HSA loss was maximal from 1 to 3 months postoperatively, in agreement with the literature [
8,
17]. The average healing time was 4.4 months; the HSA remained essentially unchanged after 6 months. Of fractures that did not heal within 3 months, significant differences were evident between the HSA in the immediate postoperative phase and the values measured over 1–3 months. We suspected that this might reflect early voluntary exercise. Most patients could endure residual pain 1 month after surgery and, after 2 months of functional exercise, limb function partially recovered. Hence, some patients may have commenced internal forward rotation, and even stretching and weight-bearing activities. Muscle action during such exercises might change the HSA. In addition, the biomechanical stability of the proximal humerus may be compromised by fracture-healing during this period. Primitive bone callus formation varies among individuals; premature exercise may change the HSA but further research is required to confirm this.
The mean HSA of the with-MS group was significantly greater than that of the without-MS group at 3 months postoperatively (
p = 0.003); MS was associated with less reduction loss. Overall, the reduction loss rate was significantly higher in the without-MS group (
p = 0.005), consistent with other studies [
18‐
20].Jung et al. [
21] reported that inadequate MS was an independent risk factor for reduction loss after surgery to treat PHFs. In the absence of MS, the fixed-angle screws must serve as perpendicular struts aiding the humeral head to resist varus displacement. Therefore, MS is important for maintenance of reduction in patients with PHFs; surgeons must maximally restore MS.
Earlier studies on the functional and radiological outcomes of PHFs reached conflicting conclusions. Yüksel et al. [
22] found no relationship between the HSA and the Constant score (which measures recovery); however, Bai et al. [
23] reported that an HSA change > 10° significantly reduced the score. We investigated the relationship between the HSA and functional recovery in PHF patients in whom locking plates were placed and found that HSA change was negatively associated with recovery. Court-Brown et al. [
24] found no association between increasing varus angulation and shoulder function; Capriccioso et al. [
25] reported that the initial surgical neck displacement (varus or valgus) did not significantly affect functional outcomes. Conversely, 16 of our patients experienced varus malunion; their mean Neer score was 73.6 ± 11.6, thus lower than that of non-varus malunion patients (84.2 ± 9.2,
p < 0.001). However, we would mention that three patients in the varus malunion group enjoyed excellent or satisfactory functional outcomes; all uninjured side HSAs were < 130°, and the Neer scores were 84, 92, and 83. A previous study found no significant between-group difference in bilateral HSAs [
26].Therefore, we used multivariable linear regression to control for confounding effects; we found that Neer type 4 fracture, the postoperative HSA difference between the injured and uninjured side, and the change in HSA to completion of follow-up may influence the Neer score. Similarly, a previous study reported that, in patients with two- and three-part fractures, the overall functional results were usually satisfactory; but this was not always the case in those with four-part fractures [
27,
28]. We suggest that the latter fractures usually develop in osteoporotic patients, often associated with proximal fractures and medial metaphyseal comminution [
29].We found that the postoperative HSA difference between the treated and uninjured side, and the change in HSA to completion of follow-up, were negatively associated with the Neer score. Thus, restoration of the native HSA afforded better functional recovery. We identified a trend toward improved function when it was possible to restore the HSA of the injured side to a value similar to that of the uninjured side. Also, we could not rule out an effect of reduction loss in patients for whom the pre-injury HSA was < 130°.
Our study had certain limitations. First, it used a retrospective design and the sample size relatively small; we also cannot exclude potential effects of unmeasured factors. Furthermore, we prescribed a general postoperative rehabilitation protocol, but some patients may have individualised needs regarding the timing of passive range-of-motion exercises; this could affect the clinical and radiographic outcomes. A future prospective study with a larger sample size is required to validate our findings.