Assessing the quality and reliability of the DEA drug identification process

Abstract

Seized drugs laboratory managers and analysts strive to produce results that are accurate and reliable by employing scientific techniques that are foundationally valid and appropriately applied. Laboratories can assess the quality and reliability of their processes by using historical performance data or by establishing quality assurance programs that include blind testing or sample re-analysis, among others. Here, an assessment of laboratory error rates within the DEA laboratory system is presented using historical proficiency test laboratory data generated during the years 2005–2016. Results indicate the DEA drug identification process is characterized by high sensitivity (99.90%) and specificity (99.12%), with very low type I (0.87%) and type II (0.092%) error rates. An overall positive likelihood ratio of 114 is calculated, providing an additional quantitative indicator of the laboratory process’ performance. Using Bayes’ theorem and population base rates estimated from historical data, a positive predictive value greater than 99.9% is obtained, further demonstrating the high degree of certainty associated with a positive drug finding.