There are a number of published programs for improving quality. Some of these are embodied in checklist certifications such as the International Organization for Standardization (ISO), NLCP, or the American Board of Forensic Toxicology. In 1986, Motorola Corporation produced an approach to achieving excellence that went beyond checklists, called Lean Six Sigma Approach. It combined efficiency plans from the automotive industry (lean) with Motorola’s own quality control in the semiconductor industry (six sigma) to form a continuous quality improvement process. “Lean” focuses on speed and efficiency of processes. Six Sigma” emphasizes precision and accuracy in processes and outcomes. Many of the concepts are applicable to drug-testing laboratories. The basic approach is for organizations to examine their current processes and define goals, develop methods of measuring processes, analyze the processes to determine the causes of poor quality solicit and implement ideas for improvement, and establish controls to reduce variability. For new processes, the approach begins in a similar fashion with defining or redefining goals. Measures are then established for capabilities and risks. Analysis follows but includes alternative methods to see if they will improve products. Once an approach is selected it is designed and proof of concept is verified before implementation.
A fundamental of the Six Sigma process is quantifying variability and establishing acceptable limits. The general concept is illustrated in Figure 4.1, where variability of a measure from an expected value is a normal distribution. The sigma values shown are standard deviations from the mean. The performance measure being monitored must be within the limits of -6s to +6s to be acceptable. This means that desirable impression is 1/6 of the tolerable error. This assumes that bias in the assay is zero. If the method has some bias, then precision must be better to keep the total error within limits. For example, consider a GC-MS assay of a QC urine specimen with 15 ng THCCOOH/mL and 15 ng/mL d3-THCCOOH as internal standard analyzed in 4 separate batches. The measured concentration of THCCOOH must not exceed +20% of 15 ng/mL. The desired coefficient of variation for the 4 test results would be ±20%/6, or 3.3%. If the matrix contributes 1 ng/mL of THCCOOH that is not in the extracted standard, then the QC would have a 1/15 bias, or 6.7%. To stay within the +20% limit for this QC, imprecision must now be less than or equal to (20 – 6.7)%/6, or 2.2%. A better solution, of course, than improving precision would be to prepare a new QC with drug-free urine and remove the bias.
This example is simple to illustrate a small part of the Six Sigma process. Laboratories have practiced this kind of QA for many years. The challenge of Six Sigma is to identify as many processes as possible that can affect results, even human processes, and find ways to quantify, monitor, and reduce variability. Some good descriptions of various processes can be found on the Westgard website.