A retrospective propensity score matched case-control study of prospectively collected data was performed with data from 2011–2018. From 2011–2016 all RTSA for PHF were cemented (Zimmer anatomical shoulder fracture system: Zimmer, Warsaw, IN, USA). With the introduction of a new shoulder arthroplasty with a more anatomical stem design in 2017 (Medacta shoulder system: Medacta, Castel San Pietro, TI, CH), we started to use uncemented RTSA for PHF. The radiographic and clinical data of all patients with RTSA for PHF are collected prospectively in our institutional registry. The regular follow-up controls are after 3 months and 1, 2, 5 and 10 years post-operatively. The functional outcome is assessed by a specifically trained study-nurse (M.M.) at each follow-up appointment and it includes the absolute and relative Constant score (CS) [17]. Clinical and radiographic complications are evaluated by one of two shoulder specialists (B.J. and C.S.) who, if indicated, also recommend revision surgery.
The data from this registry were used to compare the clinical and radiographic outcomes of patients with RTSA for PHF with uncemented stems (group nC: 2017–2018) to a matched group of patients with cemented stems (group C: 2011–2016). The inclusion criteria were: isolated PHF treated with primary RTSA [3] and a two year follow-up.
We were able to include 17 consecutive patients treated with uncemented stems to group nC and performed a 1:2 matched pair analysis by age and gender with these 17 patients; for comparison we selected 34 suitable cases from a larger patient collective of the cemented subgroup. The pathway in Fig. 1 shows the patient selection and the matching process.
Surgical techniques and prosthesis models
All operations were carried out using the delto-pectoral approach in the beach chair position; the same tuberosity refixation technique was used for both cemented and uncemented RTSA following the protocol of Fucentese et al. 2014 [18]. All operations were performed by a total of four orthopedic surgeons, all specialists for shoulder surgery. The Zimmer anatomical shoulder fracture system: (Zimmer, Warsaw, IN, USA) was used for all cemented prostheses using the 3rd generation cementation technique with PALACOS G (Haereus, Hanau, HE, DE). For the uncemented RTSA we used the Medacta shoulder system (Medacta, Castel San Pietro, TI, CH) with standard stem lengths. Postoperative rehabilitation was the same for all patients and included sling immobilization for 6 weeks with pendulum exercises after 2, functional exercises after 4 and weight bearing exercises after 12 weeks.
Radiographic analysis
All radiographic measurements were performed twice by two of the authors (M. K. and M.O.) The interobserver variability was confirmed by interclass correlation coefficient. For the final radiographic analysis a consensus reading of both raters was used. For the radiological analysis, we reassessed the fractures on CT preoperatively and on X-rays (AP and Neer) pre- and postoperatively, as well as the X-rays of 2 years after RTSA (AP neutral, AP internal rotation, axial, Neer) .
We determined bone quality on the AP X-ray of the fracture according to the deltoid tuberosity index (DTI) [19] and the fracture type according to Neer et al. 1970 [20]. The anatomical position and healing of the tuberosity were assessed on the final 2 year post-op X-rays. According to the instructions of Wright et al. 2019 [7], the tuberosity healing was rated as: healed in the anatomic position, dislocated or resorbed.
Loosening of the prosthesis was assessed according to Sperling et al. 2000 [21] based on the number and location of the lucent lines. The occurrence and the grade of scapular notching was evaluated pursuant to the Nerot-Sirveaux Classification [22, 23]. Bone resorption at the proximal humerus was assessed as stated by Aibinder for the grading, and by Denard for the location [12, 15]. Bone resorption was graded from 1 to 3. However the grading was carried out differently on the lateral and medial sides. On the lateral side, grade one is resorption of trabecular bone only, grade two is a thinning of the cortex and grade three is a complete cortical resorption down to the prosthesis. On the medial side, bone resorption was only registered when the cortex was thinned and graded from one to three according to the extent of the resorption zone. If the resorption occurred within the first third of the ingrowths surface, it was classified as grade one. If resorption happened from the second to the last third it was designated as grade two and absorption beyond the ingrowth surface was grade three [12]. The stem to humerus filling ratios were measured for all uncemented prostheses following the instructions of Denard et al. 2018 [15].
Statistical analysis
All statistical analyses were performed using R (R: A language and environment for statistical computing: R Foundation for Statistical Computing, Vienna, Austria - URL http://www.R-project.org/). We applied R MatchIt package for propensity score matching at a 1:2 ratio. Descriptive statistics included means, ranges, standard deviations and proportions. To assess the interobserver variability we calculated the interclass correlation coefficient ICC3 according to Shrout et al. 1979 [24]. Comparative statistics included t-test and Chi-square test (Wilcoxon and Fisher exact test were applied where alternatively appropriate). The confidence level for rejecting the null hypothesis was set at 95% (p-value < 0.05).