The present study provides a large-scale as well as health-economic analysis of in-hospital PRCs after TSA. With more attention on improving surgical techniques and component design, the incidence of PRCs following TSA decreased from 2010 to 2014 with an exceeding 2.5-fold reduction (Fig. 1). An overall incidence of 1% of in-hospital PRCs after TSA was firstly identified in this study because previous studies mainly focused on the specific categories of PRCs. Interestingly, this overall incidence of in-hospital PRCs following TSA was lower than that following total hip arthroplasty (THA) (1.96%) while higher than that following total knee arthroplasty (TKA) (0.69%) [38, 39]. It was found that the dislocation of prosthetic joint was the most common PRCs, followed by PJI, PPF, and mechanical loosening (Fig. 2), which were to a large extent consistent with the previous studies on implant failure after shoulder arthroplasty [12, 14, 15, 35]. Coincidentally, the prior study conducted by our authors found the same results that dislocation was the most common PRCs, followed by PJI, PPF and mechanical loosening after THA . This consistency between TSA and THA reflect the similarity of these two procedures.
A previous literature reported that male was a significant risk factor for revision shoulder arthroplasty, nonetheless in this study, female was identified as a risk factor of PRCs in logistic regression analysis . This disparity may be due to that our study was the analysis of in-hospital PRCs after TSA which was an early stage or even unsure to require revision. Besides, female patients undergoing shoulder arthroplasty were associated with receiving blood transfusion which was at high risk of PRCs in this study . Furthermore, female was reported to be susceptible to suffer from depression which was also a risk factor of PRCs in our results . Patients with PRCs were 2 years younger than those without. Besides, from the perspective of age distribution, patients younger than 64 years accounted for a greater proportion in the PRCs group. In addition, in logistic regression analysis, advanced age (≥65 years) was identified as a protective factor of PRCs. On the contrary, younger age (<64 years) could statistically be determined as a risk factor of PRCs. To a great extent, this is consistent with previous studies which identified younger age as an independent predictor of PJI, failure or revision after shoulder arthroplasty [6, 10, 11, 14, 17, 24, 28, 30]. The etiology underlying this finding is unclear, but this can be used to educate patients, inform surgeons when counseling younger patients regarding their risks, and serve as an impetus for further investigation . Reasons for the Native American patients more likely and risky to experience PRCs are unknown and likely multifactorial. It is possibly that these minority populations have lower levels of cultural and healthy literacy. Therefore, these patients tend to receive lower-quality care or have difficulty understanding and complying with postoperative instructions [41, 42].
Compared with the Northeast region of the United States, hospital in the South was associated with an increased likelihood of PRCs, similar to our prior findings discussing about the PRCs following THA . Combined with the univariate analysis and multivariate analysis, patients in voluntary hospital were less likely and risky to experience PRCs. However, the reasons for these two hospital characteristic remain unclear and require further research.
Both the median LOS was 2 days no matter whether the in-hospital PRCs occurred, which is in line with prior studies that also found an mean LOS of approximate 2 days in patients undergoing TSA [29, 33]. Although the interquartile range of LOS presented statistical significance between two groups, this small difference was not obvious and might not be clinically important or meaningful . Luckily, the occurrence of in-hospital PRCs after TSA did not incur patients to death in this study, totally unlike to our previous findings in terms of THA and TKA, suggesting that TSA is safer as well as less traumatic to individuals compared with the other two procedures [38, 39]. Even so, the presence of PRCs still increased total charge during hospitalization, due to the costly treatments and cares of these complications [8, 10].
To moderate these costly events and further reduce the need for revision, preoperative identification of patients at increased risk of in-hospital PRCs after TSA is essential [43, 44]. Logistic regression was performed and several risk factors of in-hospital PRCs following TSA were identified (Table 3, 4, and 5). Similar to our prior study on PRCs after THA, patients with alcohol abuse were at high risk of PRCs after TSA, probably because excessive alcohol consumption is associated with weakened immune system and mechanisms, impaired phagocytic function, or induced cytokine abnormalities which may have a relationship with PJI [26, 39]. Depression has previously been reported to be associated with a series of adverse outcomes, such as PJI, sepsis, wound complications, return to the operating room for irrigation and debridement, extended LOS, readmission, revision surgery, and greater health care utilization after TSA, in-hospital PRCs after THA and TKA [38–40, 45]. Although the relationship between depression and in-hospital PRCs after TSA may seem obscure, imbalance of immune system inducing PJI likely explain this. Psychological distress produces a systemic state of inflammation leading to dysregulation of the immune system and a resultant susceptible host state . Additionally, depression itself may influence T-cell phenotype, and antidepressant medications have been shown to have negative immunoregulatory effects, further causing immune susceptibility .
