Quality Assurance in Clinical Chemistry: A Touch of Statistics and A Lot of Common Sense

Abstract

Working in laboratories of clinical chemistry, we risk feeling that our personal contribution to quality is small and that statistical models and manufacturers play the major roles. It is seldom sufficiently acknowledged that personal knowledge, skills and common sense are crucial for quality assurance in the interest of patients. The employees, environment and procedures inherent to the laboratory including its interactions with the clients are crucial for the overall result of the total testing chain. As the measurement systems, reagents and procedures are gradually improved, work on the preanalytical, postanalytical and clinical phases is likely to pay the most substantial dividends in accomplishing further quality improvements. This means changing attitudes and behaviour, especially of the users of the laboratory. It requires understanding people and how to engage them in joint improvement processes. We need to use our knowledge and common sense expanded with new skills e.g. from the humanities, management, business and change sciences in order to bring this about together with the users of the laboratory.

Keywords: metrology; postanalytical error; preanalytical error; quality assurance; quality control.

Figure 1

The total testing chain in clinical chemistry involves several professionals and organizations in healthcare from the clinical decision to order a test through the pre-analytical, analytical and post-analytical phases to the value of the test result in the on-going clinical decisions and healthcare process.

Figure 2

When a mean of a result is reported, the error of the mean is influenced both by bias and random error. The standard error of the mean is inversely related to the square root of the number of replicates and thus decreases quadratically with the number of replicates. As the number of replicates is increased, the contribution of the random error to the measurement error of the mean approaches zero, thereby improving the estimate of the bias.

Figure 3

External quality control (ECQ) organizations send out stabilized quality control samples which are analyzed as singletons and evaluated centrally (depicted as dotted arrows). The use of split fresh patient samples (depicted as the solid black ring) including the use of replicate measurements facilitates finding bias and thereby minimization of measurement uncertainty in an organization or a geographical area. A laboratory represented by the yellow circle may preferably serve as a mentor for a certain measurand for the other laboratories in the conglomerate of laboratories serving a certain population (28, 34).

Figure 4

The diagnostic uncertainty of a measurement result in a patient sample is a property of the measurement result itself, influenced by several uncertainty components, including biological variation, preanalytical variation, analytical variation (including uncertainty of the calibration) and postanalytical variation. The total error of an external quality control sample, in contrast, is influenced by substantially fewer and smaller uncertainty components and therefore represents a property of the measurement system itself. The total error is commonly used for regulatory or accreditation purposes. EQC = extrenal quality control, TE = Total error.