Stability of relative and absolute metrics: empirical evidence from pulmonology

摘要

It has been widely argued that absolute treatment effect measurements (such as risk difference) reveal the “clinical benefit” of an intervention. Yet, many previous experience with binary endpoints have shown that they are unlikely to be transportable between populations. As absolute metrics are usually derived from baseline risk and relative metric (such as odds ratio), it seems logical to rather measure relative metrics, assuming they are stable. In the present study, a continuous endpoint was used to assess the stability of both relative and absolute metrics using a empirical data from pulmonology. Results are preliminary due to the low baseline variability, yet, the difference was significantly correlated with the baseline, unlike the ratio, which is in line with previous experience with binary endpoints. Further research is needed to explore the stability with continuous endpoints.