Anatomic versus reverse shoulder arthroplasty for patients with cuff-intact arthritis: similar outcomes, different complications

Loose glenoid in aTSA

Baseplate failure in RSA

Dislocation in RSA

Acromial fracture in RSA

Summary

Conventionally, anatomic total shoulder arthroplasty (aTSA) has been used for patients with arthritis and a functioning rotator cuff. The reverse total shoulder arthroplasty (RSA) was developed for shoulders whose rotator cuff had failed. RSA now accounts for about seven of every ten shoulder replacements in the United States. More than half of them are used for patients with cuff-intact arthritic shoulders. 

The question surgeons and patients would like answered is: how do the clinical outcomes of these two procedures compare for shoulders with cuff-intact arthritis?

The short answer: the existing evidence does not show major differences in clinical outcomes between the two procedures. However, the published literature does show differences in the types, seriousness and frequency of complications. These differences are not reflected in revision rates because of the relatively poor outcomes for revision of a failed RSA, which contrasts with the better outcomes for revision of a failed aTSA.

Considering only studies limited to the treatment of cuff-intact arthritis, the evidence suggests: (1) on numeric scores (SANE, ASES, VAS), the two are about the same; (2) on motion and certain specific  tasks, the anatomic often does better; (3) on durability, controlled studies are in the works and the results are as yet unknown.

Numeric scores look alike. Matched comparisons from single practices, a national English registry of more than 7,000 operations, and several pooled reviews all found that shoulder scores after the two operations differed by less than the amount that patients would notice. 

The best results may differ. One study looked only at patients who rated their shoulder 95 or more out of 100. Among patients with osteoarthritis and an intact cuff, 41% of those with an anatomic and 35% of those with a reverse reached that level. Among those who did, patients receiving aTSA could more often lift 10 pounds overhead without difficulty (73% vs 55%) and carry 20 pounds at the side (86% vs 70%). 

Time may change the comparison. The longest comparison came from one surgeon, in shoulders with advanced wear of the socket that were followed for about eight years. The anatomic was better at two years. By eight years the reverse had slightly better scores, more satisfied patients (95% vs 82%) and fewer revisions (2% vs 8%). 

The complications differ, and revision rates do not show it. The complications differ in type: after an aTSA, mainly subscapularis failure early and glenoid loosening and cuff failure later; after an RSA, mainly acromial and scapular spine fractures, instability, and baseplate failure. These complications differ in seriousness: a failed anatomic can usually be successfully converted to a reverse, while a failed reverse is harder to salvage and some of its complications are not revised at all. In a national registry, fewer than one in 20 patients with an unsatisfactory result after an RSA was revised, about half the rate for aTSA. 

Can patients forget their shoulder? A new questionnaire, the Forgotten Joint Score, has now been applied to shoulder replacement, but no one has yet used it to compare aTSA and RSA.

What we know and what we don't. Every comparison of anatomic and reverse so far is observational. Surgeons chose the operation based on their experience, the characteristics of the patient and the characteristics of the shoulder. These factors strongly influence the outcome—perhaps more than which type of arthroplasty is used. 

Much needed randomized clinical trials are underway in the United Kingdom and Switzerland. Until these studies are complete, the relative value to the patient of aTSA and RSA will remain uncertain. 

Now for the details

Clinical Outcomes Are Not Significantly Different Between 

aTSA and RSA for Cuff-Intact Arthritis

The reverse is now the most common shoulder arthroplasty. In a national cohort it rose from about 55% to about 70% of shoulder arthroplasties between 2016 and 2020, while the anatomic fell from about 37% to 27%. Of 95,808 reverses, 53% (51,150) were done for glenohumeral arthritis with an intact cuff (Mayfield et al. [1]).

This post asks what recent clinical studies show about function after an anatomic versus a reverse for the arthritic shoulder with an intact cuff, including whether patients come to forget the shoulder. It also covers what happens when the initial arthroplasty fails.

Only clinical studies of primary osteoarthritis with an intact rotator cuff are considered: studies of patients with outcomes reported by patients or measured on them. Cadaver, finite element and computer models were not included.

Anatomic or reverse for the arthritic shoulder with an intact cuff

In studies of the cuff-intact arthritic shoulder, no difference in patient-reported scores exceeded the minimal clinically important difference (MCID). Where motion differed, it favored the anatomic, but in a matched single-center series the difference was small.

Study	Design (all cuff-intact OA)	What it showed
Leinweber et al., JSES 2026[2]	One surgeon; propensity-matched, 61 pairs; mean follow-up about 2 and 5 years	Over 96% in both groups reached the ASES MCID at both times. Substantial clinical benefit: 95.1% aTSA vs 80.3% rTSA early (P = .027), 91.8% vs 78.7% at midterm (P = .074). Internal and external rotation better early after aTSA, not at midterm.
Baek et al., JSEA 2026 [3]	One center; propensity-matched, 44 pairs, also matched for cuff fatty infiltration; minimum 3 years	No difference in ASES, Constant, SANE, pain, or elevation, external or internal rotation. Same result at age 70 or younger and over 70. Complications 18.2% after rTSA (4 acromial or spine fractures, 3 notching) vs 4.5% after aTSA (P = .089); one revision in each group. Note the difference in the ratio of complications to revisions for each procedure.
Cuff et al., JSES 2023 [4]	One surgeon; B2, B3 or C glenoids; 101 aTSA with high-side reaming vs 93 rTSA; minimum 7 years; not matched	At 2 years the anatomic was better: ASES 85 vs 80, elevation 162° vs 151°. At a mean of about 8 years the reverse was better: ASES 80 vs 77, satisfaction 95% vs 82%, revision 2% vs 8% of the starting cohorts. Note again that revision rate is relatively insensitive to complication rate.
Valsamis et al., BMJ 2024[5]	National Joint Registry for England; propensity-matched, 7,124 procedures, age 60 or older	Fewer reoperations at 12 months after rTSA (absolute difference 0.51%). No clinically important difference in long-term revision-free survival. Similar change in Oxford Shoulder Score at 6 months.
Shah et al., Bone Jt Open 2026 [6]	Systematic review, 18 studies	Better motion after aTSA. No significant difference in ASES, pain, complications or revision.
Desouza et al., World J Orthop 2026[7]	Meta-analysis of retrospective studies	rTSA scored higher on UCLA, SST and SPADI. None of the differences exceeded the MCID.
Brazilian Society of Shoulder and Elbow Surgery, 2026[8]	Evidence review, patients 70 or older	Very low-quality evidence that rTSA has a lower reoperation rate.

 

Cuff's series is the only one with more than five years of follow-up. The ASES differences were smaller than the MCID, and patients who were revised or lost were omitted from the final outcome comparison. In the Valsamis study, patient-reported outcomes were measured only at six months.

