[In search of the true dosage: statistics applied to the analysis of the cocaine].

A method of separation by gas chromatography with a flame ionization detector was developed for quantifying cocaine in powders seized by the police. The method was validated by studying parameters of calibration, trueness, precision based on trueness error (or systematic bias) and random error. Total error, which is the combination of these errors, is used to confirm the method adequacy with the objectives fixed by the analyst. Accuracy profile is an efficient decision tool to do it. Results obtained with weighted regression model allow concluding that the method fits quantitation of heroin and cocaine in powders on 2 to 100% concentration (w/w) domain with 10% limits of acceptation and a risk of 5%.