Up to 25% of hospitalized patients in a psychiatric department exhibit troubles linked to cannabis use. Weaning patients with psychiatric disorders off drugs of abuse requires specific care to improve their clinical outcome. The present study aims to develop a predictive model of urinary excretion of creatinine-normalized cannabinoids (UCNC ) and to determine UCNC thresholds corresponding to the widely used cut-offs of 20 ng/mL and 50 ng/mL for cannabinoids. One hundred thirty-two patients with 452 urine samples were included between 2013 and 2017. Urinary cannabinoids and UCNC elimination curves were computed for each patient. Using a mono-exponential mixed effect model with 88 samples from 26 subjects exhibiting at least 3 decreasing UCNC in a row, the average calculated elimination rate constant was 0.0108 ± 0.0026 h-1 , corresponding to a mean elimination half-life of 64 ± 12 hours. The use of UCNC is of particular interest because of a high inter- and intra-individual variability of urinary creatinine concentration (from 0.06 to 3.81 mg/mL). Moreover, UCNC allows for the detection of diluted or concentrated urine specimens that may lead to false positive (FP) or false negative (FN) results. Receiver operator characteristic (ROC) curves were used to assess UCNC thresholds of 32.4 and 124.7 ng/mg that provide a strong discrimination between positive and negative samples for cannabinoids cut-offs of 20 and 50 ng/mL respectively. The developed model and the defined UCNC thresholds allowed for the accurate prediction of the time needed to reach a negative UCNC result that could be used by clinicians to optimize clinical care.
Keywords: cannabinoids; creatinine; modeling; monitoring; psychiatry.
© 2019 John Wiley & Sons, Ltd.