Every one of these comparisons is observational. The surgeon chose the procedure, and the reasons for that choice also affect the outcome. Propensity matching cannot control for variables the surgeon considered but did not record. 

Randomized trials are underway, including RAPSODI-UK [9] and a Swiss trial [10]. 

Until the results are published, the relative value of aTSA and RSA for the patient with cuff-intact arthritis will remain uncertain.

The forgotten shoulder

Numerical scores hide what matters to the active patient: does the replaced shoulder feel and work like a normal one? A shoulder version of the 12-item Forgotten Joint Score (FJS-12) was tested in 60 patients after arthroplasty for osteoarthritis, 49 anatomic and 11 reverse (Feik et al., JSEA 2026 [11]). Mean scores were 64 ± 33 at six months and 70 ± 29 at one year, out of 100. The minimal detectable change was about 14 points. The wide spread shows that many patients are still aware of the shoulder at one year, even when standard scores sit near their ceiling. At present no published study compares a forgotten-joint score between anatomic and reverse arthroplasty.

Near normal

Beleckas et al. (JSES 2025) [12] studied patients who rated their shoulder at least 95 of 100 on the Single Assessment Numeric Evaluation (SANE), at a minimum of two years. In the subgroup treated for osteoarthritis without a cuff tear, 41% of anatomic and 35% of reverse patients reached that level. Among those who did, the anatomic patients had higher ASES (98 vs 95) and SST (12 vs 10) scores. More of them could lift 10 pounds above the shoulder without difficulty (73% vs 55%), lift 8 pounds to shoulder level (86% vs 66%) and carry 20 pounds at the side (86% vs 70%). External rotation was 60° vs 45°. Pain did not differ. The ASES difference was smaller than the MCID; the differences in specific tasks were ones patients noticed. Note that the anatomic patients were younger (median 70 vs 74 years) and followed longer, and the groups were not matched.

Complications Are Different Between 

aTSA and RSA for Cuff-Intact Arthritis

When neither operation has a complication, the outcomes are similar. The procedures differ in how often complications occur, what kind they are, how serious they are, and which of them can be  successfully revised. 

How often. Short-term complication rates are low after both operations, and the direction of the difference varies between series. In a matched comparison of 370 anatomic and 370 reverse arthroplasties for cuff-intact arthritis in patients without previous surgery (mean age 73, mean follow-up 41 months), complications occurred in 4.9% after the anatomic and 2.2% after the reverse (P = .045), and revisions in 3.2% and 1.4% (P = .086; the study had 37% power to detect a difference in revisions) (Friedman et al., JAAOS 2022 [13]; single manufacturer’s registry, several authors with financial ties to the manufacturer). In Baek’s matched series the direction was the opposite: complications in 18.2% after the reverse and 4.5% after the anatomic [3]. A meta-analysis of eight comparative studies in patients 70 or older (1,716 patients, follow-up under five years) found more complications after the anatomic (risk ratio 1.75, 95% CI 1.04–2.96) and more revisions (risk ratio 3.96, 95% CI 1.54–10.15) (Gupta et al., JSES 2026 [14]). In the two included studies that gave reasons, anatomic revisions were mostly for glenoid loosening (9 of 21) and cuff tears (6 of 21); the three reverse revisions were for infection (2) and humeral component dissociation (1). The authors caution that reverses are often chosen for older patients less likely to undergo revision, and that symptomatic reverses managed without surgery are not counted as failures. 

What kind. After an anatomic, the early complication is mainly subscapularis failure; the late ones are glenoid component loosening and secondary cuff failure. After a reverse, the complications are mainly acromial and scapular spine fractures, instability, scapular notching, and baseplate failure [3]. A loose anatomic glenoid can usually be revised to a reverse; an acromial fracture after a reverse is seldom revised at all. For this reason, complications should be reported by type, together with revisions.

Revised or not. A national registry study of 21,918 shoulder arthroplasties with Oxford Shoulder Scores asked how often patients with an unsatisfactory result (mean score below 29 of 48) were revised. Unsatisfactory scores were more common after a reverse than after a total shoulder (27% vs 15%), yet fewer of these patients were revised (4.9% vs 10.6%) (O’Malley et al., PLoS One 2025 [15]). The authors attribute this to a lower threshold for revising an aTSA, which has a reasonable salvage option in conversion to a reverse. The study included all indications rather than only cuff-intact arthritis, and the reverses included shoulders with cuff tear arthropathy and fractures; the results were similar when only elective indications were analyzed. Only about a third of patients had postoperative scores recorded. Even so, it shows directly that the threshold for revising a reverse is higher: a lower revision rate after a reverse does not mean fewer patients doing poorly. The gap between complications and revisions after a reverse is not new: in a systematic review of 782 reverses, mostly for cuff deficiency, complications likely to worsen the final outcome occurred in 24%, but only 10% were revised (Zumstein et al., JSES 2011 [16]).

When. The main late complication for aTSA accumulates slowly. In a meta-analysis of 15 studies of cemented all-polyethylene glenoids, mostly for primary osteoarthritis but including some rheumatoid arthritis, followed for a mean of 11.8 years, radiographic glenoid loosening was present in 41% and cuff deficiency in 44%; 17% had been revised. Implant survival was 93% at 10 years, 85% at 15 years and 82.5% at 20 years (Ranieri et al., Int Orthop 2026 [17]). Revision depends on design, technique and patient selection as well as time. In our consecutive series of 458 anatomics for cuff-intact arthritis with a cemented, pegged all-polyethylene glenoid (mean follow-up 6.7 years), 2.6% were revised: seven in the first two years, four between two and five years and one between five and ten years. Six were for subscapularis failure and only one for glenoid loosening, and Simple Shoulder Test scores did not decline after two years (Schiffman et al., Int Orthop 2026 [18]). 

Most comparisons of anatomic and reverse stop at two to five years, before many of these failures appear. Comparable long-term data for the reverse in cuff-intact arthritis do not yet exist.

When the first operation fails, and what comes after

Patients undergoing revision reverse arthroplasty had significantly worse motion, patient-reported outcomes and satisfaction than patients undergoing primary reverse arthroplasty (Tobin et al., JSES 2026 [19]; single manufacturer's registry). These are different patients, and revision patients start with worse shoulders, so the comparison is observational. In Dr. Frankle's single-surgeon registry of 267 reverses done after a failed anatomic, 10.5% were revised again at a mean of 33 months, most often for baseplate failure. In the 188 with one-year follow-up, the Simple Shoulder Test rose from 2.4 to 6.0 of 12 and the ASES score from 30 to 66 (Narain et al., JSEA 2026 [20]). In a larger series from the same group, 320 first-time revision reverses were done for mechanical failure between 2002 and 2022. Implant survival at 10 years was 82.1% when the failed implant was an anatomic and 59.5% when it was a reverse (P < .001). ASES scores at one year were 64 and 53 (P = .008) (Patel et al., JSES 2026 [21]). The worse result after revising a reverse may make surgeons and patients more reluctant to revise one, so a lower revision rate after a reverse does not by itself mean fewer failures. In one series of 202 reverses, baseplate loosening occurred after 1.2% of primary reverses but after 10% of revision reverses (Bitzer et al., JSES 2018 [22]).

