Increased revision rates in shoulder arthroplasty following shoulder arthroscopy

Abstract

Introduction

Total shoulder arthroplasty (TSA) and reverse TSA (rTSA) are successful treatments for end-stage shoulder arthritis. However, it is unknown whether prior arthroscopy is associated with an increased risk for revision surgery. This study investigates if undergoing a shoulder arthroscopy in the year prior to primary arthroplasty increases risk of revision surgery within 2 years.

Methods

Patients who underwent TSA or rTSA between 2005 and 2017 were identified in a natinal claims database and stratified into two cohorts: (1) individuals with a history of shoulder arthroscopy prior to arthroplasty and (2) individuals with no documented history of arthroscopy prior to arthroplasty. These cohorts were propensity matched based on demographic and comorbidity factors. Univariate analysis was used to determine differences in revision rates, aseptic loosening, periprosthetic fracture, and infection between the two cohorts.

Results

Seven hundred and eighty-eight patients were successfully matched from the two cohorts. Revision surgery (3.4% vs. 1.4%, p  =  0.001) and aseptic loosening (2.2% vs. 0.8% p  =  0.021) were significantly more common in the arthroscopy cohort. Periprosthetic fracture and periprosthetic infection were not found to be significantly different between cohorts.

Discussion

Shoulder arthroscopy in the year prior to shoulder arthroplasty is associated with an increased risk of complications, including revision and aseptic loosening.

Introduction

Total shoulder arthroplasty (TSA) and reverse TSA (rTSA) are efficacious treatments for patients with end-stage glenohumeral osteoarthritis (GHOA). The prevalence of GHOA has been estimated at 4–26%,^1,2 and the number of TSA and rTSA procedures performed each year in the United States has been increasing.^3,4 Risks associated with total joint arthroplasty include infection, instability, fracture, implant failure and aseptic loosening, and when treatment failure does occur, revision TSA is often required.

One recently published study observed rates of revision TSA to be 6.7% following TSA and 15% following rTSA at an average follow up of 8.7 years. ⁵ Revision arthroplasty is a complex procedure relative to primary arthroplasty, ⁶ and it is important to identify pre-operative patient parameters that may be associated with an increased risk of revision surgery. ⁷ It is currently unknown if prior shoulder arthroscopy (SA) is associated with an increased risk of adverse outcomes or revision arthroplasty.

The clinical implications of arthroscopy prior to arthroplasty have recently been investigated in the shoulder as well as other joints.^8,9,10,11 Studies suggest, SA performed within 2 years before shoulder arthroplasty is associated with a higher rate of prosthetic joint infection both within 90 days ¹² and the first year following shoulder arthroplasty. ¹¹ Although SA may improve short-term and medium-term patient reported outcomes and delay arthroplasty in some patients,^13,14 the relationship between prior SA and outcomes following shoulder arthroplasty merits investigation based on this early clinical data and the significant costs and morbidities associated with revision arthroplasty.

The purpose of this study was to determine if an association exists between SA in the year prior to TSA/rTSA and 2-year postoperative outcomes following arthroplasty for GHOA. While it is known that SA prior to shoulder arthroplasty is associated with greater risk of prosthetic joint infection, it has not been studied if there is any association of postoperative complications, such as aseptic loosening and periprosthetic fracture, following revision surgery in addition to periprosthetic infections. We hypothesized that arthroscopy prior to TSA/rTSA will be found to be associated with an increased rate of revision surgery and selected postoperative complications.

Materials and methods

Data source

A retrospective cohort study was conducted using data gathered from the Humana national insurance claims database using the PearlDiver Patient Records Database (PearlDiver Technologies, Fort Wayne, IN, USA, www.pearldiverinc.com). This database contains patient records, billing records, and information regarding prescribed medications for over 120 million patients. The database includes all-payer (commercial insurance, Medicare, Medicaid, self-pay, government, etc.) claims information from across the United States with patients from all age ranges and socioeconomic status. Additionally, the database uniquely tracks patients within the dataset, permitting quantification of any follow-up period limited only to the 10-year period provided. Queries for the database can be made using International Classification of Disease (ICD) codes, Current Procedural Terminology (CPT) codes, demographic, geographical, and other parameters.

