Laboratory identification of drug use based on observable signs and symptoms.

The heuristics of the Drug Evaluation and Classification Program (DECP) were evaluated for detecting drug impairment and identifying the type of the impairing drug. A quantitative model, based on stepwise logistic regression of systematic observations made by Drug Recognition Experts (DREs), provided higher levels of drug identification, and lower rates of errors than the DREs. The regression models' Phi correlations with the actual drug dosing were three times as high as the DREs' correlations with respect to cannabis, twice as high with respect to alprazolam, equal (and poor) with respect to codeine, and over 50 times as high with respect to amphetamine (where the DREs performance was worse than chance). In conclusion, the DECP is a useful tool for drug detection and identification, though to be effective its training procedures should focus on better integration of information from signs and symptoms indicative of drug impairment.