Where this leaves us

The existing evidence does not show major differences in clinical outcomes between the anatomic and the reverse for the patient with cuff-intact arthritis. It does show differences in the types, seriousness and frequency of complications, and these differences are not reflected in revision rates. 

In deciding between the two operations, patients need to know what can go wrong with each, how often, and what can be done if it happens. 

Considering the options.

Sibling pileated woodpeckers

Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133/

References

[1] Mayfield CK, et al. Volume, indications, and number of surgeons performing reverse total shoulder arthroplasty continue to expand: a nationwide cohort analysis from 2016-2020. JSES Int. 2023;7(5):827-834. PubMed

[2] Leinweber KA, Bowler AR, Diestel DR, et al., Jawa A. Reverse and anatomic total shoulder arthroplasty for glenohumeral osteoarthritis: a propensity-matched comparison at early and midterm follow-up. J Shoulder Elbow Surg. 2026;35(7):1632-1641. PubMed

[3] Baek CH, Lim C, Kim JG, Kim BT, Kim SJ. Comparative outcomes of anatomic and reverse total shoulder arthroplasty for rotator cuff-intact glenohumeral osteoarthritis: a propensity score-matched, age-stratified exploratory analysis. J Shoulder Elb Arthroplast. 2026;10:100050. doi

[4] Cuff DJ, Simon P, Patel JS, Munassi SD. Anatomic shoulder arthroplasty with high side reaming versus reverse shoulder arthroplasty for eccentric glenoid wear patterns with an intact rotator cuff: comparing early versus midterm outcomes with minimum 7 years of follow-up. J Shoulder Elbow Surg. 2023;32:972-979. doi

[5] Valsamis EM, Prats-Uribe A, Koblbauer I, Cole S, Sayers A, Whitehouse MR, Coward G, Collins GS, Pinedo-Villanueva R, Prieto-Alhambra D, Rees JL. Reverse total shoulder replacement versus anatomical total shoulder replacement for osteoarthritis: population based cohort study using data from the National Joint Registry and Hospital Episode Statistics for England. BMJ. 2024;385:e077939. doi

[6] Shah PP, et al. Anatomical versus reverse total shoulder arthroplasty for primary glenohumeral osteoarthritis with an intact rotator cuff: a systematic review. Bone Jt Open. 2026;7(8):1056-1063. PubMed

[7] Desouza C, et al. Outcomes of reverse vs anatomic total shoulder arthroplasty in glenohumeral osteoarthritis without rotator cuff deficiency: a meta-analysis. World J Orthop. 2026;17(1):110188. PubMed

[8] Fernandes MR, et al. Evidence-based review and recommendations from the Brazilian Society of Shoulder and Elbow Surgery: reverse versus anatomic total arthroplasty in patients aged 70 years or older. Acta Ortop Bras. 2026;34(Spe1):e306296. PubMed

[9] Rodrick HL, et al. RAPSODI-UK randomised controlled trial protocol. BMJ Open. 2025;15(12):e106740. doi

[10] Anatomic versus reverse shoulder arthroplasty for primary glenohumeral osteoarthritis: a randomized prospective study. NCT04629391. ClinicalTrials.gov

[11] Feik ML, Smith AZ, Chen KK, Gehring ZA, Myers NL, Gregory JM. The reliability of the forgotten joint score for shoulder arthroplasty. J Shoulder Elb Arthroplast. 2026;10:100059. PubMed

[12] Beleckas CM, Schodlbauer DF, Mousad AD, Levy JC. Evaluation of new normal after shoulder arthroplasty: comparison of anatomic vs. reverse total shoulder arthroplasty. J Shoulder Elbow Surg. 2025;34:S43-S49. doi

[13] Friedman RJ, Schoch BS, Eichinger JK, et al. Comparison of reverse and anatomic total shoulder arthroplasty in patients with an intact rotator cuff and no previous surgery. J Am Acad Orthop Surg. 2022;30(19):941-948. doi

[14] Gupta MS, Krishan A, Rashid A, Lee MH. Reverse versus anatomic total shoulder arthroplasty in patients over 70 with a competent rotator cuff and glenohumeral osteoarthritis: a meta-analysis. J Shoulder Elbow Surg. 2026;35:1370-1386. doi

[15] O’Malley O, Davies A, Rangan A, Sabharwal S, Reilly P. Is there a difference in thresholds for revision between shoulder arthroplasty types? A National Joint Registry study. PLoS One. 2025;20(8):e0330975. doi

[16] Zumstein MA, Pinedo M, Old J, Boileau P. Problems, complications, reoperations, and revisions in reverse total shoulder arthroplasty: a systematic review. J Shoulder Elbow Surg. 2011;20:146-157. doi

[17] Ranieri R, Illuminati M, Delle Rose G, Borroni M, Conti M, Fontanarosa A, Castagna A, Garofalo R. Long-term glenoid loosening, radiological and clinical outcomes in cemented all-polyethylene anatomic total shoulder arthroplasty: a systematic review and meta-analysis. Int Orthop. 2026;50(7):1683-1696. doi

[18] Schiffman C, Chin E, Whitson A, Hsu J, Matsen FA III. Anatomic shoulder arthroplasty: a single surgeon’s consecutive series of four hundred and fifty eight patients. Int Orthop. 2026;50:2635-2642. doi

[19] Tobin JG, et al. Revision reverse total shoulder arthroplasty: clinical and radiographic outcomes compared to primary reverse total shoulder arthroplasty. J Shoulder Elbow Surg. 2026;35(8):1890-1895. PubMed

[20] Narain AS, Larson JH, Hancock JU, Christmas KN, MacFarlane AJ, Nieboer MJ, Frankle MA. Outcomes of revision reverse total shoulder arthroplasty after failed anatomic total shoulder arthroplasty. J Shoulder Elb Arthroplast. 2026;10(4):100060. PubMed

[21] Patel R, Kucharik MP, Schmidt CM 2nd, Kolakowski L, Christmas KN, Simon P, Frankle MA. Revision reverse shoulder arthroplasty after failed shoulder arthroplasty: long-term survivorship and risk factors for failure. J Shoulder Elbow Surg. 2026;35(10):2346-2353. PubMed

