**4. Discussion**

Patients with RSA after head-split PHF showed very good clinical and radiographic results, and the revision rate was 4% for those patients with clinical follow-up. Our cohort shows that patients who sustained a head-split fractures resulting from high-impact trauma had better results regardless of tuberosity healing. In addition, patients with a classic type 1 GT adjacent head-split showed better outcomes over those with multifragmentation of the articular face.

A current meta-analysis including 1303 PHF patients found the average anterior forward elevation flexion of about 122◦ with an average CS of 59 points and a total complication rate of 11% [22]. In our population, we achieved a similar complication rate, but our patients scored 74 points, on average, for the CS and achieved 148◦ elevation.

Healing of the GT led to favorable external rotation. Although GT healing was achieved in 61% of our cohort, this factor did not influence anterior forward elevation or any other functional outcome parameter besides external rotation, in contrast to a current meta-analysis [23]. This result could be biased by the small sample size in our cohort. However, the presence of the ER lag sign seems to be a prognostic factor not only for poorer external rotation but for subjective performance (SSV) since external rotation is involved in many activities of daily life.

Since our study population is older than 50 years, this study presents the advantages of RSA treatment for older patients. While complication rates for joint-preserving therapy options are high, young patients should be treated as soon as possible to minimize the risk of avascular necrosis [8,15]. High-impact trauma resulted in humeral head splitting that was first documented in dislocation fractures [4,6,8,24], ye<sup>t</sup> we observed a collateral glenoid rim fracture in 5 out of 45 patients. We hypothesize that this is due to the trauma mechanism of the humeral head being forced against the glenoid, which causes the head-split fracture but may also result in glenoid rim fractures due to either extremely high shearing forces or poor bone quality.

The first published consecutive cohort included eight patients (3x ORIF, 3x missed, 1x CRIF, 1x HA): the oldest patient was a 56-year-old female who sustained a low-energy trauma fracture that was initially missed on radiographic examination and left untreated [6]. The outcome was a stiff and painful shoulder with extremely poor function. Conversely, younger patients (19- and a 21-year-old males) within the cohort who both suffered highenergy traumas achieved excellent functional results after early open or closed reduction and internal fixation (CS was 89 and 100 points, respectively) [6,8].

Although our study population is older by far, this finding concurs with our data including patients who had experienced a high-energy trauma and achieved a better outcome post-RSA. We hypothesize that head-split fractures resulting from high-impact and low-impact injuries are two different entities. Active patients who are confident to cycle or ski regularly can anticipate an increased risk of experiencing a high-impact accident. These patients could be considered biologically young as their active lifestyle results in good bone quality according to Wolff's law [25]. In such cases, high shearing forces result in head-splitting fractures, but due to the grea<sup>t</sup> regenerative potential of vital tissue, patients can achieve better outcomes after RSA. Patients that sustain head-split fractures due to a fall from standing height were, on average, ten years older in our cohort and were not participating in an active lifestyle. For these cases, the fracture patterns are the result of poor bone quality and poor bone density due to immobility or osteoporosis.

Based on our cohort, articular-faced comminution of the humerus presents a serious treatment challenge for surgeons because very poor postoperative outcome can be expected. Our type IV patients had significantly poorer outcomes in all clinical scores measured. In addition, abduction and internal rotation were significantly lower for type IV fractures; external rotation was not affected by fracture type.

Although HA offers comparable results for head-split fractures (diagnosed on radiographs) compared to conventional three- or four-part PHF at short- to long-term followup [2,26], the complication rate of 36% and a revision rate of 12.5% should not be underestimated [2]. Compared to the cohort that has been treated with HA (*n* = 8), short-term results are comparable even though patients with RSA perform better in flexion but worse in external rotation [2]. Nowadays, RSA has limited the use of HA for PHF due to the current development and progress in shoulder arthroplasty [27,28].

Our study has several limitations, such as the retrospective design of this study as well as its small cohort. Differences between the different fracture types should not be considered as significant results; rather as trends. Our cohort analysis showed that headsplit fracture patients were quite old with many in their mid-70s at the time of surgery, which resulted in a high rate of loss to follow-up due to death alone (24%). A strength of this study was that all patients were treated by only one of two senior surgeons in the same operative technique and that the head-split fracture was diagnosed on CT scans.

Finally, while joint-preserving therapy is the precedent for young patients with unreconstructable PHF, the high complication rates of 44% in cases aged under 55 years and up to 50% in general dictate the greater likelihood for secondary surgery due to osteonecrosis or nonunion [15,29]. As RSA techniques develop and push the boundaries of age due to good results in complicated fracture situations, long-term studies must continue to monitor whether young patients benefit more from early primary or later secondary RSA treatment.
