Previous rotator cuff repair increases the risk of revision surgery for periprosthetic joint infection after reverse shoulder arthroplasty

Background: Previous studies have indicated an increased risk of periprosthetic joint infection (PJI) in patients treated with reverse shoulder arthroplasty (RSA) compared with patients treated with anatomic total shoulder arthroplasty. The reason for this is unclear but may be related to a high prevalence of previous rotator cuff repair in patients who are treated with RSA. The purpose of this study was to determine previous non-arthroplasty surgery as a risk factor for revision owing to PJI after RSA for cuff tear arthropathy, massive irreparable rotator cuff tears, or osteoarthritis. Methods: Data were retrieved from the Danish Shoulder Arthroplasty Registry and medical records. We included 2217 patients who underwent RSA forcufftear arthropathy,massiveirreparable rotatorcufftears, orosteoarthritisbetween 2006 and2019.PJIwasdeﬁnedas (cid:1) 3 of 5 tissue samples positive for the same bacteria or as deﬁnite or probable PJI evaluated based on criteria from the International Consensus Meeting. The Kaplan-Meier method was used to illustrate the unadjusted 14-year cumulative rates of revision. ACox regression model was used to report hazard for revision owing to PJI. Results were adjusted for previous non-arthroplasty surgery, sex, diagnosis, and age. Results: Revision was performed in 88 shoulders (4.0%), of which 40 (1.8%) underwent revision owing to PJI. There were 272 patients (12.3%) who underwent previous rotator cuff repair, of whom 11 (4.0%) underwent revision owing to PJI. The 14-year cumulative rate of revision owing to PJI was 14.1% for patients with previous rotator cuff repair and 2.7% for patients without previous surgery. The adjusted hazard ratio for revision owing to PJI for patients with previous rotator cuff repair was 2.2 (95% conﬁdence interval, 1.04-4.60) compared with patients without previous surgery. Conclusion: There is an increased risk of revision owing to PJI after RSA for patients with previous rotator cuff repair. We recommend that patients with previous rotator cuff repair be regarded as high-risk patients when considering RSA.

Reverse shoulder arthroplasty (RSA), since it was introduced in the 1980s, 5 has shown good and reliable outcomes for patients with cuff tear arthropathy (CTA). 4Subsequently, the use of RSA has increased during the past few decades, 1,8 and the indications have expanded to include massive irreparable rotator cuff tears, proximal humeral fractures, glenohumeral osteoarthritis (OA) with severe glenoid bone loss, and failed anatomic arthroplasty. 13SA is associated with a higher revision rate owing to periprosthetic joint infection (PJI) compared with anatomic total shoulder arthroplasty (TSA). 4,9PJI is the most common reason for revision of RSA, and most revisions are performed shortly after the primary RSA. 8A previous study from the Nordic Arthroplasty Registry Association (NARA) reported a 10-year cumulative rate of revision owing to PJI of 1.4% for all arthroplasties but 3.1% for RSAs.When adjusted for age, sex, and indication, the risk of revision owing to PJI after RSA was 2.4 times higher than after anatomic TSA. 9 The higher risk of revision owing to PJI after RSA may be related to a high proportion of patients who previously underwent non-arthroplasty surgery.Werthel et al 16 reported a hazard ratio (HR) of 1.8 for revision owing to PJI after shoulder arthroplasty in patients who had undergone any type of previous non-arthroplasty surgery.For RSA in particular, the risk may be substantial because of a high prevalence of previous rotator cuff repair.Information about previous rotator cuff repair as a risk factor for revision because of PJI after RSA is, however, limited.Because revision owing to infection is a rare event, a large cohort of patients is needed to perform an adequate risk analysis.Such cohorts may only be achieved using registry data.
The purpose of this study was to determine previous non-arthroplasty surgery as a risk factor for revision owing to PJI after RSA for CTA, massive irreparable rotator cuff tears, or OA.The hypothesis was that patients with rotator cuff repair prior to RSA would have a higher risk of revision owing to PJI compared with patients without any previous surgery.Thus, the insertion of suture anchors and multiple non-dissolvable sutures may be associated with low-grade infection, and if an infection after rotator cuff repair is not recognized and addressed, there will be a high risk of revision owing to PJI after the RSA.In contrast, we did not expect to find an increased risk of revision owing to PJI in patients with previous subacromial decompression and/or acromioclavicular (AC) joint resection.