[22] Bitzer A, Rojas J, Patten IS, Joseph J, McFarland EG. Incidence and risk factors for aseptic baseplate loosening of reverse total shoulder arthroplasty. J Shoulder Elbow Surg. 2018;27:2145-2152. doi

When tests for PJI are reassuring, but the shoulder still hurts

Our prior post,When should a failed shoulder arthroplasty be treated for infection? brought up a number of questions about the management of the possibly infected shoulder. One of our shoulder fellows, Brad Graefe, summarized a recent case illustrating some of the challenges. The case is presented with the consent of the patient and the operating surgeon. Here's the story in brief

A patient did well for years after a shoulder arthroplasty and then developed unexplained pain that did not go away. At first, the clinical examination was essentially unchanged, and the radiographs were read as unchanged. The sedimentation rate was 3 mm/h, the C-reactive protein 2.8 mg/L, and the peripheral white count 5.74 ×10³/µL. A joint aspirate returned 3,295 nucleated cells per microliter with 12% neutrophils, 57% lymphocytes and 31% macrophages. At revision surgery, a frozen section of the collar membrane showed two neutrophils per high-power field and no organisms.

Nothing remarkable.

Within days of that surgery, all five specimens grew Cutibacterium acnes.

This post is about the delayed presentation of a Cutibacterium periprosthetic infection after a “honeymoon” period of good function.

The full story

A man in his late fifties came in with advanced glenohumeral arthritis in one shoulder. Years before, he had an os acromiale treated with ORIF. He wanted to keep lifting weights, so he chose a ream-and-run arthroplasty — humeral hemiarthroplasty with concentric reaming of the glenoid — together with an open biceps tenodesis.

Left shoulder radiographs at his 6-month post-operative visit were unremarkable

He did well. More than four years later he reported no pain in that shoulder, was lifting weights and doing push-ups, and his radiographs remained unremarkable. On the strength of that outcome, he had the same operation on the other side, and that shoulder did well too.

 

Left shoulder radiographs at 4-year post-operative visit were again unremarkable.

About six and a half years after the first operation he developed soreness in both shoulders after a day of yard work. The right shoulder settled. The left did not. His examination was unchanged. His radiographs were interpreted as showing no significant interval change compared with prior films. In retrospect there was some evidence of osteolysis of the calcar as seen below.

Left shoulder radiographs at the 6-year post-operative visit 

 

Three months later, his examination had changed: he had pain at the end of active elevation and of rotation in both directions. Active and passive elevation were both 130 degrees, external rotation at the side 20 degrees, external rotation in abduction 30 degrees, internal rotation in abduction 80 degrees, and internal rotation behind the back to the back pocket. Supraspinatus strength was 4/5 and external rotation strength 5−/5. 

Nonoperative treatment was continued, but the pain persisted, so his joint was aspirated and serum markers were drawn, with these results:

·     Synovial nucleated cell count 3,295 per microliter, with 12% neutrophils, 57% lymphocytes and 31% macrophages

·     Erythrocyte sedimentation rate 3 mm/h

·     C-reactive protein 2.8 mg/L

·     Peripheral white blood cell count 5.74 ×10³/µL

The serum markers were normal. The synovial count was slightly above 3,000 cells per microliter, but the differential was lymphocyte-predominant rather than neutrophil-predominant — not the pattern of a septic joint.

Notwithstanding these unimpressive results, the patient had a revision for persistent unexplained pain and because of new radiographs showing osteolysis along the medial calcar, the greater tuberosity, and along the glenoid as shown below.

Radiographs of the left shoulder before revision, showing humeral and glenoid osteolysis 

 

What the operation showed

At surgery, aspiration of the shoulder yielded no joint fluid. There was no gross purulence. There was extensive osteolysis of both the humerus and the glenoid, with bone that looked necrotic. The glenoid had multiple deep areas of osteolysis. 

A frozen section of the collar membrane showed two neutrophils per high-power field and no appreciable bacteria — below the threshold usually taken as evidence of infection. Five specimens were each sent for aerobic, anaerobic, acid-fast and fungal culture.

So at the moment a decision had to be made, the serum markers were normal, the aspirate was not inflammatory, the frozen section was negative, and there was no pus. What was abnormal was the bone.

What was done

The appearance of the tissue was enough to proceed as though this were an infection. The humeral components were removed, the shoulder was extensively debrided and irrigated, and a wound VAC was placed. A postoperative CT was obtained to assess the remaining bone stock.

CT after debridement, obtained to assess the remaining bone stock

Two days later a revision to a reverse total shoulder arthroplasty was carried out using a monoblock baseplate, a hooded glenosphere to load share with the remaining native bone, and a cemented humeral component. The greater tuberosity fractured during instrumentation of the humeral canal which was not surprising given the amount of bone loss from osteolysis. The greater tuberosity was securely repaired by suturing to the implant.

Radiograph after revision to a reverse total shoulder arthroplasty

Three days after the first debridement, one of the operative cultures began to grow Cutibacterium acnes; by post-operative day 8, all five were positive.

Intraoperative Culture Results

Specimen	Culture result	Growth
Humeral head explant	Cutibacterium acnes	1+
Proximal humeral bone	Cutibacterium acnes	1+
Explanted humeral stem	Cutibacterium acnes	1+
Collar membrane	Cutibacterium acnes	1+
Periglenoid tissue	Cutibacterium acnes	1+
Total	5 of 5 specimens positive (100%)	All 1+

 

 

He was discharged on oral amoxicillin with a planned course of about three months and follow-up with an orthopaedic infectious disease specialist.  He is being followed closely.

 

What a new multicenter model says about this case

The ASES Revision Shoulder Arthroplasty and PJI Multicenter Research Group has reported a prediction model for clinically apparent shoulder periprosthetic joint infection, built on 861 consecutive revision arthroplasties performed by 40 surgeons at 20 institutions [1]. Ninety-seven of those shoulders (11%) met the study’s definition of infection: a sinus tract communicating with the prosthesis, or intra-articular purulence. Seven variables entered the final model, which had a concordance statistic of 0.90 in the development cohort. Using this prediction model, a nomogram was constructed, and weights were assigned to each variable. 

 

The model was built to predict infection that is clinically apparent. This patient had neither a sinus tract nor purulence, so by the study’s outcome definition he would have been counted among the shoulders without infection — and he grew Cutibacterium from five of five specimens. The limitations section of the article anticipates patients like this one: low-grade or indolent infections may follow a different biological pathway and may be associated with a different set of predictive factors. What those factors are is not yet known.

What we can take from this case

·     Pain that persists without an explanation is a finding that may suggest infection, even if it surfaces months or years after the index arthroplasty.

