Statistics in publishing: the (mis)use of the p-value (part 1)

Abstract

In hand surgery research, most studies, whether observational studies or randomized controlled trials (RCT), are aimed at finding out whether there is an effect (association or difference) of a certain determinant on a specific outcome. This is usually determined using null-hypothesis significance testing (NHST), in which a p-value <0.05 is considered as evidence that the findings are significant. Although this method is widely used, it has been criticized since its inception. The critique has been mainly focused on the misuse of NHST, but also more conceptually on the method itself. In part 1 of this two-part article, we discuss some examples of how the p-value can be misused, using a simulated dataset partly based on real data from an RCT (Broekstra et al., 2022). In part 2, we will discuss the conceptual criticism and offer some guidance on alternatives. In this example study, women with a distal radial fracture were randomized either to an intervention (cast þ rehabilitation programme) or control (cast only) group in a 1:1 ratio. The intervention was aimed at restoring hand function, which was measured using the Patient-Rated Wrist Evaluation (PRWE), a validated patient-reported outcome measure for determining hand function in patients with wrist problems, with a score ranging between 0 (no problems) and 100 (severe problems).