Diabetic patients with vulnerable defenses against bacteria, or impaired wound healing because microangiopathic changes could reduce the tissue concentrations of antibiotics as well as cause local tissue ischemia, consequently are susceptible to PJI, and in-hospital PRCs following TSA, THA and TKA [38, 39, 46]. In concordance with our previous results, metastatic cancer in this study conferred the highest OR value (Table 3). This severe comorbidity possibly predisposes patients to PJI because of immunosuppressive conditions [38, 39]. Patients with either neurological disorders or Parkinson disease were at increased risk of PRCs likely because of the increased tone of the shoulder girdle musculature, the difficulties with rehabilitation, and stretching of the rotator cuff-capsule arthrotomy site, particularly the rotator interval . Besides, patients with Parkinson’s disease lack complete volitional muscular control and have asynchronous motor function, they appear to place the shoulder at high risk for instability [14, 39]. Additionally, constant tremors may result in implant loosening, which further increases the dislocation risk . Intriguingly, fluid and electrolyte disorders not only increase the odds of PRCs after THA and TKA, but also had an increased risk of PRCs after TSA [38, 39]. Renal failure has been reported as a significant risks factor for surgical site infection and revision following TSA [6, 8]. Patients with these two comorbidities to some degree may reflect the weakened status and insufficient immune function, and hence surgeons should be more cognizant of perioperative nephrotoxic medications, intraoperative hypotensive anesthesia, and postoperative fluid management in this complex patient population .
As expected, patients with indications for TSA such as AN, RA, or RCTA had higher odds of PRCs compared with osteoarthritis. Consistently, AN has been associated with a significantly increased risk for postoperative infection, dislocation, fracture, and revision surgery after TSA . Despite our authors previously reported that RA was associated with in-hospital PRCs following THA, it was firstly found in this study that patients RA undergoing TSA were risky to experience in-hospital PRCs . Corticosteroid therapy, alcohol abuse, immunosuppressive conditions, or posttraumatic poor soft tissue bed may partially explain the association of in-hospital PRCs with AN and RA [8, 16, 17, 26, 39]. Patients with RCTA commonly underwent prior soft tissue mobilization and rotator cuff manipulation, which are thought to affect prosthetic implant stabilization and may play a role in the increased dislocation, infection, and bleeding rates .
Arthroscopic intervention for shoulder osteoarthritis has been used as a measure to temporize pain or mechanical symptoms in order to delay joint arthroplasty and expedite time to return to recreational activities and physically demanding jobs . However, it was found that patients with history of PSA were risky to PRCs after TSA, similar to our prior finding that prior knee arthroscopy conferred high risk of PRCs after TKA . Furthermore, PSA has also been reported to be associated with a higher risk of infection after shoulder arthroplasty . On the basis of this association, surgeons should proceed with increased caution before performing an arthroplasty procedure in a patient with a history of PSA, and consider a lower threshold to rule out infection as well as the perioperative usage of antibiotic . It has been shown that blood transfusion confers increased risk of PJI and mechanical complications within 2 years after shoulder arthroplasty and is associated with in-hospital PRCs after THA or TKA [22, 38, 39]. Consistent with previous literatures, blood transfusion was the only one factor in this study significantly associated with in-hospital PRCs following TSA among the perioperative complications. Specifically, allogeneic blood transfusion may have an immunomodulatory effect that may lower the threshold for PJI through several mechanisms [22, 38].
The main strengths of the current study include its both large-scale sample and national representativeness with the power to investigate rare events, and the utilization of multivariable regression modeling to reduce confounding [31, 33, 38]. However, several limitations still require mention, mainly inherent to the use of the NIS database. First, misclassification or discrepancy in the process of coding and documentation may be produced as with any large administrative database [16, 17, 33, 38, 39]. Second, only in-hospital information of each patient is recorded, meaning any complication or adverse outcome that occurs after discharge such as readmission, functional status, and long-term follow-up can not be obtained in this database. This limitation may cause underestimating the incidence of PRCs [8, 9, 26, 29, 33, 38, 39]. Furthermore, only variables provided by the NIS database could be assessed. There are other potential procedural and component characteristics that possibly affect PRCs were unable to capture via the NIS database, such as surgical approach, length of surgery, type of anesthesia, amount of blood loss, cemented or uncemented components, and implant design [7–9, 15, 35, 38, 39, 47].