2.1 Patient selection

We retrospectively reviewed the TRON database, which includes orthopedic trauma cases from 18 participating hospitals. Between January 2010 and December 2022, a total of 304 patients were identified with a diagnosis of PPF classified as Vancouver type B1, B2, or B3.

The inclusion criteria for this study were as follows: (1) surgical treatment with ORIF and (2) ≥ 60 years of age. Initially, 125 patients met these criteria. We then applied the following exclusion criteria: (1) follow-up period <6 months, (2) pre-injury Parker score ≤2, and (3) initiation of full weight-bearing at ≥ 13 weeks postoperatively. After applying these criteria, 42 patients were excluded. Consequently, 93 patients remained. We then estimated propensity scores using a logistic regression model including age, sex, BMI, pre-injury Parker score, Vancouver classification, and history of fragility fractures. Patients were matched 1:1 using nearest-neighbor matching without replacement with a caliper of 0.2 standard deviations of the logit of the propensity score. Finally, a total of 62 patients were included in the final analysis. After propensity score matching, 62 patients were included (EWB group, n = 31; DWB group, n = 31). The patient selection process is illustrated in a flowchart (Fig. 1).

Close

Fig. 1 Flowchart of patient selection.

Export to PPT

2.2 Data collection

Patient demographic data included age, sex, and body mass index. Pre-existing conditions and risk factors included the American Society of Anesthesiologists Physical Status (ASA-PS) classification,¹⁴ the presence of dementia, smoking status, medication for osteoporosis, and any history of prior fragility fractures (proximal femur, distal radius, thoracolumbar vertebra, or proximal humerus fractures). We also recorded the Parker Mobility Score¹⁵ before the injury. Fracture-specific data included the side of the injury and the Vancouver classification.¹⁶ Details regarding the primary arthroplasty, such as the type (bipolar hemiarthroplasty or total hip arthroplasty) and femoral component fixation (cemented or cementless), were also collected.

2.3 Surgical procedure

The choice of surgical procedure was determined by the operating surgeon. Even in cases classified as Vancouver type B2 or B3 with stem loosening, ORIF was performed in selected patients based on a comprehensive assessment of the fracture pattern and general condition. The fixation method was also determined by the operating surgeon. In most cases, fixation was achieved using a polyaxial locking plate system designed for periprosthetic fractures. The primary implants used across the participating institutions included the LCP Periprosthetic Proximal Femur Plate (DePuy Synthes, Raynham, MA, USA), the NCB Periprosthetic Femur Plate System, Integral GTR plate (Zimmer Biomet, Warsaw, IN, USA), and the Variax™ Periprosthetic Plating System (Stryker, Kalamazoo, MI, USA). Supplemental cerclage cables or wires to aid in fracture reduction and enhance stability were used as necessary.

2.4 Postoperative rehabilitation

Postoperative rehabilitation was initiated on the first day after surgery, with patients being mobilized (e.g., sitting and range-of-motion exercises) as soon as they were medically stable. The progression of mobilization, from using a wheelchair to a walker and then to crutches or canes, was supervised by a multidisciplinary team that included physicians, nurses, and physiotherapists. The timing for initiating full weight-bearing was determined by the individual treating surgeon and was generally performed with weight-bearing as tolerated while monitoring the patient's pain level.

2.5 Clinical evaluation

We evaluated the Merle d’Aubigné Pain Score¹⁷ and the Parker Mobility Score at the last follow-up examination. We collected data on complications, including fracture-related infection (FRI),¹⁸ deep vein thrombosis, implant failure, and reoperation. Implant failure was defined as the breakage of any implant component, including plates, screws, or wires. The Radiographic Union Score for Tibial fractures (RUST)¹⁹ was evaluated six months after surgery. The RUST Score was originally developed to assess bone union in the tibia; however, it has also been shown to be useful in evaluating bone union in other long bones.²⁰ Two observers evaluated the radiographs and the agreement of the diagnosis was assessed. The Kappa values were 0.87 (95% confidence interval: 0.82–0.91). If the diagnoses differed, the results of the first author were used.