Data source
In Denmark, health care is governmentfunded and available to all persons without cost.All Danish citizens are registered by a unique civil registration system (CPR) number given at birth or immigration.The CPR number is used by hospitals to reimburse expenses for hospital visits, which is done by reporting the type of health care provided and the CPR number to the Danish National Patient Register (DNPR).
This was a retrospective study, in which data were obtained from the Danish Shoulder Arthroplasty Registry (DSR).The DSR was established in 2004.Reporting has been mandatory for all public and private hospitals since 2006.The completeness of reporting in 2004 and 2005 was low, and these 2 years are regarded as a trial period, which is why the inclusion period started in 2006.The surgeon reports data related to the patient and the surgical procedure at the time of the operation. 12The completeness of reporting is calculated by comparing data from the DSR with data from the DNPR.The rate of completeness of reporting has been shown to be 94%. 11To maintain high completeness of reporting, data from the DSR are compared with data from the DNPR every 3 months.If data do not correlate, a reminder of registration will be sent to the hospitals. 11The CPR number is used by the DSR to accurately link index and revision arthroplasties and to obtain information from the DNPR about death or emigration.
For this study, previous surgery was categorized into 3 groups: (1) rotator cuff repair; (2) other types of previous surgery (biceps tenotomy, biceps tenodesis, surgery for instability, synovectomy, subacromial decompression, AC joint resection, arthroscopy, or unknown); and (3) no previous surgery (Fig. 1).If >1 type of previous surgery was reported, the aforementioned categorization was used as a hierarchy, so only 1 type of previous surgery was registered.The DSR does not collect detailed information about rotator cuff surgery (open or arthroscopic procedure, number of tendons involved, time of surgery, and so on).
In case of revision, the surgeon reports surgical data including the reason for revision at the time of surgery.Thus, reporting of PJI to the DSR is based on preoperative data and perioperative findings and not on positive culture samples or composite criteria such as the criteria from the International Consensus Meeting (ICM) 3 (Fig. 2).

Study population
A total of 14,073 primary arthroplasties were reported to the DSR from 2006-2019.Patients who were treated with other arthroplasty types (n ¼ 10,510) and patients who were treated with RSA because of rheumatic arthritis (n ¼ 491), avascular necrosis (n ¼ 91), fracture sequelae (n ¼ 108), acute fracture (n ¼ 312), or other reasons (n ¼ 114), as well as those who had a missing diagnosis (n ¼ 15), were excluded.Patients who were recorded to have undergone surgery for infection prior to the primary arthroplasty (n ¼ 6) were excluded.Bilateral RSA was noted in 209 patients.To avoid dependency issues in the analyses, the second arthroplasty was excluded (Fig. 1).Thus, 2217 patients with a primary RSA for CTA, massive irreparable rotator cuff tears, or OA were included in the study (Table I).

Outcome
The outcome was first-time revision defined as the removal, addition, or exchange of any component.For patients with available medical records, PJI was defined as !3 of 5 soft tissue samples positive for the same bacteria or as definite or probable PJI according to the ICM criteria 3 (Fig. 2).The ICM criteria were developed to establish a uniform definition of PJI for use in research and clinical decision making.On the basis of macroscopic observations and laboratory results, each case can be categorized as definite, probable, possible, or unlikely PJI or as no PJI (Fig. 2).Patient records were available for 69 patients (78%).For the remaining 19 patients, the diagnosis of PJI relied on the indication for revision reported to the DSR by the surgeon at the time of the revision procedure.
We estimated the validity of the reporting of PJI as the indication for revision reported to the DSR by comparing information from the DSR with information retrieved from medical records for the 69 revision arthroplasties for which medical records were available.With this use of medical records as the gold standard, the positive predictive value (PPV) and negative predictive value (NPV) of PJI reported to the DSR were 93% and 83%, respectively (Table II).

Statistics
Descriptive statistics were used to report demographic data including sex, age, previous surgery, diagnosis, year of surgery, follow-up time, RSA brand, time to revision, and indication for revision.To adjust for differences in follow-up time, timedependent analyses were used.The Kaplan-Meier method was used to report unadjusted 14-year cumulative rates of revision overall and rates of revision owing to PJI.A Cox regression model was used to report the HR for revision owing to PJI.Sex, age, diagnosis, and type of previous non-arthroplasty surgery were included in the multivariate model.The end of follow-up was defined as first-time revision, death, emigration, or December 31, 2019, whichever came first.The proportional hazard assumption was found to be appropriate when assessed graphically for each risk factor.P < .05 was considered statistically significant, and P values were 2-tailed.SPSS statistical software (version 25.0; IBM, Armonk, NY, USA) was used.The PPV was calculated as the number of registrations in the DSR with their key variable confirmed in the medical records divided by the number of positive registrations in that key variable in the DSR: PPV ¼ (Confirmed positive key variable registrations in DSR / All positive key variable registrations in DSR) Â 100.
There were 484 patients (21.8%) who underwent previous non-arthroplasty surgery, of whom 272 (12.3%) underwent rotator cuff repair (Table I).For 6 patients, there was no information on previous surgery; none of these patients underwent revision surgery.The mean time from index arthroplasty to revision was 44.8 months (SD, 33.9 months) for the entire population, 44.2 months (SD, 33.1 months) for patients with no previous surgery, 52.8 months (SD, 37.5 months) for patients with other types of previous surgery, and 42.4 months (SD, 34.5 months) for patients with previous rotator cuff repair.
The unadjusted HR for revision owing to PJI for patients with previous rotator cuff repair was 2.8 (95% CI, 1.40-5.77;P ¼ .004)compared with patients without previous surgery.When age, sex, and diagnosis were included in the multivariate model, the HR for risk of revision owing to PJI was 2.2 (95% CI, 1.04-4.60;P ¼ .04)(Table VI).Additionally, we found that male sex was associated with an increased risk of revision owing to PJI (HR, 3.2; 95% CI, 1.65-6.28,P ¼ .001)(Table VI).