Patient selection

Patients with a primary diagnosis of GHOA, as determined through the use of the ICD codes (Supplemantal Appendix A), who underwent primary TSA or rTSA between 2005 and 2017, were identified using CPT codes (Supplemental Appendix B). Patients undergoing hemiarthroplasty or revisions were not included in this study. Anatomic versus reverse shoulder arthroplasty procedures could not be differentiated by the CPT code used to identify primary TSA. Individuals undergoing TSA/rTSA for fracture, trauma, or neoplasms were excluded and only patients with a primary diagnosis of GHOA were included. Patients who underwent prior SA were identified using CPT codes listed in Supplemental Appendix B.

Postoperative outcomes

Differences in revision rates, aseptic loosening (mechanical loosening of prosthetic joint in the absence of infection), periprosthetic fracture, and infection between the patients who underwent SA before TSA or rTSA were compared with those who did not undergo prior SA. Outcomes were assessed at 2-year follow up after the index TSA or rTSA procedure.

Propensity matching

Propensity-matched cohorts of patients that had an SA procedure before TSA or rTSA and those who did not have an SA before TSA or rTSA were made to control for measured covariates and limit confounding. The propensity score was defined as the conditional probability of having undergone a prior SA procedure based on age, Charlson comorbidity index, obesity status, and smoking status. Matching was conducted using a 1:2 nearest neighbor matching ratio by univariate analysis. This was done using demographics collected at the time of TSA among the SA and non-SA cohorts. Propensity score matching was conducted using R software provided by PearlDiver.

Statistical analysis

Patient demographics, complications, and revision surgery were collected and subsequently compared between propensity-matched cohorts with univariate analysis using R software provided by PearlDiver. Univariate analysis and statistical comparisons were then performed using the group of interest against the control group. A p-value of <0.05 was used as the cutoff for significance for univariate analysis.

Results

Study cohort

In total, 788 TSA patients with prior SA and 20,171 TSA patients without prior SA were identified in the time period from 2005 to 2017.

The prematch cohort for those who did not undergo prior arthroscopy had several statistically significant differences as compared to the arthroscopy cohort. The prematch control cohort had a higher proportion of females (60.1% vs. 55.3%, p < 0.001), a different distribution of number of documented comorbidities (p < 0.001), was generally older (p < 0.001), and had lower rates of obesity (12.5% vs. 47.6%, p < 0.001) and smoking (6.1% vs. 21.5%, p < 0.001) (Table 1).

Table 1.

Comparison of prior arthroscopy cohort with prematch, or arthroplasty before adjusting the sample for specified demographic characteristics.

	Arthroscopy (n  788)	Percentage	Control–prematch (n  =  20,171)	Percentage	p-Value
Gender					<0.001
Male	352	44.7	8049	39.9
Female	436	55.3	12122	60.1
CCI					<0.001
0	197	25.0	5542	27.5
1	180	22.8	4271	21.2
2	129	16.4	3198	15.9
3	94	11.9	2321	11.5
≥3	188	23.9	4839	24.0
Age					<0.001
<60	86	10.9	847	4.2
60–69	276	35.0	5975	29.6
70–79	344	43.7	9779	48.5
≥80	82	10.4	3570	17.7
Obesity (BMI ≥ 30)	375	47.6	2523	12.5	<0.001
Smoking	169	21.5	1226	6.1	<0.001

CCI: Charlson comorbidity index; BMI: body mass index.

Following successful propensity matching between the cohorts, there were no observed differences in matched parameters for demographics or comorbidities between patients who had SA before TSA or rTSA and those who did not have prior SA (p > 0.05). Following matching, there were 788 patients in each cohort, those with and without SA before TSA or rTSA (Table 2).

Table 2.

Comparison of prior arthroscopy cohort with postmatch cohort following propensity matching for gender, CCI, age, obesity, and smoking.

