1 Introduction

The indications for anatomic total shoulder arthroplasty (aTSA) implants have expanded to include osteoarthritis, rheumatoid arthritis, osteonecrosis, avascular necrosis, proximal humerus fractures, and irreparable rotator cuff injuries.¹ Likewise, while the average age for shoulder replacement is still between 65 and 70 years old, the operative age range for aTSA has also expanded. However, the use of shoulder replacement in a young population (i.e., <50 years) and an older population (i.e., >80 years) is somewhat controversial due to increased risks, such as subsequent revision arthroplasty or increased comorbidity rates, respectively.

In other orthopedic (sub)specialties, age has been negatively correlated to short-term postoperative outcomes.^2–4 For example, in outpatient hand and elbow surgery, increased age was indicated as an independent predictor of 30-day readmission rates.³ In arthroplasty procedures, such as total knee (TKA) and hip arthroplasty (THA), short-term studies have shown that increased age on the date of surgery leads to more 90-day readmissions, prolonged LOS, discharge to rehabilitation centers, and overall complications.^2,5–7 At the national level, the impact age at surgery has on TSA outcomes is slightly inconsistent. Most studies found that increased age is associated with higher complication (e.g., cardiac, psychosis, anemia), mortality, and readmission rates, LOS, and healthcare charges.^8–11 Yet, Schairer et al. showed that age at surgery had no impact on risk for readmission after TSA in a sample of 26,218 patients.¹² Nevertheless, functional outcomes at mid-to long-term follow-up after TSA have proven to be favorable regardless of age at surgery.^13–16

To date, no study to our knowledge has reported on the impact age has on short-term shoulder specific complications (i.e., dislocation, hardware/prosthetic complications) at the national level. Moreover, the most recent study in the current literature that we identified used data up through 2016. For these reasons, the present study attempted to characterize the impact age has on complication and readmission rates following aTSA using updated 2016–2018 data consisting of more than 42,000 aTSA patients. We hypothesized that age would be positively correlated with overall total charges, length of stay (LOS), complication rates, and 30-/90-day readmission rates in individuals undergoing aTSA.

2 Materials and methods

The Healthcare Cost and Utilization Project National Readmissions Database (HCUP-NRD) from 2016 to 2018 was queried for patients undergoing aTSA. The dataset includes an estimated 17 million admission records per year, of which 42,505 met inclusion criteria based on the International Classification of Disease Tenth Revision-Clinical Modification (ICD-10-CM) codes (0RRJ0JZ, 0RRK0JZ). Data collected included demographic data such as sex and age, procedural cost data, and billing codes.

We excluded patients under the age of 18, those who died upon first admission, and patients who underwent a nonelective surgery. Additionally, due to the importance of the 90-day readmission metric, we excluded all patients with a discharge date after September of each year. Patients were divided into five categories based on age, 18–49 years old (y.o.), 50–59 y.o., 60–69 y.o., 70–79 y.o., and 80+ y.o.. A sub-analysis was also run comparing patients under 65 y.o. to those older than 65 y.o. in order to provide clarity to the outcomes for age groups on either side of the average aTSA age at surgery. Demographic variables collected included sex, insurance type, and Charlson comorbidity index. Outcome variables collected were discharge location, length of stay (LOS), total charges, 30-/90-day readmission, hardware related complications, intraoperative complications, neurologic complications, peripheral vascular complications, respiratory complications, genitourinary complications, gastrointestinal complications, postoperative shock, postoperative infection, foreign body reaction, wound dehiscence, deep vein thrombosis, pulmonary embolism, dislocation, surgical site infection/joint infection, prosthetic complications, and noninfectious wound complications. The term “intra-operative complications” here refers to postprocedural and or intraoperative complications of the musculoskeletal system as defined by the NRD and classified by the ICD-10 codes beginning with “M96”. Further, the term “post-operative infection” refers to any infection following a procedure while “surgical site/joint infection” refers to postoperative infections of the joint prosthesis or the incision site in particular.