Discussion
The 14-year cumulative rate of revision owing to PJI was 14.1% for patients with previous rotator cuff repair and 2.7% for patients without previous surgery.In the multivariate Cox regression model, patients with previous rotator cuff repair were found to have an increased risk of revision owing to PJI with an HR of 2.2 compared with patients without previous surgery.Previous surgery with  We were able to obtain information from medical records for 69 of 88 patients.The PPV, NPV, sensitivity, and specificity are presented for reporting of PJI to the DSR for these 69 patients.
subacromial decompression and/or AC joint resection was not associated with an increased risk of revision owing to PJI (HR, 1.1, 95% CI, 0.40-3.29).
The reason for the increased risk of revision owing to PJI cannot be deducted from this observational study, and there is no adequate explanation in the literature.Several factors can, in theory, increase the risk of PJI.The insertion of suture anchors and multiple non-dissolvable sutures may be associated with low-grade infection, and if an infection after rotator cuff repair is not recognized and addressed, there will be a high risk of revision owing to PJI after RSA.This would explain why there is no increased risk for patients with previous subacromial decompression and/or AC joint resection.Furthermore, the rotator cuff constitutes an extra layer between the arthroplasty and the skin.If the rotator cuff is not intact, the same barrier is not present to prevent PJI.Finally, in contrast to subacromial decompression, rotator cuff repair is not always performed arthroscopically.Thus, the use of an open approach may increase the risk of infection compared with arthroscopic procedures.
To our knowledge, this is the first study to focus on revision owing to PJI after RSA and previous rotator cuff repair.Fourteen-year cumulative rate of revision for any reason for all patients treated with reverse shoulder arthroplasty for cuff tear arthropathy or osteoarthritis with 95% confidence intervals (dashed lines).The cumulative rate was 2.6% at 1 year, 4.1% at 5 years, 7.9% at 10 years, and 7.9% at 14 years.