	Arthroscopy (n  =  788)	Percentage	Control–postmatch (n  =  788)	Percentage
Gender
Male	352	44.7	352	44.7
Female	436	55.3	436	55.3
CCI
0	197	25.0	197	25.0
1	180	22.8	180	22.8
2	129	16.4	129	16.4
3	94	11.9	94	11.9
≥3	188	23.9	188	23.9
Age
<60	86	10.9	86	10.9
60–69	276	35.0	276	35.0
70–79	344	43.7	344	43.7
≥80	82	10.4	82	10.4
Obesity (BMI ≥ 30)	375	47.6	375	47.6
Smoking	169	21.5	169	21.5

CCI: Charlson comorbidity index; BMI: body mass index.

Postoperative complications

Revision surgery (3.4% vs. 1.4%, p  =  0.001) and aseptic loosening (2.2% vs. 0.8% p  =  0.021) in 2 years following arthroplasty were significantly more common in the arthroscopy cohort (Table 3). Periprosthetic fracture (1.1% vs. 0.9%, p  =  0.615) and periprosthetic infection (2.5% vs. 1.4%, p  =  0.102) were not significantly different between the two cohorts.

Table 3.

Incidence of complications within 2 years of total shoulder arthroplasty.

	Arthroscopy (n  =  788)	Percentage	Control–postmatch (n  =  788)	Percentage	p-Value
Revision	27	3.4	11	1.4	0.001
Aseptic loosening	17	2.2	6	0.8	0.021
Periprosthetic fracture	9	1.1	7	0.9	0.615
Periprosthetic joint Infection	20	2.5	11	1.4	0.102

Discussion

The present study, which used a cohort of 1576 propensity-matched patients (with 788 patients in the control cohort), demonstrated that those with a prior history of SA are at increased risk of complications including aseptic loosening and revision surgery within 2 years of TSA compared with patients with no history of SA. There were no significant differences in the rate of periprosthetic fracture or prosthetic infection between groups.

Joint preservation procedures for arthritis of the shoulder are increasingly popular. ¹⁴ As surgeons treating GHOA encounter an ageing population, the decision to proceed with arthroscopy or arthroplasty is of increasing importance. Palliative arthroscopy for GHOA, including arthroscopic debridement and capsular release, have been recommended for younger patients ¹⁵ with less severe arthritis. ¹⁶ In a Markov decision model, Spiegl et al., ¹⁷ demonstrated that the benefits of SA are most clearly defined for patients under the age of 47, and other authors have advocated for an age cutoff of 50 years for joint preservation techniques. ¹⁴ In our study, individuals 70 years of age and older constituted 54.1% of the arthroscopy cohort as compared to 66.2% of those in the prematch control (nonarthroscopy) group. These data suggest that while surgeons are using arthroscopy less frequently in older patients, a significant number of individuals over the proposed cutoff age undergo SA in the year prior to primary arthroplasty.

The rates of revision surgery (3.4% vs. 1.4%, p < 0.001) and aseptic loosening (2.2% vs. 0.8%, p  =  0.021) observed in our study were significantly higher in the group that had undergone prior SA. The rate of periprosthetic infection (2.5% vs. 1.4%, p  =  0.102) in those with prior SA was not statistically significant. Regardless of cohort, the rate of revision surgery within 2 years of arthroplasty was lower than that which has been recently reported in the literature. Matsen et al. reported a 2-year revision rate of 5.6% (19 of 337) in a sample that included hemiarthroplasty in addition to TSA/rTSA and included indications beyond GHOA for primary arthroplasty. ¹⁸

In a case-control study by Kiet et al., ¹⁹ comparing outcomes following TSA and rTSA, 2-year revision rates were 11% and 9%, respectively. One reason for the lower revision rate observed in our study is the exclusion of fracture as an indication for surgery. Although excluding patients with proximal humerus fractures may lead to the exclusion of some rTSA patients, it is necessary in this study to obtain a more homogenous patient cohort. Additionally, only 14% of patients undergoing rTSA do so for fractures, while the majority of rTSA cases are for cuff tear arthropathy. ³

