**4. Discussion**

Evaluating the value of health care is of paramount importance to keep improving patients' quality of life and optimizing associated costs. Hospitals' digitalization is still ongoing and offers a grea<sup>t</sup> potential for patients' evaluation along their entire care path. Beyond this, the real challenge that often arises in VBHC discussions is the absence of external benchmarks which compels us to compare our results within our institution or at different time intervals. The authors of the present study therefore created a new value-based dashboard for TSA, which allows an objective comparison with standard references.

According to our results, 78% of the TSAs performed at our institution offered a substantial value to patients. It is worth noting that 41 patients (35%) had a substantial delivered value although they had either a quality of care below the expectations or an excessive direct cost (Figure 2). This emphasizes the importance of evaluating both indicators together rather than interpreting them independently from each other.

Different authors recently evaluated the value delivered by TSA at short term using different methods [12–14]. Menendez et al. [14] defined the delivered value as the postoperative ASES score divided by the hospitalization time-driven activity-based costs. More comparable to our value calculation method, Berglund et al. [13] divided the ratio of PROM improvement (in units of MCID) by the total hospitalization cost. Both aforementioned studies found that reverse TSA was associated with a decreased delivered value compared to anatomic TSA, which corroborates our findings. Although it was expected given that reverse TSA has a higher cost associated with the managemen<sup>t</sup> of rotator cuff deficiencies, it is important to note that such an association can be reversed at some point since different studies already revealed mid- or long-term concerns on anatomic TSA (glenoid loosening, difficult revision procedures, and disappointing outcomes) [15–17]. Furthermore, the indications for these two procedures can be different and further analyses with matched cohorts are needed [18].

In our study, the delivered value was higher for shoulders with a lower preoperative function or higher pre-operative pain since greatest clinical improvements are usually observed for patients with worse preoperative health [19]. Our analyses also revealed that current or former smokers had a lower delivered value compared to non-smokers. The negative impact of tobacco use on outcomes after TSA is well reported [20–23] and emphasizes smoking cessation programs [24]. In the next decades, machine learning algorithms might be able to accurately predict postoperative patient outcomes based on their preoperative characteristics [25]. Such prognostic tools would help manage patient expectations [26] and avoid surgery for patients who would not benefit from it, thereby reducing associated risks for the patients while lowering costs for the health care system.

Different authors already worked on the creation of VBHC dashboards/scorecards [2,27]. Riley et al. published an innovative method to illustrate patient value following total hip and knee arthroplasties [2]. This method consisted of comparing the results of different surgeons within the same institution, which motivates them to outperform for the sake of their patients. However, the use of internal references such as orthopedic department averages for direct costs or PROMs can be misleading. For instance, implementing this method in small institutions where only one surgeon works in a specific medical field would be unwarranted. Furthermore, this method could reveal outstanding results for a surgeon even though the entire department has bad outcomes. In our study, we proposed to use SCB thresholds for the interpretation of PROMs improvements and to estimate the direct cost reference by using the DRG-based standard reimbursement system. The proposed dashboard can guide toward patient value improvement before a new methodology and strong external benchmarks using data from several hospitals are created.

Continuous improvements based on measuring the own performance in order to provide the best possible value to customers has been a key success factor for successful companies across all industries. VBHC is bringing this principle into health care, to the grea<sup>t</sup> benefit of patients and the system. The mentality of the different health care players is changing, and the competition slowly shifts from micro-costing only to patient outcome and cost optimization. It is setting the stage for a new way of thinking, collaborating, and competing, thereby opening new opportunities to reinforce excellence in care. The combination of medical expertise with an open mindset for change and self-evaluation is essential. In this sense, VBHC is redefining the basis of what leadership is for healthcare professionals. An essential development will be the emergence of new reimbursement models rewarding better outcomes. This will again require a fundamental change in people's mindset, while providing a grea<sup>t</sup> opportunity for early adopters to accelerate change.

Our study has several limitations. First, our analyses only illustrate the delivered patient value at short follow-up. Furthermore, patients for whom complications were noted might have been double-penalized since such clinical outcomes might also affect PROMs. To reduce the aforementioned bias, an artificial floor was used for patients who had a negative change in PROMs. Second, the weighting of clinical outcomes was solely based on the senior surgeon's experience. The logic was to attribute an equal weight (0.2) to the five principal outcomes that are crucial for shoulder arthroplasty success: global function (ASES and Constant scores), pain, complication (intra- or post-operatively), PJI, and revision. If a principal outcome comprised different sub-outcomes (e.g., Global function), the weight was then split proportionally to have a similar weighting between sub-outcomes (e.g., 0.1 for the ASES score and 0.1 for the Constant score). A Delphi method engaging the patients, insurance providers, and other key important players would have been more appropriate. Third, our outcome and cost indicators were not risk-adjusted, which can represent a bias if comparisons are made between two surgeons with differences in case mix and patient populations. Fourth, the direct cost reference was estimated to be 44% of the standard TSA reimbursement based on our patients with basic insurance only. A thorough analysis of the DRG-based standard reimbursement system should be performed and published so that each institution knows the theoretical amount supposed to cover direct costs. Fifth, the MCID and SCB values might change across different patient populations. Lastly, a broader analysis focusing on a specific pathology (e.g., glenohumeral arthritis) rather than on a particular treatment (e.g., TSA) would be more in line the VBHC concept.