·     The lack of a host response by lab values (normal serum inflammatory markers, synovial values, and frozen section) cannot exclude a Cutibacterium PJI.

·     Progressive osteolysis was the principal sign of infection; the laboratory tests did not support it.

·     The necrotic-appearing bone seen at surgery was striking, especially in the absence of purulence and a better explanation for osteolysis such as polyethylene wear.

·     Indolent infections may follow a different physiologic pathway that is not fully captured by a predictive model based on pathognomonic signs of infection as a reference standard.  

 

The bottom line

The objective preoperative signs were a changed examination and progressive osteolysis, first visible at the calcar on films initially read as showing no significant interval change. Every laboratory test was falsely reassuring.

Two questions: 

(1) Is there a better way to determine the risk of infection prior to revision so that treatment can be modified accordingly? 

(2) Where did these Cutibacterium come from – were they present for six years after the index arthroplasty?

Getting things straight

Red shafted northern flicker

Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133/

Reference

[1]   Hsu JE, Ricchetti ET, Tashjian R, Namdari S, Collins AP, Fedorka C, Kim HM, Zhang C, Presson AP, Whitson AJ; American Shoulder and Elbow Surgeons (ASES) Revision Arthroplasty/Periprosthetic Joint Infection (PJI) Multicenter Group. Predictive modeling of clinically apparent shoulder periprosthetic joint infection (PJI). J Shoulder Elbow Surg. Published online October 12, 2026. doi:10.1016/j.jse.2026.10.004

 

 

When should a failed shoulder arthroplasty be treated for infection?

What is the problem?

We know that periprosthetic infections (PJIs) can be devastating for the patient and costly for the health care system. Yet in spite of years of great research by many thoughtful individuals, we’re still pretty much in the dark about the treatment of patients that may have an infected arthroplasty.

A recent JBJS review, The Persistent Challenges of Diagnosing Orthopaedic Implant-Related Infections, concluded that no single test can confirm or exclude implant infection and that the field is moving toward estimating the probability of infection rather than declaring it present or absent. [1] The review is about diagnosis. 

We ask, should the surgeon’s approach to a patient with a 75% probability of a periprosthetic infection (PJI) be different than that for a patient with a 25% probability of PJI? Many patients seem to lie in this intermediate zone.

The article does not address the question the surgeon and patient face at revision: should this patient’s shoulder be treated as if it were infected, and how?

Consider, for example, a patient presents with a failed shoulder arthroplasty: pain, stiffness, a loose component, a worn glenoid. Before we revise it, we'd like to know if we should treat the shoulder as a periprosthetic joint infection, recognizing that the results of cultures obtained at surgery will not be finalized until weeks after the wound is closed.

The size of the problem

Shoulder PJI may be uncommon in any one practice but is costly in total.

•       Rate. In the US Nationwide Inpatient Sample, the PJI rate after primary shoulder arthroplasty was 0.98% from 2002 to 2011 and did not change significantly over that period. [2] From 2011 to 2018, the rate was 1.1%, rising from 0.8% in 2011 to 1.4% in 2018. [3] Registry data with longer follow-up give a different picture by implant type: in 17,730 Nordic primary arthroplasties, the ten-year cumulative rate of revision for infection was 1.4% overall, 3.1% for reverse arthroplasties and 8.0% for reverse arthroplasties in men. [4] In the US inpatient data, by contrast, anatomic total shoulders had the highest proportion of infections (2.0%), followed by hemiarthroplasty (1.0%) and reverse (0.3%). [3] We cannot yet say with confidence which implant type carries the higher risk.

•       Cost per patient. At a single high-volume center, the mean hospital cost per shoulder for two-stage reimplantation was $35,824, compared with $16,068 for a primary arthroplasty, in 2013 dollars. [5] These figures leave out outpatient costs, antibiotics between stages, therapy, extended-care stays, reoperations after the second stage and lost work. [5] The cost of a PJI is incurred on top of the cost of the original arthroplasty. [2]

•       National cost. Hospital charges for shoulder PJI in the US rose 324%, from $44.8 million in 2011 to $190.3 million in 2018, and are projected to approach $500 million annually by 2030. [3] Charges are not costs, and administrative codes can miss the second stage of a two-stage exchange, so these totals are estimates. [5]. The cost of physician time and patient hardship are not included in these numbers.

These numbers do not tell us what we most need to know to manage it: for a patient with a given presentation, which treatment leads to the best comfort and function over the long haul. Registries record a revision for infection as an endpoint; they do not record how the shoulder works afterward. The studies that report treatment results use inconsistent definitions of infection [6, 7], few report patient-reported outcomes, and none compares treatments in matched patients. 

For a complication of this frequency and expense, we have remarkably little operational intelligence on how to manage patients at risk for it.

What is an infection?

Bacterial infection is not the presence of bacteria; it is bacteria doing harm. The 2025 International Consensus Meeting on Musculoskeletal Infection (ICM) takes a similar position. It notes that joints carry their own microbiome, calls the distinction between pathogens and normal flora “an important but unresolved question,” and suggests that the host’s inflammatory status, not the mere detection of an organism, be an important test of whether an infection is present. [8]

We carry about 38 trillion bacteria, roughly as many as our own cells. [9] Most of them keep us healthy, and Cutibacterium is one of them. It is the most abundant bacterium on the healthy, oily skin of the face, chest and back, where it helps keep the surface acidic, competes with potential pathogens for space and nutrients, and limits oxidative stress. [10] In laboratory cultures it modifies the tight junctions between skin cells that form the skin's barrier. [11] It is the loss of Cutibacterium, or of the diversity of its strains, that is associated with skin disorders; people with acne carry about as much Cutibacterium as people without. [10] Removal of Cutibacterium can make the skin less healthy.

Around the shoulder, Cutibacterium lives in the sebaceous glands, below the reach of skin preparation. The incision for shoulder arthroplasty cuts through these glands: after chlorhexidine preparation, 40% of dermal biopsies at open shoulder surgery grew Cutibacterium. [12] Every shoulder arthroplasty wound is inoculated with Cutibacterium.

When a failed arthroplasty is revised, three questions are often blended into one:

1.     Are bacteria present?

2.     Are those bacteria contributing to the pain, stiffness or loosening?

3.     Will treating this shoulder as a PJI change the patient's outcome?

Attempting to “define” PJI does not settle the matter. A systematic review of 20 studies of revision shoulder arthroplasty found that 6 (27%) gave no explicit definition of infection. [6] A 2025 review of 231 shoulder studies found the diagnostic workup still inconsistent. [7] A 2026 review in JBJS notes that there is no true diagnostic gold standard and that up to 15% of hip and knee cases remain inconclusive under the 2018 ICM criteria, which most diagnostic studies still use. [1] The 2025 ICM changed its recommendations on 37 (22%) of the 172 questions it shared with the 2018 meeting; 23% of those shared questions concerned the shoulder. [13]

How often failed shoulders grow bacteria

About half of failed shoulder arthroplasties will grow bacteria at revision, most often Cutibacterium, and more often in men.