For statistical analysis, categorical variables were analyzed using Chi-square test and continuous variables were compared using an ANOVA. Multivariate logistic regression controlling for sex, myocardial infarction, congestive heart failure, peripheral vascular disease, COPD, rheumatoid disease, diabetes without complications, diabetes with complications, and renal disease was performed with age as an independent factor. Logistic regressions were only performed on subgroups where the total number of complications exceeded 100 to ensure appropriate statistical power and model convergence. Chi-square tests were not performed if a group had zero data points. P-values were reported for all statistical tests with <0.05 being considered significant. Odds ratios were also reported for the multivariate logistic regressions. All tests were conducted using the SciPy 1.6.1 python package.

3 Results

In total, we captured 42,505 patients that met our inclusion criteria. There were 1,541, 6,552, 16,364, 14,694, 3,354, patients in the 18–49, 50–59, 60–69, 70–79, and 80+ cohorts respectively. Older age cohorts had a significantly lower proportion of males (p < 0.001), lower proportion of Medicaid (p < 0.001), private (p < 0.001), self-pay (p < 0.001), and other (p < 0.001) insurance, and higher proportion of Medicare insurance (p < 0.001). The Charlson 0 (p < 0.001), 1–2 (p < 0.001), and 3–4 (p < 0.001) scores varied significantly among the age cohorts but the Charlson ≥5 score did not (p = 0.18). Home health care discharge was most common in the 80+ cohort (24.7%; p < 0.001) as was transfer to other discharge (17.8%; p < 0.001). Routine discharge was most common in the 18–49 cohort (87.6%; p < 0.001). Length of stay varied significantly among the age cohorts (p < 0.001) with the 80+ cohort exhibiting the highest average LOS (1.9 days) followed by the 70–79 (1.6 days), the 60–69 cohort (1.4 days), the 50–59 cohort (1.4 days) and the 18–49 cohort (1.4 days). Total charges (p = 0.40) and discharge against medical advice (p = 0.66) did not vary significantly across the cohorts (Table 1).

Age cohorts varied significantly in the incidence of hardware related complications (p < 0.001), surgical site infections (p < 0.001), prosthetic complications (p < 0.001), 30-day readmissions (p < 0.001), and 90-day readmissions (p < 0.001); for a graphical representation of the 30- and 90-day readmission rates see Fig. 1. The 80+ cohort experienced the highest incidence of both 30- and 90-day readmission (4.0% and 6.9%, respectively), while the 18–49 cohort experienced the highest incidence of hardware related complications (5.4%), surgical site infections (0.9%), and prosthetic complications (4.2%). Age cohorts did not vary significantly in the incidence of peripheral vascular complications (p = 0.86), gastrointestinal complications (p = 0.80), wound dehiscence (p = 0.94), deep vein thrombosis (p = 0.10), or noninfectious wound complication (p = 0.65). They did differ with respect to intraoperative complications (p = 0.029), pulmonary embolism (p = 0.044), shoulder dislocation (p = 0.003), and postoperative infection (p = 0.034) (Table 2).

Close

Fig. 1 Bar plot of 30-day and 90-day readmission rates for each age group (18–49, 50–59, 60–69, 70–79, and 80+ y.o.). P-values indicate significant differences between groups with chi-square testing. * and ** indicates that the 30-day and 90-day readmission rates for the 80+ subgroup were significantly higher when compared to all other subgroups. All other subgroup comparisons were not statistically significant.

Export to PPT

The multivariate logistic regression found that age as an independent variable was significantly associated with 30-day readmission (OR: 1.02; CI: 1.009–1.025; p < 0.001), 90-day readmission (OR: 1.01; CI: 1.001–1.011; p = 0.020), pulmonary embolism (OR: 1.03; CI: 1.008–1.048; p = 0.006), hardware-related complications (OR: 0.98; CI: 0.971–0.983; p < 0.001), rate of general infection (OR: 1.01; CI: 1.002–1.027; p = 0.022), prosthetic complications (OR: 0.97; CI: 0.965–0.980; p < 0.001), surgical site/prosthetic infection (OR: 0.97; CI: 0.954–0.982; p < 0.001), and shoulder dislocation (OR: 0.99; CI: 0.977–0.999; p = 0.026). No association was found between age and intraoperative complications (p = 0.65), or respiratory complications (p = 0.076) (Table 3).

