There is a phenomenon in the medical literature called publication bias. It results from two phenomena – authors are more likely to submit the results of trials with positive results, and editors tend to publish articles with positive results. This results in all sorts of flaws with regard to the composition of the scientific literature, and exerts a particularly troubling hidden effect in meta-analyses and systematic reviews.
I comment upon this in the context of yet another cardiovascular assay article that has – essentially – negative results that are spun to be positive. Copeptin, as I've discussed before, is another acute phase indicator of myocardial demise – but sacrificing specificity for sensitivity. These authors combine copeptin with hs-TnT for evaluation of chest pain in the Emergency Department, and report several favorable findings in their abstract and the text of their discussion.
In reality only one of the findings they focus on is truly positive – an increase in sensitivity from 76% to 96%. The NPV increases from 95% (90.4-98.3) to 98.9% (94.2-100) and is not truly a positive result. More importantly, the authors report copeptin "adds incremental value" – when the area under the receiver operating curve is statistically identical at 0.886 (0.85-0.922) vs. 0.928 (0.89-0.967).
Perhaps copeptin will someday be proven to add true clinical value in an algorithm for the rapid assessment of chest pain in the Emergency Department. This paper, however, seems to have exaggerated the positivity of its results. Considering the spate of other recent "positive" copeptin articles – I foresee systematic reviews and meta-analyses of the test characteristics further perpetuating any unremarkable reported advantage in test characteristics.
"Early rule out of acute myocardial infarction in ED patients: value of combined high-sensitivity cardiac troponin T and ultrasensitive copeptin assays at admission"