately end any attempt to apply this study to NSW roadside testing:

The study did not evaluate the device used by NSW Police.

Instead, it assessed commercially available versions of the Securetec DrugWipe 5S and the Dräger DrugTest 5000.

This is not a minor limitation. It is a complete break in relevance.

Immunoassay devices are not interchangeable. Small differences in antibody design, buffer systems, or cut-off calibration can materially alter performance. Treating one device as a proxy for another is scientifically indefensible.

Any argument that this study demonstrates flaws in NSW roadside testing is therefore built on a false premise.

A Controlled Lab Is Not the Roadside

The study involved:

• 14 participants

• Infrequent cannabis users

• Standardised vaporised dosing

• Strictly controlled conditions

This is a pharmacokinetic experiment, not an operational evaluation.

Real-world roadside testing involves:

• Chronic and heavy users

• Mixed routes of administration

• Variable saliva composition and flow

• Environmental and procedural variability

The study does not replicate these conditions. It does not attempt to. Yet its findings are routinely stretched into that domain.

The Known Problem: Oral Contamination

The study confirms what forensic toxicologists have long understood:

Peak oral fluid THC occurs within minutes of use and declines rapidly thereafter.

This reflects oral contamination, not systemic drug levels. It explains both the high early concentrations and the rapid drop-off.

Using this as a basis to critique “accuracy” conflates detection of recent exposure with measurement of impairment. No oral fluid device (regardless of manufacturer) resolves that biological limitation.

Threshold Effects Are Not Device Failure

The study reports reduced accuracy around the 10 ng/mL cut-off. This is presented as a limitation of the devices.

In reality, it is a fundamental property of any threshold-based screening system.

Near any decision boundary, small analytical or biological variations will produce classification instability. Lower the confirmatory cut-off and false negatives increase. Raise it and false positives increase.

This is not evidence of defective technology. It is basic analytical chemistry.

What the Study Actually Shows

When stripped of overinterpretation, the findings are unremarkable:

• Oral fluid THC is a marker of recent use

• Concentrations decline rapidly post-administration

• Screening devices require confirmatory testing

These are established principles, not revelations. These have in fact been part of Australian Standard 4760 (Oral Fluid testing) since its inception in 2006.

Where the Narrative Breaks Down

Problems arise when a controlled laboratory study, conducted on a small and selective cohort using devices not deployed in NSW, is used to make broad claims about the reliability of a specific roadside testing program.

That leap is not supported by the data.

It is an extrapolation that ignores the study’s own limitations.

The Bottom Line

This study has value within its scope. It contributes to understanding oral fluid THC dynamics and the behaviour of specific commercial devices under controlled conditions.

What it does not provide is a direct, transferable critique of NSW roadside drug testing.

Using it as such is not a matter of scientific interpretation. It is a misapplication of evidence.

In forensic science, context is everything. Remove it, and even good data becomes misleading.

Perhaps the most ironic aspect of this study is what it actually shows when read honestly. Oral fluid THC levels peak quickly and fall just as fast, with detection becoming unreliable within a few hours. In other words, the same paper often cited to attack roadside drug testing also suggests that positive results are largely confined to very recent use, not days later. That leaves a simple question: which is it? Either this is a fatally flawed study that cannot be relied upon to criticise roadside testing, or it is evidence that such testing primarily detects recent consumption (the past 3 hours). It cannot be both. Selectively invoking its conclusions when convenient, and ignoring them when they are not, is not scientific debate. It is advocacy.