Previous studies have reported the risk of PJI after different
Figure 4 Fourteen-year cumulative rate of revision owing to periprosthetic joint infection (PJI) for all patients treated with reverse shoulder arthroplasty for cuff tear arthropathy or osteoarthritis with 95% confidence intervals (dashed lines).The cumulative rate was 1.1% at 1 year, 1.8% at 5 years, 4.0% at 10 years, and 4.0% at 14 years.
reduction and internal fixation make it difficult to compare their results with ours.We found male sex to be a risk factor for revision owing to PJI, which supports results reported in previous studies. 9,15he reason for this may be related to a higher prevalence of C acnes around the shoulders in male individuals because of a higher density of hair follicles, where the presence of C acnes is more prevalent. 10We found that C acnes was the most common reason for PJI, which supports the results of previous studies. 14,15Up to 56% of PJIs in shoulders have been reported to be caused by C acnes. 7sing information from medical records as the gold standard, we found a PPV of 93% and NPV of 83% for the reporting of revision owing to PJI to the DSR.We were unable to retrieve information from the medical records about the reason for revision in 19 patients (22%).For these patients, we relied on data reported to the DSR by the surgeon at the time of the revision procedure.On the basis of the PPV of 93% and the NPV of 83%, we find it unlikely that the results are significantly influenced by the lack of information from the medical records.
The major strengths of this study are the large population and the high grade of completeness, making it possible to focus on a specific type of previous surgery and a specific arthroplasty type.Merging data for different types of arthroplasties and different types of previous surgery, as done in some previous studies, makes it difficult to give robust estimates when patients are being informed about their individual risk profile prior to a shoulder arthroplasty procedure.Another advantage of this study is that in the majority of cases, the PJI was reported based on the number of positive tissue samples and the ICM criteria. 3It is, however, important to bear in mind that not all information related to the minor criteria was available mainly because preoperative blood samples and joint aspiration were not routinely used during the study period.Finally, the study was based on a complete nationwide cohort and a long period of inclusion, which leave the study with high external validity.
The limitations of our study include the inherent limitations identified for observational studies.Our results may also have been biased by inaccurate reporting of previous surgery to the DSR.Data on comorbidities were not available, and any systematic differences in diabetes, inflammatory diseases, medication, and so on may have biased the results.
We were unable to distinguish between open and arthroscopic rotator cuff repair.Both techniques are used in Denmark.This is important because some previous studies have shown an increased risk of infection after open rotator cuff repair in comparison with arthroscopic repair. 2,6espite the use of data from a nationwide registry, the Figure 5 Fourteen-year cumulative rate of revision with 95% confidence interval (dashed lines) for different types of previous nonarthroplasty surgery.The cumulative rates for rotator cuff repair (RCR) were 4.0% at 1 year, 7.6% at 5 years, 21.4% at 10 years, and 21.4% at 14 years.The cumulative rates for others were 3.4% at 1 year, 6.8% at 5 years, 6.8% at 10 years, and 6.8% at 14 years.The cumulative rates for none were 2.1% at 1 year, 3.2% at 5 years, 5.9% at 10 years, and 5.9% at 14 years.number of patients with previous rotator cuff repair and revision of RSA owing to PJI was too small to analyze the risk of revision owing to PJI at different time points.It can be hypothesized that the increased risk of revision owing to PJI for patients with previous rotator cuff is primarily related to low-grade infection that is only recognized years after the RSA.Thus, the risk of revision owing to PJI for patients with previous rotator cuff repair and patients with no previous surgery may be similar in the short term.A future study with a larger data set could explore early vs. late onset of revision owing to PJI.
Because of the long-lasting sick leave, as well as multiple hospitalizations and surgical procedures, PJIs after shoulder arthroplasty entail economic consequences for society and major consequences for the individual patient, who is often left with a poor-functioning shoulder.Given Figure 6 Fourteen-year cumulative rate of revision owing to periprosthetic joint infection (PJI) with 95% confidence intervals (dashed lines) for different types of previous non-arthroplasty surgery.The cumulative rates for rotator cuff repair (RCR) were 1.6% at 1 year, 4.2% at 5 years, 14.1% at 10 years, and 14.1% at 14 years.The cumulative rates for others were 1.4% at 1 year, 2.1% at 5 years, 2.1% at 10 years, and 2.1% at 14 years.The cumulative rates for none were 0.9% at 1 year, 1.4% at 5 years, 2.7% at 10 years, and 2.7% at 14 years.the serious consequences, there should be awareness of the increased risk of revision of RSA owing to PJI after previous rotator cuff repair.

Conclusion
We found a significantly increased risk of PJI after RSA in patients who had undergone previous rotator cuff repair compared with patients with no previous surgery.On the basis of the results from this study and the existing literature, we recommend that patients with previous rotator cuff repair be regarded as high-risk patients when considering RSA.This observational study with no intervention does not allow recommendations regarding the prevention of PJI.However, surgeons could consider the perioperative collection of tissue samples and administration of oral antibiotics until the results are available, irrigation with saline water mixed with antibiotics, use of antibiotic-collagen sponges, or similar approaches.

Disclaimers:
Funding: No funding was disclosed by the authors.Conflicts of interest: Dr Steen L. Jensen reports payment for expert testimony from the European Commission Expert Panel for Medical Device Regulation and is a fiduciary for the Danish Shoulder Arthroplasty Registry.Bo S. Olsen reports grants for other projects from the manufacturers DePuy Synthes and Zimmer Biomet, is a member of the panel of the Danish Shoulder Arthroplasty Registry, and is past-president of the Danish Orthopaedic Society.
The other authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.

Figure 1
Figure 1 CONSORT (Consolidated Standards of Reporting Trials) flow diagram of inclusion of patients.DSR, Danish Shoulder Arthroplasty Registry; RSA, reverse shoulder arthroplasty; PJI, periprosthetic joint infection.
8 for revision owing to PJI after shoulder arthroplasty for patients who had undergone any type of previous non-arthroplasty surgery compared with patients without previous surgery.Most of the patients were treated with anatomic TSA, and previous surgery was defined as cuff repair, open reduction and internal fixation, d ebridement, acromioplasty, capsular repair, and others.The inclusion of different types of arthroplasties and especially the inclusion of fracture patients with previous open

Figure 3
Figure3Fourteen-year cumulative rate of revision for any reason for all patients treated with reverse shoulder arthroplasty for cuff tear arthropathy or osteoarthritis with 95% confidence intervals (dashed lines).The cumulative rate was 2.6% at 1 year, 4.1% at 5 years, 7.9% at 10 years, and 7.9% at 14 years.

Table V
Microbiology of revisions owing to infections PJI, periprosthetic joint infection.There were 40 patients with PJI but only 33 with cultures.There were cases with different types of bacteria; hence, the total is >40.

Table III
Reasons for revision reported in patient medical journals 16pes of arthroplasties and various types of previous surgery.Werthel et al16reported an HR of 1.