•       Of 193 revisions for pain, loosening or stiffness, 108 (56%) were culture positive. Of the positive cultures, 70% grew Cutibacterium and 55% took more than a week to grow. Male sex, humeral osteolysis and cloudy fluid each increased the likelihood of a positive Cutibacterium culture dramatically. [14]

•       Of 221 revisions without obvious infection, at least one positive culture was found in 54% of loose glenoid components and 51% of secure ones (P = .77). [15]

•       Deep cultures at revision were positive in 12 of 20 men and 1 of 20 women. [16]

•       In a multicenter series of 490 revisions, 14% met ICM criteria for definite infection and 57% were classed as unlikely. Cutibacterium was the most common organism in both groups: 25% of definite and 27% of non-definite cases. The definite infections that grew Cutibacterium presented with gross purulence 72% of the time. [17]

Two points follow. The chance of bacteria in a failed shoulder is high before we know anything else about it. And the organism found in "definite" infections is the same one found in shoulders classified as "probable," "possible" or "unlikely" infections. Authors suggest that labeling organisms "virulent" or "low virulence" may need reconsideration, and that classification might rest better on the clinical presentation than on the species. [17]

Why the tests cannot settle it

An “obvious” infection may be signaled by pus, draining sinus, abnormal inflammatory labs, fever, erythema, and/or positive aspirate culture for virulent organisms. In a failed shoulder without obvious infection, no test available before or during surgery is accurate enough to rule bacteria in or out at the moment the surgical and postoperative antibiotic decisions need to be made.

Serum tests. In 579 revision shoulder arthroplasties, CRP discriminated poorly even in definite infections: at its best cutoff of 5.9 mg/L, the area under the curve was 0.66, with 82% sensitivity and 54% specificity. In the remaining revisions, ESR performed worse than chance. [18] The 2025 ICM recommends CRP and ESR before every hip and knee revision, while noting that serology is often normal in infections with Cutibacterium and coagulase-negative staphylococci. [19]

Synovial fluid. Aspiration of the shoulder often yields little or no fluid. Where fluid was obtained in 425 revisions, the synovial neutrophil percentage discriminated best, with cutoffs between 72% and 78%. [20] In 105 aspirated shoulders, alpha-defensin had 75% sensitivity and 96% specificity; leukocyte esterase had 50% and 87%. [21] Aspiration is likely to miss Cutibacterium. [22] In a four-center series, aspiration detected none of the infections found at open biopsy; percutaneous synovial biopsy detected 37%. [23] For hip and knee revisions, the 2025 ICM named leukocyte esterase the best intraoperative test (pooled sensitivity 82.5%), with the caveat that its sensitivity is limited in low-grade infection; [24] in the shoulder its sensitivity was 50%. [21]

Frozen section. In the same 425 revisions, frozen section detected only 20% to 28% of the shoulders with bacteria among the cases other than those with definite infection, which are the cases in which we need help. [20]

Nuclear imaging. Labeled white-cell SPECT/CT failed to diagnose chronic shoulder infection. [25]

What this means in numbers. Start with a hypothetical patient with a 50% chance of bacterial presence. Suppose the CRP is below 5.9 mg/L. With 82% sensitivity and 54% specificity, a normal CRP has a likelihood ratio of about 0.34, and the chance of bacteria falls from 50% to about 25%. One shoulder in four has not been ruled out, and those figures come from the definite infections, where CRP performs best and where we need it the least. 

How useful is this test in determining treatment?

Even the cultures are ambiguous

A positive culture does not prove infection, and a negative set does not exclude it.

Positive cultures without infection. In primary open shoulder surgery, 7 of 54 sterile control specimens (13%) grew bacteria, and 17% of shoulders without prior surgery had at least one positive Cutibacterium culture. [26] Air swabs exposed for 5 seconds in laminar-flow operating rooms grew Cutibacterium in 6 of 40 cases (15%); none of those patients had an infection at 90 days. [27] Positive cultures are common even at primary shoulder arthroplasty. [28]

Negative cultures with bacteria present. In 137 Cutibacterium-positive revisions with at least four specimens, a mean of 43% of the specimens from each positive shoulder showed no growth. [29] The organism is unevenly distributed, so the number and source of specimens change the answer; the rate of positive shoulders rises with the number of specimens taken. [14] In a hip and knee series, ultrasonic disruption of biofilm during revision raised the rate of positive cultures from the surgical-field fluid from 40% to 76%, revealing bacteria that standard sampling had missed. [30]

What helps sort them. Load and timing. Every probable true-positive Cutibacterium culture turned positive by day 11, while 4 of 9 probable contaminants turned positive later; true positives also had more positive specimens. [31]

This is why we take five deep tissue or explant specimens, culture them for 14 days on aerobic and anaerobic media, and grade each plate from 0 to 4+. The result is a measure of bacterial load and where the bugs are – better information than characterizing a shoulder as “culture positive” or “culture negative”. The 2025 ICM recommendations are similar: four to six specimens, at least three of them tissue; anaerobic incubation for up to 14 days; and both solid and liquid media for Cutibacterium. [8]

Do positive cultures change the outcome?

When failed shoulders are revised in a single stage and given antibiotics, those that turn out to be culture positive do as well as those that do not.

•       In 55 single-stage revisions without obvious infection, 27 shoulders with at least two Cutibacterium-positive cultures (89% male) were compared with 28 with zero or one. All received antibiotics adjusted to the cultures. In the culture-positive group, the Simple Shoulder Test score rose from 3.2 to 7.8 at a mean of 46 months, at least as good as in controls. Three patients (11%) in each group needed further surgery for pain or stiffness; none of those re-revisions was culture positive. Fourteen patients reported antibiotic side effects. [32]

•       In 92 revisions, patients judged at high risk received intravenous antibiotics and those at low risk received oral antibiotics, both adjusted to the cultures. At a mean of 4.1 years, infection-free survival was 91%; infection-free survival, revision-free survival and patient-reported outcomes were similar between groups, but the intravenous group had more antibiotic adverse events. [33]

•       In 124 revisions for reasons other than suspected infection, 27 (22%) had unexpected positive cultures, mostly Cutibacterium. Gains in shoulder score and motion did not differ from those of culture-negative shoulders. [34]

•       In 65 conversions of anatomic to reverse arthroplasty with the humeral stem retained, all classed as unlikely infection by ICM criteria, 9 (14%) had positive cultures, 8 of them Cutibacterium. Only 6 patients received postoperative antibiotics, so at least 3 culture-positive shoulders went untreated. Two patients (3%) needed reoperation for infection; neither had a positive culture at the index revision. [35]

•       In an early series of 28 revisions presumed aseptic, 8 (29%) were culture positive, and 2 of those 8 later developed infection. [36]

What these studies demonstrate: Culture-positive failed shoulders revised in one stage and given antibiotics reach outcomes similar to those of culture-negative shoulders. No study has randomized culture-positive shoulders to antibiotics or none, so the benefit of the antibiotics themselves is inferred, not shown. The stem-retention series suggests that some low-load positives do well without treatment; the early series reminds us that some unexpected positives go on to declare themselves. Both series are small. Here our practice departs from the 2025 ICM, which advises that an organism found without an inflammatory response not be interpreted as a pathogen. [8] The reason for the difference is in the threshold argument below.

