BackgroundRetrograde extrapolation of drug concentrations in blood can be relevant in cases of drug-impaired driving and is regularly used in forensic toxicology in Norway. Δ9-tetrahydrocannabinol (THC) has complex, multi-compartmental pharmacokinetics, which makes retrograde extrapolation of blood THC concentrations problematic. In the present study, we evaluated an approach to retrograde extrapolation in which momentary rates of decrease of THC were estimated from two consecutive blood samples in apprehended drivers. Material and methodsData were collected from apprehended drivers in Norway 2000–2020. We included 548 cases in which THC was detected in two consecutive blood samples collected ≥ 20 min apart. THC concentrations were measured by GC-MS and UHPLC-MS/MS. In each case, THC concentrations and the time between the two sampling points (Δt) were used to estimate the rate constant k. The relationship between THC concentration and k was modelled by linear regression. ResultsThe median Δt was 31 min (interquartile range, IQR = 9). The median blood THC concentration was 2.4 μg/L (IQR = 3.4) at the first sampling point and 2.3 μg/L (IQR =3.1) at the second. The concentration decreased in 62% and increased in 38% of all cases. However, considering measurement uncertainty, the changes were not statistically significant in 87% of cases. The mean of k was 0.12 h-1, corresponding to an apparent t1/2 of 6.0 h. The t1/2 predicted from linear regression of k against THC concentration ranged from 0.93 to 13 h for the highest and lowest concentrations observed (36 and 0.63 μg/L, respectively). The time from driving to blood collection had a median of 1.7 h (IQR = 1.5), and did not correlate with k. ConclusionsThe apparent t1/2 of THC calculated from the mean of k was 6.0 h, which is shorter than the terminal elimination t1/2 suggested in previous population studies. This indicates that blood samples were often taken during the late distribution phase of THC. Because Δt was short relative to the rates of decrease expected in the late distribution and elimination phases, the underlying true concentration changes related to in vivo pharmacokinetics were small and masked by the relatively larger “false” changes introduced by random analytical and pre-analytical error. Therefore, individual values of k calculated from only two blood samples taken a short time apart are unreliable, and a two-sample approach to retrograde extrapolation of THC cannot be recommended.