The pathophysiological explanation of increased risk for postoperative complications among shoulder arthroplasty patients with prior SA is not clear. While prior studies have shown that patients undergoing arthroplasty following shoulder surgery have outcomes inferior to those without prior surgery, a clear causation has yet to be identified.^20,21 In a study of Medicare and private insurance patients, SA performed within 2 years before shoulder arthroplasty was associated with a higher rate of prosthetic joint infection in the first year following shoulder arthroplasty. ¹¹ A similar study also demonstrated these findings in the 90 days postshoulder arthroplasty. ¹² Analogous observations have also been made in other orthopedic subspecialties indicating that this phenomenon may not be limited to shoulder procedures only. For instance, investigators have examined the role of knee arthroscopy prior to total knee arthroplasty (TKA) and found an association between prior arthroscopy and an increased risk of complication following TKA.^8,9,10 While a relationship was observed between prior arthroscopy and increased risk of complications, the mechanism has not been elucidated.

An additional finding in this study is the percentage of patients who are obese (body mass index  > 30) and those who smoke among the matched cohorts. While the percentage of obese patients in the matched cohorts (47.6%) is similar to the age-adjusted national prevalence of obesity among adults in the United States (41.9%) as reported by the Center for Disease Control and Prevention, ²² the percentage of patients who smoke (21.5%) is higher than the adults who currently smoke in the United States (12.5%). ²³ These findings are important to note as both smoking and obesity have been associated with rotator cuff pathology.^24–27 Further, obesity and smoking have both independently been shown to affect the healing of rotator cuff repair leading to failure.^28,29 Additional studies are necessary to investigate the relationship between smoking and obesity among patients undergoing SA prior to shoulder arthroplasty and any associated risk of complications.

This study was subject to several limitations. First, given that the data is from an insurance claims database there is the possibility of data entry errors, miscoding, or missed diagnosis. While the sample size of our study was relatively large, this study may have been underpowered to detect a difference in periprosthetic infection. While there was a trend toward increased risk of periprosthetic infection, this rare outcome may require a larger sample size to confidently draw conclusions regarding any additional risk that prior SA may impose. Specific timing of SA prior to arthroplasty was not analyzed in this study, and the use of this data may represent an opportunity to better characterize the temporal relationship between SA and the risk of poor outcomes following arthroplasty. Further limitations of our data were that it did not characterize the exact indication for SA and it did not specify which patients underwent TSA and rTSA either. There could be a disproportionate number of TSA or rTSA cases in the matched cohort. Due to the limitations in coding, we are unable to determine the exact number of TSA or rTSA cases nor were we able to match patients by type of procedure. While previous work has shown a higher revision rate in rTSA patients the most common causes for revision were glenoid failure and instability. ⁵ A history of SA prior to arthroplasty was not investigated. Therefore, while our work is limited by a lack of precise procedural classification, we believe it adds to the existing literature on shoulder arthroplasty by identifying a potential risk factor for revision surgeries among TSA patients. Further research must be conducted to elucidate the impact of SA on revision surgery by the type of shoulder arthroplasty performed. Additionally, due to limitations in the ICD coding, we are unable to clarify if the aseptic loosening of the joint was due to the loosening of the humeral or glenoid component. Finally, there is the possibility of selection bias in this study, for instance, if a patient underwent SA and avoided TSA/rTSA as a result. The degree of GHOA was also not analyzed in this study, which may represent an important factor regarding which patients would be more likely to require arthroplasty in the future. The strengths of this study include the large sample size and the 2-year duration of follow up, although complications that arise beyond 2 years are not accounted for. Further studies are needed to validate a causal relationship between SA and increased revision rates and to better evaluate pathoanatomical features that may contribute to this increased risk.

Conclusions

SA in the year prior to shoulder arthroplasty is associated with an increased risk of complications, including revision surgery and aseptic loosening. Patients considering arthroscopy should be counseled about the association of an increased risk of loosening and possible future revision surgery when deciding on how best to manage symptomatic glenohumeral arthritis.