2.6 Statistical analysis

All statistical analyses were performed using EZR (ver. 1.66; Saitama Medical Center, Jichi Medical University, Saitama, Japan),²¹ which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria). For the purpose of comparing outcomes, patients were retrospectively classified into two groups based on the timing of the initiation of full weight-bearing. The early weight-bearing (EWB) group (n = 31) included patients who began full weight-bearing ≤5 weeks after surgery. The delayed weight-bearing (DWB) group (n = 31) included patients who began full weight-bearing ≥6 weeks after surgery. We compared baseline characteristics and clinical outcomes between the EWB and DWB groups. Continuous variables were compared using the Mann-Whitney U test. Categorical variables were assessed using the Fisher's exact test. P values of <0.05 were considered to indicate statistical significance.

Subsequently, a subgroup analysis was performed by stratifying patients based on prosthesis stability. Patients with a Vancouver type B1 fracture (stable prosthesis) were analyzed separately from those with Vancouver type B2 or B3 fractures (loose prosthesis). The same statistical comparisons as described above were performed within each subgroup.

3.1 Patient characteristics

Baseline demographics and clinical characteristics were well balanced between the groups (Table 1). The mean age was 77.29 ± 10.01 years in the EWB group and 78.19 ± 8.70 years in the DWB group (p = 0.706), and the proportion of males was 35.5% vs. 29.0% (p = 0.786). The numbers of Vancouver type B1/B2/B3 fractures in the EWB group were 22/7/2, and those in the DWB group were 19/11/1, respectively (P = 0.541). There were no significant differences in patient background between the two groups.

3.2 Functional outcomes

We showed clinical outcome score in Table 2. At final follow-up, the median Parker Mobility Score was 5 in the EWB group and 6 in the DWB group (p = 0.175). The median change from pre-injury Parker score (Delta Parker Mobility Score) was −2 vs. −1, respectively (p = 0.251).

The median Merle d’Aubigné score was significantly higher in the DWB group than in the EWB group (13 vs. 15, p = 0.00377). The DWB group showed higher walking (3 vs. 4, p = 0.0153) and ROM subscores (5 vs. 6, p = 0.0158), while the pain subscore did not differ significantly (5 vs. 6, p = 0.156).

3.3 Radiographic outcomes

The median RUST Scores at six months did not differ significantly between the groups (8 vs. 8, p = 0.659) (Table 2).

3.4 Complications

Complications were rare and did not differ between groups (Table 3). Reoperation was performed in 1 patient in the EWB group and 2 patients in the DWB group (p = 1.000). In the EWB group, the reoperation was performed for a new fracture. In the DWB group, one reoperation was performed for FRI and the other for a new fracture. Both new fractures occurred more than 1 year after ORIF.

3.5 Subgroup analysis

・Vancouver B1 fractures

In Vancouver B1 fractures, baseline characteristics were comparable except for BMI (p = 0.046) (Supplement Table 1). The DWB group showed significantly higher Merle d’Aubigné total scores (p = 0.000744), with significantly higher walking (p = 0.00538) and ROM subscores (p = 0.0215), while the pain subscore did not differ significantly (p = 0.0904) (Supplement Table 2). Parker Mobility Score, delta Parker Mobility Score, the RUST score at 6 months, and complication rates were not significantly different.・Vancouver types B2/B3 fractures

In Vancouver B2/B3 fractures, baseline characteristics were comparable except for dementia (p = 0.045) (Supplement 3). In the matched cohort, no significant intergroup differences were observed in functional outcomes, radiographic outcomes, or complications (Supplement 4).

4.1 Study limitations

This study has several limitations. First, its retrospective design carries an inherent risk of selection bias and residual confounding. However, in this analysis, we adjusted for imbalances in major baseline characteristics as much as possible using propensity score matching, which likely reduced these effects to some extent. Nevertheless, confounding due to unmeasured factors may still remain. Second, the sample size—although larger than previous reports—was still insufficient to detect small differences in rare complications such as implant failure or FRI. Third, although surgical and rehabilitation protocols were relatively standardized across the affiliated institutions, some variability likely remained. Fourth, patient-specific risk factors for complications (e.g., degree of osteoporosis, nutritional status) were not fully accounted for in this analysis.

Accordingly, prospective, adequately powered studies are warranted to validate the safety of early full weight-bearing in this patient population and to identify subgroups that may benefit most. Future research should also aim to refine patient selection criteria, incorporating both fracture-specific factors (e.g., fixation stability, prosthesis status) and patient-specific characteristics (e.g., bone quality, comorbidities).