When comparing patients age of 65 and lower (n = 14,856) to those 65 and older (n = 27,649), we found that patients in the ≥65 cohort had significantly lower incidence of hardware related complications, shoulder dislocations, surgical site infections, and prosthetic complications. Alternatively, the ≥65 cohort had higher rates of respiratory complications, postoperative infection (general), pulmonary embolism, 30-day readmissions, and 90-day readmissions (Table 4). Further, age of ≥65 was an independent predictor for decreased hardware-related complication, prosthetic complication, dislocation, and surgical site-infection events and increased postoperative infection (general) and pulmonary embolism events (Table 5).

4 Discussion

Shoulder replacement surgery has seen continued growth and usage in both younger and older patients. A 2015 study predicted that the demand for shoulder replacement will increase by 330% and 755% by the year 2030 for patients younger than 55 years of age and older than 55 years of age respectively.¹⁷ Due to this high demand, it is imperative that we understand the impact a patient's age at the time of surgery has on short-term outcomes following shoulder replacement. In this study of 42,505 patients, we showed that older patients had higher rates of non-routine discharge, intraoperative complications, postoperative infections, and 30-/90-day readmissions but were less likely to experience hardware complications and prosthetic complications. Overall, total charges remained stable regardless of a patient's age at surgery.

Excellent mid-to long-term clinical outcomes have been reported in the current literature for younger (<60 years) and older (>80 years) patients following shoulder replacement.^13–16 However, due to the stress incurred on the body during arthroplasty, complications often occur within the first few postoperative months. In a report by Fox et al., the complication rate within 30-days of shoulder replacement was noted to be 5.6%.⁸ This rate has been reported to be higher amongst older patients. For example, Lovy et al. showed that age was associated with increased short-term severe adverse events (e.g., thromboembolic event) in a study of 5801 aTSA and reverse TSA (rTSA) patients (OR (95% CI): 1.04 (1.02–1.06)).¹¹ Similarly, Griffin et al. showed that in their sample size of 58,790 shoulder arthroplasty patients, those ≥80 years old had higher rates of anemia (11.6%), psychosis (1.9%), and cardiac complications (1.1%) compared to younger patients (p < 0.001).¹⁰ In this study, older age (i.e., 70–79 and ≥ 80 years old) was associated with higher rates of intraoperative complications, postoperative infection, and pulmonary embolism. However, age was only an independent predictor for pulmonary embolism occurrence. Increased rates for short-term complications such as the ones noted here are likely due to the additional comorbidities present in older populations, highlighted by the fact that not every significant finding in the univariate-translated to the controlled multivariate-analysis. However, as the rates of the aforementioned complications, both here and in the literature, were shown to differ between age groups, surgeons should address these potential complexities with patients preoperatively. Additionally, the use of anticoagulation must be strongly considered in elderly individuals who are more likely to experience a thrombotic event. This is especially true as Lovy et al. showed pulmonary embolisms to be the cause of 7.6% of readmissions after TSA.¹¹

Moreover, in this study increasing age was associated with less overall shoulder specific complication events (i.e., hardware-related, prosthetic, dislocation). It is possible that younger populations who are more active put additional stress on their new prosthesis leading to increased complications. Similarly, patients presenting at younger ages likely have advanced disease leading to their initial presentation. Previous reports have indicated that dislocations are amongst the most common surgically related causes for readmission.^11,12 At mid-term follow-up Wagner et al. also saw that younger patients had higher rates of mechanical failures comparatively.¹⁸ Thus, surgeons must be particularly aware of the increased risk of prosthetic complications such as dislocations in their young and old patients alike.

Importantly, age was seen to be an independent predictor for increased rates of readmissions at both the 30- and 90-day postoperative mark. Previous studies looking at shoulder replacement have found similar findings.^11,12,19 These prior studies postulated the negative correlation with age to be associated with increased health complications in older demographics. We agree with the prior research in this regard. Additionally, we suspect that there is a component of polypharmacy, prior hospitalizations, and functional status at play. For example, prior hospitalizations have been associated with increased risks for readmissions overall.²⁰ With a larger proportion of older individuals who underwent aTSA having non-routine discharges in this study, it is a worthwhile consideration for surgeons to include increased postoperative aid into the recovery regimen of their geriatric patients.

