Cohort groups
Based on a retrospective cohort study between 2004 and 2012, unilateral RHA using a modular smooth-stemmed radial head implant (Evolve; Wright Medical Technology) was identified in 49 patients with comminuted radial head fractures (Mason type III and IV). Patients with less than 7-year follow up, other ipsilateral arm fracture, Essex Lopresti injury, insufficient medical record, and poor radiographic quality were excluded. A total of 33 patients were enrolled in this report. All the surgeries were approved preoperatively by the audit committee in our department with surgical indication well documented in the medical records. There were 13 female and 20 male patients with an average age of 44.76 ± 13.25 years (range, 24 to 75 years). RHA was the primary surgery after injury in 25 patients and, revision surgery after previous fixation failure (from 1 to 3 surgeries) in 8 patients. All had regular follow-up for more than 2 years postoperatively. The latest survey was at an average of 9.03 ± 1.74 years (range, 7 to 15 years). The 33 patients were divided into two groups based on the type of instability. Group A was valgus-type injury and consisted of 14 patients. Group B was radial head fracture dislocation in 19 patients.
Surgical procedure
All surgery was performed in general anesthesia and supine position by one single surgeon. Radial head fracture was explored with lateral Kocher approach. All patients underwent RHA with an uncemented modular prosthesis (EVOLVE radial head system, Wright Medical Group, Arlington, TN), which included a head segment and a smooth stem with options for head thickness and neck length. The size of the head segment was 1 to 2-mm downsized by reassembling the major fragments of the radial head on a sizing tray. The stem diameter was determined after sequential reaming of radius canal and calcar trimming. The final height of implanted prosthesis was adjusted by the combination of proper head thickness and neck length and was set with the proximal margin to reach the horizontal level of coronoid tip, which was confirmed by the mini c-arm image intensifier during surgery.
Lateral ligament-capsular structure was torn in 16 patients, and reattached using Mitek GII anchor (Mitek Surgical Product or Twinfix Ti anchor (Smith & Nephew Endoscopy, Andover, MA) suture fixation and No.2 ethibond (Ethicon, Somerville, NJ, USA) suture augmentation after completion of radial head replacement. Medial collateral ligament was explored and fixed with suture anchor repair in three patients (two in group A and 1 in group B) who presented grade III or more instability on valgus test after radial head prosthesis implantation.
Clinical evaluation
Study approval from Institutional Review Board (IRB 201800206B0) was obtained for patient data retrieval and invitation to patients in returning for clinical evaluation. Implant data of radial head prostheses were located through the National Health Insurance Administration Register, which contain original registration files and claim records for reimbursement. Clinical data review and collection was performed by one of the co-authors, who was blinded to the patients’ individual files. Functional survey was performed using the Mayo Elbow Performance Score (MEPS) and shortened Disabilities of the Arm, Shoulder, and Hand (QuickDASH) score. Residual pain around the involved elbow was recorded in the visual analog scale (VAS), ranging from 0 to 10.
Radiographic investigation
Radiographs of anteroposterior (AP) and lateral projections were performed for each elbow. AP view was taken with forearm in maximal supination and elbow in maximal extension; lateral view, with forearm in neutral rotation and elbow in 90° flexion. Two of co-authors performed radiographic evaluation including radiolucency around the prosthesis stem, presence of osteoarthrosis, and heterotopic ossification. High resolution images of AP and lateral views were meticulously compared with postoperative radiographs to identify the location of periprosthetic osteolysis (Figure 1). Evaluation of periprosthetic osteolysis on radiographs was performed by drawing a line across the stem at the location with maximal width of radiolucency, which was then calibrated with head and stem size (Figure 2). The sum of maximal width of radiolucency in AP and lateral views was recorded for each measurement. Radiolucency score of each elbow was defined as the average of measurement with two evaluators [13].
Statistical analysis
Descriptive statistics were calculated for analysis of the key variables with comparison between two groups. In the primary outcome survey, a chi-square test was used for calculating categorical data (sex, injured side, injured hand and complication rate); an independent t-test, for normally distributed data (patient age, time from injury to surgery and surgical times). For the secondary outcome measurement, the Mann-Whitney rank sum test was used in comparing the data that were not normally distributed (VAS, motion range, radiolucency sore, MEPS, and QuickDASH). A p-value of < 0.05 was considered statistically significant. Pearson correlation was used to explore the correlation between radiolucency and functional scoring.