Response to letter to the editor regarding “Does primary treatment of proximal humerus fractures show favourable functional outcomes over secondary treatment with reverse shoulder arthroplasty?”

We appreciate the correspondence regarding our manuscript “Does primary treatment of proximal humerus fractures show favourable functional outcomes over secondary treatment with reverse shoulder arthroplasty?” ¹ In our response, we seek to clarify the specific points put forward.

The first point raised by the reviewers addresses the potential impact of psychological factors on the outcomes of the published analyses regarding the comparison of primary and secondary RSA in the treatment of proximal humerus fractures (PHFs). We acknowledge that psychological factors are influential on the outcome of both patients who are primarily treated for a PHF as well as patients who are treated surgically due to a failed primary treatment. However, we want to emphasise that the included patients in the delayed group all had a fracture sequela, which is specified in the article. Contrary to what the authors suggest, secondary treatment is not a ‘choice’.

If psychological factors were to be incorporated into the analysis, it is anticipated that these factors would primarily affect the patient-reported outcome measures (PROMs). Therefore, we complemented the PROMs with ROM to further add to the objectivity of the measures being compared. These psychological factors could be equally present in both groups though, as the theory put forward by the responders that these might be overly represented in the secondary treated group is insufficiently substantiated. Furthermore, the fact that psychological factors are an important element in delayed treatment would suggest preference for primary treatment. Overall, because of the aforementioned arguments, we consider our study results sufficiently robust in the ability to withstand the influence of psychological factors.

The second point concerns biomechanical stability in conservative treatment. Since we conducted a retrospective study, we lack data on the preoperative rotator cuff function. This could, of course, influence the outcomes after RSA. However, the concept of Grammont's principle posits that with an RSA, arm function is less dependent on the rotator cuff. ² Naturally, if the rotator cuff remains well-preserved, function improves. One of the critical factors in this regard is the healing of the tuberosities. If they have not healed properly, there is a diminished rotational function, which is why we specifically examined this aspect. Additionally, the impact of scapula mobility on PHFs has not been widely acknowledged to be a large factor at play, among other things since scapula position does not change because of a PHF. We do know this entity (scapula dyskinesia) is consequential to a malunited (shortened) clavicular fracture for example.

Thirdly the authors indicate that it would be interesting to know how many surgical procedures could be avoided with initial conservative treatment. Clearly, there exists a subset of patients for whom conservative treatment is an excellent option. The fractures considered for primary surgical treatment, however, are complex fracture types (headsplit fractures, fracture-dislocations, severely displaced three- or four-part fractures, etc.). Our study centres on a cohort of RSAs for fracture sequelae compared to patients who received a primary RSA for the above-mentioned indications.

Given the retrospective nature of our study, we don’t have preoperative functional data. However, other studies have preoperative functional outcomes compared to postoperative treatment, revealing improved outcomes following secondary RSA. ³ Our objective is not to advocate a one-size-fits-all approach but rather to underscore the relevance of considering primary RSA, particularly in cases of complex fracture patterns in elderly patients. This perspective aligns with the data presented in our article.

The fourth point also is a reaction to the retrospective nature of this study. As you can see in our baseline characteristics, the primary treated patients had more complex fractures compared to the secondary group. If anything, the potential bias due to the retrospective nature of the study would yield an even larger advantage for the primary treated group. Moreover, we tried to include medical comorbidities by including the ASA classification (which also considers smoking and other comorbidities) at the baseline characteristics. Furthermore, we tried to minimise the confounding effects of the baseline characteristics by performing a linear regression analysis.

Based on the points of the correspondents, we conclude that retrospective studies always carry a risk of bias, and ideally, an RCT would be preferred. However, conducting an RCT for studies of this nature would likely necessitate a substantial timeframe, spanning many years and the results would potentially be outdated by the time they got out due to new (operative) possibilities. A carefully conducted retrospective study with statistical correction of slight baseline differences is the next best study design. With our study, we maintain that, despite the points raised that always surround a retrospective study design, it does give valuable insights into the decision-making in the treatment of PHFs.