I’ve been doing an assignment where I have to go through published evaluations to see what the state of the art is for a bunch of analytes. The most important parameter – as always – is simply the distribution of differences between the published assay and its comparison assay. But for most of the publications I’ve seen, it’s impossible to determine this parameter. There is because
- Difference plots (Bland-Altman plots) are not present – instead, there are scatter plots which are mostly empty space
- Regression results lack the estimated sy.x
Besides the inadequate method comparison results, there are things like linearity, precision, and carryover with which one cannot use to estimate the distribution of differences.
It’s a shame that in this day and age, with so many standards, that the studies have such low information content.