Increasing THC Positivity Did Not Mean the Study’s 2016 Breakpoint Accelerated the Trend

forensic toxicology working group members October 31, 2026 Forensic Toxicology

THC positivity increased over the observation period, but the paper’s chosen pre/post-2016 interaction did not show a significant acceleration. An overall rise and a change in the rate of that rise are different statistical questions.

The study evaluated more than a before-and-after count

The researchers examined monthly patterns in positive toxicology among screened trauma patients. They also fitted a model incorporating an interaction between time and their chosen 2016 dividing point. That term asked whether the trajectory changed across the designated periods.

The direct change comparison was null

The paper reported a persistent increase over time, while the interaction did not show a significant differential change, with p=0.923. A higher proportion late in the series does not independently establish that the trend accelerated at the dividing point. The direct interaction result must accompany the broader time trend.

The model does not establish a policy’s complete effect

This is a statement about the statistical comparison used in the paper, not a legal account of enactment dates or a causal estimate of every policy consequence. Screening selection, co-occurring changes and the chosen time model constrain inference. The null interaction also does not prove that no population effect of any size was possible. A defensible summary distinguishes the observed rise from an unsupported claim of a demonstrated slope change. It should also preserve that the outcome was toxicology positivity among selected trauma patients, rather than measured impairment or population crash incidence.

Related analysis: injury severity and population crash risk.