Every treatment carries its own risk

The treatments for infection range from a few weeks of oral pills to months with removal of a shoulder implant; the more aggressive treatments endanger bone and soft tissue, imposing their harms on every patient who receives them, infected or not.

•       Two-stage exchange. Of 49 shoulders scheduled for two-stage exchange, 14 (29%) were never reimplanted. Mortality was 10% within 90 days of implant removal and 25% at latest follow-up. [37]

•       One stage versus two in the shoulder. Across 13 single-stage and 30 two-stage series, reinfection was 6.3% versus 10.1%, a difference that was not significant; two-stage cases had more MRSA (9.7% versus 2.5%), so the comparison is confounded. [38] A 2024 meta-analysis found reinfection of 10.9% versus 12.9% and an average of 1.16 versus 2.25 further operations, with selection bias as a major limitation. [39]

•       One stage versus two, randomized, in the hip and knee. In 323 chronic hip and knee infections randomized to one or two stages, success at 2 years was 97% (131 of 135) versus 91% (112 of 123), meeting the criterion for noninferiority. Prior revisions, immunosuppressed patients, and culture-negative and fungal infections were excluded. There is no such trial in the shoulder. [40] A single-arm hip and knee series of single-stage revision with ultrasonic debridement reported 96% apparent freedom from infection at a mean of 29 months. [30] Of course there is no test for shown the absence of bacteria.

•       Consensus. The 2025 ICM found no absolute contraindications to one-stage exchange for chronic hip and knee infection (90% delegate agreement), considered it viable for culture-negative infections and, when the soft tissues can be closed, for joints with a sinus tract, and recommended at least 6 weeks of antibiotics after a one-stage exchange. [41] It found too little evidence to set the duration of antibiotics between the stages of a two-stage exchange or to support an antibiotic holiday before reimplantation. [42]

•       Oral versus intravenous antibiotics, randomized. In 1,054 bone and joint infections, treatment failure at 1 year was 14.6% with intravenous and 13.2% with oral antibiotics; oral therapy was noninferior. [43]

•       Antibiotic harms after shoulder revision. Among 175 revisions, 19% of patients had adverse effects from postoperative antibiotics, more often with the intravenous protocol. [44]

The shoulder comparisons come from observational series in which sicker patients with more virulent organisms were chosen for two stages, which makes two-stage results look worse than they would in matched patients. Even so, nothing in the shoulder literature shows that two stages produce better comfort or function.

Tipping the balance: the host and the operating room

Infection is a balance between bacteria and host, and part of that balance is in our hands before and during surgery.

•       The 2025 ICM found preoperative anemia to be a risk factor for infection (in shoulder arthroplasty, odds ratio 2.04) and allogeneic transfusion to be associated with infection (risk ratio 1.94). It found tranexamic acid associated with less periprosthetic infection (odds ratio 0.69), although delegate agreement on that point was only 63%; it favored aspirin as the thromboprophylaxis with the lowest infection risk. [45]

•       In 7,713 shoulder arthroplasties, cefazolin prophylaxis, compared with vancomycin or clindamycin, was associated with 69% less periprosthetic infection and 78% less Cutibacterium infection. [46]

•       The 2025 ICM concluded that a corticosteroid injection within 3 months of shoulder arthroplasty is associated with a higher risk of PJI. [13]

•       In 33,470 total shoulder arthroplasties, surgical-site infection rose once operative time passed 180 minutes. [47]

•       Neither chlorhexidine wash nor benzoyl peroxide soap used at home removed Cutibacterium from the shoulder skin, [48] and after chlorhexidine preparation, Cutibacterium repopulated the skin surface within an hour. [49] Hydrogen peroxide added to the skin preparation reduced but did not eliminate Cutibacterium in a controlled trial. [50]

•       Dilute povidone-iodine lavage before closure was associated with fewer acute infections after hip and knee arthroplasty (0.97% versus 0.15%). [51]

•       Vancomycin prevented Cutibacterium growth in a laboratory model of shoulder arthroplasty, [52] and a cost analysis concluded that topical vancomycin pays for itself if it prevents only a small number of infections. [53]

Most of these are associations from observational data, and few are specific to shoulder revision.  

When to treat: a threshold, not a diagnosis

Consider treating a failed shoulder arthroplasty as if it were infected, and scale the intensity of treatment to the likelihood of infection, the virulence of the organism and what the patient, their bone and soft tissues can tolerate.

The idea comes from clinical decision analysis. Any treatment does some good in the patients who need it and some harm in the patients who do not. Whether it is worth giving depends on how likely it is that this patient needs it.

Here is a hypothetical illustration. Imagine 100 failed shoulders. Suppose that treating an infected shoulder with oral amoxicillin prevents 10 units of harm, such as a recurrent failure and another operation. Suppose that treating an uninfected shoulder causes 1 unit of harm, from side effects, cost and resistance. If 9 of the 100 shoulders are infected, treating all 100 prevents 90 units of harm and causes 91: roughly a draw. If more than 9 in 100 are infected, treating does more good than harm; if fewer, it does more harm than good. That break-even point, about 9%, is the treatment threshold.

Neither number has been measured for the shoulder, but the logic explains why the answer depends on the treatment. When the treatment carries little harm, the threshold is low. A failed shoulder starts near a 50% chance of having bacteria in it, far above that threshold, and no test we have can bring it below. That supports a single-stage exchange with oral antibiotics for every revised shoulder as long as it can be accomplished without significant bone or soft tissue damage. A two-stage exchange is different. For an uninfected patient, months without a shoulder arthroplasty, a failed reimplantation or death is a harm nearly as large as the benefit to an infected one. When the harm equals the benefit, the threshold is 50%; when it is larger, the threshold is higher still. Thus, two-stage exchange may be appropriate for shoulders in which infection is obvious: purulence, a sinus tract or sepsis.