An interesting result seen in this analysis was that of “infections”. While increasing age was an independent predictor of postoperative infections it was also a predictor for less surgical site/prosthetic joint infections. In regards to postoperative infection, previous research has shown non-correlations between increased age and postoperative infection in shoulder arthroplasty.¹⁰ Thus, it is possible that the findings we appreciated may be due to lower overall health status in older populations leading to infection development that is not necessarily related to the surgical site or prosthesis. Alternatively, studies have shown that younger patients experience increased prosthetic infections after shoulder replacement.¹⁸ Yet, Diamond et al. showed that ages 70–74 years old had the highest rate of prosthetic joint infection after TSA.²¹ With regards to site/prosthetic infections here, there may be an influence of increased LOS in older adults on infection development. In this cohort LOS increased with age as did the rate of non-routine discharge. Thus, older adults may have fewer surgical site infections due to their increased stays and monitoring in healthcare facilities. Ultimately, given that infection is one of the primary reasons for TSA readmission, more research is needed to further parse out the reason for the differences in the infection rates appreciated here.¹²

To this end, previous studies have also shown increasing age to be associated with longer LOS after shoulder replacement.^10,22 For example, Griffin et al. showed that the average LOS in patients <50, 51–79, ≥80 years old was 2.9-, 2.9-, and 4.0-days respectively after shoulder replacement.¹⁰ Similar findings have been seen across different arthroplasty procedures.⁵ This is likely due to the increased postoperative needs of older patients including delirium and aspiration precautions. Older patients in this study had higher Charlson Comorbidity scores, indicating an increased comorbidity burden. Nevertheless, the overall healthcare charges did not differ between the five age cohorts. Older patients should be made aware that they are at an increased risk for extended stays following their surgery, but reassured that overall costs will not be impacted because of their age.

We also performed a two-cohort sub-analysis at the 65-years of age cut-off. This was done with the intention of characterizing complication rates in individuals who are age-eligible for Medicare benefits. In this sub-analysis, age ≥65 was found to be an independent predictor for increased general infection and pulmonary embolism rates; however, ≥65 was a protective factor for hardware related complication, prosthetic complication, dislocation, and surgical site infection events. We believe it is important for surgeons and healthcare systems to be aware of these complications in the Medicare demographic population should shoulder arthroplasty be considered in the hospital readmissions reductions program in the future.

Certain limitations of this retrospective study need to be addressed. Of note, the cohorts in this study were not matched in terms of their sample sizes. Additionally, using a national database limits our understanding of longer-term outcomes, surgical complexity, and or a patient's baseline health. Further, a final consideration to make is whether or not odds ratios falling between 0.97 and 1.03 are clinically significant. Nevertheless, this study provides substantial insight into the impact age has on short-term complications and readmissions at the national level. Future studies should seek to compare short-term outcomes after aTSA using the most common age range (i.e., 60–70 years) as a reference point to compare against. Additionally, future researchers may wish to expand upon the present study into comparisons of aTSA, rTSA, and hemiarthroplasty in order to determine which procedure is most appropriate based on age at presentation for arthroplasty.

5 Conclusion

Age has varying effects on complication rates following aTSA. In this analysis of 42,505 patients, an increasing age was shown to be an independent predictor for increased readmission, postoperative infection, and complication rates yet it decreased a patient's risk of developing implant-related complications and surgical site infections. Additionally, age was shown to be associated with longer lengths of hospitalization but had no impact on overall healthcare charges. Ultimately, this study identified important variables to consider when caring for both younger and older patients requiring shoulder replacement surgery. Shoulder surgeons may choose to include these elements in their preoperative informed consent conversations with patients.

Author contributions

Christopher White – conceptualization, data curation, formal analysis, investigation, methodology, writing original draft, review & editing.

Akiro Duey – conceptualization, data curation, formal analysis, investigation, methodology, writing original draft.

Bashar Zaidat – data curation, formal analysis, investigation, methodology, review & editing.

Troy Li – data curation, formal analysis, investigation, methodology, review & editing.

Samuel Cho – conceptualization, investigation, methodology, project administration, resources, review & editing, validation, visualization.

Jun Kim - conceptualization, investigation, methodology, project administration, resources, review & editing, validation, visualization.

Paul Cagle – conceptualization, investigation, methodology, project administration, resources, review & editing, validation, visualization.

Funding/sponsorship

This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

Informed consent

N/a.

Institutional ethical committee approval

N/a.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.