Methodological considerations on repeated measures and sub-LOQ values in passive cannabis hair contamination

Giorgetti et al. report a carefully designed in vivo cross-over experiment showing that brief passive exposures to cannabis or "light cannabis" smoke inside a vehicle can deposit detectable Δ9-tetrahydrocannabinol (Δ9-THC) or cannabidiol (CBD) into hair, even after a validated dichloromethane washing step. Their findings are forensically relevant because washed-hair concentrations sometimes approached or exceeded Society of Hair Testing cut-offs used to infer active use. The work advances beyond in vitro models by using realistic exposure durations. However, two statistical practices warrant further examination: the repeated-measures structure from four volunteers contributing multiple samples was not fully accounted for in primary tests, and values below the limit of quantification were replaced by a constant equal to the limit of detection. These choices may underestimate variability and produce narrow confidence intervals. This discussion reanalyses the authors' tabulated data using linear mixed-effects models and maximum-likelihood Tobit regression. The qualitative conclusions remain consistent, but uncertainty estimates widen materially, suggesting the value of more conservative reporting in small-cohort forensic toxicology studies.