Patients Aren’t p Values — Proximal Humerus Fractures Study Oversells Findings

Imperceptible benefits — even if observable in statistics — won’t matter to patients

What’s the Claim?

A recent retrospective, comparative study of patients over the age of 55 who had isolated proximal humerus fractures treated either without surgery or with a locked plate, and matched them for age, comorbidities, and fracture type, suggested surgery resulted in “decreased pain and improved passive ROM” in the first six months after injury. The authors said that patients treated with ORIF also “trended toward a higher secondary reoperation rate (15.5% vs. 5.0%)” compared to those treated nonsurgically, but statistics didn’t bear out this difference.

How’s It Stack Up?

Most readers will be familiar with the ProFHER (Proximal Fracture of the Humerus) family of studies, including the seminal RCT that found no advantage to surgery over nonsurgical care at either two or five years, the main findings of which suggest that many patients who have surgery for this diagnosis may do so to their disadvantage. Authors of the current study wondered whether looking at the longer term, as ProFHER did, might result in surgeons denying patients some early advantages that surgery confer involving quicker relief of pain or restoration of function, and they suggested that better matching of patients and injuries than was possible in ProFHER RCTs would be worth doing.

What’s Our Take?

That stated rationale certainly justified the study, and the authors did about as well as possible in a retrospective study of this sort. It would have been nice to know whether the substantial proportion of patients missing (15% or so) were unevenly distributed in the study’s comparator groups. But our main concerns about this study centered on interpretation, not design. Almost all the differences the authors claimed, based on p values, would’ve been imperceptible or unimportant to patients, and it does not seem likely that a patient would choose major surgery to achieve those small benefits for just a few weeks. For example, the authors claimed surgery provided about 20° of additional passive forward flexion between 2 and 12 weeks, with a 16° advantage of active flexion appearing only at six weeks, but not before or after that point. Is that worth the knife? And if we define two points out of 10 as a clinically important difference on the VAS pain scale, as is commonly done (we actually think it ought to be much larger than that to justify major surgery), the only clinically important difference between the groups was seen two weeks after injury, where patients treated without surgery had a VAS pain score of 6.2 out of 10 and those who had surgery reported 2.9 out of 10. Differences on the PROMIS outcomes tool were even smaller, and no differences in any endpoint favored surgery by six months (right in line with ProFHER’s findings). Notable, really, given that most of the patients in this series had surgery after ProFHER was published, so there was a good chance that those who underwent surgery were the ones surgeons thought might really benefit from it. Regardless, here’s the nugget — patients don’t perceive p values, they perceive effect sizes. Any benefits found here were nearly imperceptible in size. There may be reasons to perform surgery in some older patients with proximal humerus fractures, but authors’ protestations to the contrary notwithstanding, they were not convincingly identified in this study.


Samborski SA, Haws BE, Karnyski S, et al. Early Outcomes of Proximal Humerus Fractures in Adults Treated With Locked Plate Fixation Compared With Nonoperative Treatment: An Age-, Comorbidity-, and Fracture Morphology-Matched Analysis. J Orthop Trauma. 2023;37:142-148.