In this fourth article in his current series, Stephen MacDonald moves on to provide a guide to dealing with the results of statistical tests initially that look reassuring yet still hide a clinically important problem.
Every quantitative test has limits. Push a specimen too high or too low and the result stops behaving proportionally to the amount of measurand actually present. Signal saturation and reagent depletion blunt the response at the top end. Imprecision and detection capability limit it at the bottom. Before a laboratory reports a number, it needs to know where that number can be trusted.
Linearity studies are how we find out. The name is a bit misleading. The instrument doesn’t need to produce a literal straight line internally, and many calibration curves are deliberately non-linear. What matters is whether the final reported result stays acceptably proportional to concentration across the interval you intend to use. A pretty straight-looking graph, a correlation coefficient near 1, or a non-significant curvature test can all look reassuring and still hide a clinically important problem, which is really the point of this article.
Linearity, validation and verification – The essentials
Log in or register FREE to read the rest
This story is Premium Content and is only available to registered users. Please log in at the top of the page to view the full text.
If you don't already have an account, please register with us completely free of charge.