External validity in a multicenter randomized clinical trial of proximal humeral fractures: the DelPhi trial

BACKGROUND: Randomized controlled trials (RCT) are regarded as the gold standard for effect evaluation in clinical interventions. However, RCTs may not produce relevant results to all patient groups. We aimed to assess the external validity of a multicenter RCT (DelPhi trial). METHODS: The DelPhi RCT investigated whether elderly patients with displaced proximal humeral fractures (PHFs) receiving reversed total shoulder prosthetic replacement (RTSA) gained better functional outcomes compared to open reduction and internal fixation (ORIF) using an proximal humerus locking plate (PHILOS). Eligible patients were between 65 and 85 years old with severely displaced 11-B2 or 11-C2 fractures (AO/OTA-classification, 2007). We compared baseline and follow-up data of patients for two of the seven hospitals that were included in the DelPhi trial (n = 54) with non-included patients (n = 69). Comparisons were made based on reviewing medical records regarding demographic, health and fracture parameters. RESULTS: Forty-four percent of the eligible patients were included in the DelPhi trial. Comparing included and non-included patients indicated higher incidences of serious heart disease (P = 0.044) and a tendency toward higher tobacco intake (P = 0.067) in non-included patients. Furthermore, non-included patients were older (P = 0.040) and had higher ASA classification (P < 0.001) and were in more need for resident aid (in-home assistance) (P = 0.022) than included patients. The cause of PHF was more frequently related to fall indoors in non-included vs. included patients (P = 0.018) and non-included patients were more prone to other concomitant fractures (P = 0.004). Having concomitant fractures was associated with osteoporosis (P = 0.014). We observed no significant differences in rates of complications or deaths between included and non-included patients within 3 months after treatment. In descending order, non-included patients were treated conservatively, with PHILOS, RTSA, anatomic hemi-prothesis or an alternative type of ORIF. RTSA was the preferred treatment choice for C2-type fractures (P < 0.001). CONCLUSIONS: Results from the DelPhi RCT may not directly apply to older PHFs patients with lower health status or concomitant fractures. LEVEL OF EVIDENCE: Level 4.