In practice:

1.     Before surgery, estimate the starting probability. Male sex, humeral osteolysis, onset of pain or stiffness after a good early result, and prior surgery raise it.

2.     Decide where the shoulder lies on the spectrum. Fever, a sinus tract, purulence or sepsis, whatever the organism, suggests aggressive debridement matched to what the patient can tolerate.

3.     For most cases, consider the single stage. Take five deep specimens and a frozen section, remove the components if this can be accomplished without significant bone and soft tissue damage, debride thoroughly, irrigate copiously, lavage with povidone-iodine, apply topical vancomycin powder and reimplant without cement.

4.     Start oral antibiotics at once. We use amoxicillin 1 g three times daily for 3 weeks while the cultures incubate, extended to 6 months if they are positive.

5.     Support the host. Correct anemia and malnutrition; we recommend 2 g of protein per kg per day and 2,000 IU of vitamin D daily until healed.

The cultures arrive after the important decision has been made. Their job is assure that the antibiotics initiated are appropriate the culture findings and to set the duration of antibiotic treatment. Infectious disease consultation is often helpful in making these decisions.

What we still do not know

Several links in this reasoning have never been tested directly.

•       No trial has randomized culture-positive shoulders, once the cultures are known, to stop or continue antibiotics, so we do not know how much the antibiotics contribute, or whether low-load positives need treatment at all.

•       No matched trial has compared single-stage and two-stage revision in the shoulder.

•       The optimal duration of oral antibiotics after a culture-positive revision is unknown.

•       The literature says almost nothing about the failed shoulder that is not being revised: the patient managed without surgery or with surgery that does not include implant exchange.

•       Most importantly, data are lacking on the clinical outcomes (i.e., patient comfort and function) for patients with different sets of findings (e.g., abnormal lab tests, purulence, draining sinus vs none of these; bacteria recovered; bacterial load) having different treatments (e.g., non-operative, debridement only, single-stage revision, two-stage revision).

Conclusion

A lot of good research has gone into the definition of shoulder periprosthetic infection. 

What is lacking is evidence on clinical outcomes–– patient comfort and function––, for the common clinical presentations managed with the common treatment strategies. The 2025 ICM reached a similar view: with 87% delegate agreement, it recommended that patient-reported outcomes take priority over infection eradication, reoperation and continued antibiotics for reporting the success of PJI treatment. It found that only one randomized trial of PJI treatment had used a patient-reported outcome as its primary endpoint. [55] 

The path forward may lie in (1) developing a classification system for the most common clinical presentations, for example by history, physical findings, laboratory findings, the presence of purulence or a draining sinus, and bacterial load; (2) developing a classification system for the most common treatment strategies, from non-operative care to debridement, single-stage and two-stage revision; and (3) mapping the outcomes of each strategy onto each presentation.

As a rough draft of a system for classifying common presentations using the most dominant feature:

A. Unexpected positive cultures

B. Otherwise unexpected pain and stiffness after a honeymoon period of normal post arthroplasty recovery

C. Loose humeral component

D. Osteolysis

E. Pus

F. Draining sinus

G. Sepsis

Are there bugs in there? How much does it matter?

Lewis's Woodpecker

Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133/

References

1.          Lum ZC, Cohen-Rosenblum A, Yao JJ, et al. The persistent challenges of diagnosing orthopaedic implant-related infections. J Bone Joint Surg Am. 2026;108(15):1142-8. doi:10.2106/JBJS.25.01516

2.         Padegimas EM, Maltenfort M, Ramsey ML, Williams GR, Parvizi J, Namdari S. Periprosthetic shoulder infection in the United States: incidence and economic burden. J Shoulder Elbow Surg. 2015;24(5):741-6. doi:10.1016/j.jse.2014.11.044

3.         Schick S, Elphingstone J, Murali S, et al. The incidence of shoulder arthroplasty infection presents a substantial economic burden in the United States: a predictive model. JSES Int. 2023;7(4):636-41. doi:10.1016/j.jseint.2023.03.013

4.         Moeini S, Rasmussen JV, Salomonsson B, et al. Reverse shoulder arthroplasty has a higher risk of revision due to infection than anatomical shoulder arthroplasty: 17 730 primary shoulder arthroplasties from the Nordic Arthroplasty Register Association. Bone Joint J. 2019;101-B(6):702-7. doi:10.1302/0301-620X.101B6.BJJ-2018-1348.R1

5.         Baghdadi YMK, Maradit-Kremers H, Dennison T, et al. The hospital cost of two-stage reimplantation for deep infection after shoulder arthroplasty. JSES Open Access. 2017;1(1):15-8. doi:10.1016/j.jses.2017.02.001

6.         Hsu JE, Somerson JS, Vo KV, et al. What is a "periprosthetic shoulder infection"? A systematic review of two decades of publications. Int Orthop. 2017;41(4):813-22. doi:10.1007/s00264-017-3421-6

7.         Clifford AL, Hurley E, Rowe D, et al. How is infection diagnostic criteria for shoulder periprosthetic joint infection reported in literature: systematic review. JSES Int. 2025;9(1):219-25. doi:10.1016/j.jseint.2024.09.022

8.         Rottman M, Gordon AM, Kang KS, et al. 2025 ICM: sample collection and agent identification. J Arthroplasty. 2025;41(1 Suppl 1):S157-63. doi:10.1016/j.arth.2025.08.040

9.         Sender R, Fuchs S, Milo R. Revised estimates for the number of human and bacteria cells in the body. PLoS Biol. 2016;14(8):e1002533. doi:10.1371/journal.pbio.1002533

10.       Rozas M, Hart de Ruijter A, Fabrega MJ, et al. From dysbiosis to healthy skin: major contributions of Cutibacterium acnes to skin homeostasis. Microorganisms. 2021;9(3):628. doi:10.3390/microorganisms9030628

11.        Bolla BS, Erdei L, Urbán E, et al. Cutibacterium acnes regulates the epidermal barrier properties of HPV-KER human immortalized keratinocyte cultures. Sci Rep. 2020;10(1):12815. doi:10.1038/s41598-020-69677-6

12.       Phadnis J, Gordon D, Krishnan J, et al. Frequent isolation of Propionibacterium acnes from the shoulder dermis despite skin preparation and prophylactic antibiotics. J Shoulder Elbow Surg. 2016;25(2):304-10. doi:10.1016/j.jse.2015.08.002

13.       Ferrini A, He M, Bernaus M, Veloso M, Mont MA, Parvizi J. International Consensus Meeting on Orthopaedic Infection: differences between ICM 2018 and ICM 2025. J Arthroplasty. 2026;41(2):304-7. doi:10.1016/j.arth.2025.10.113

14.       Pottinger P, Butler-Wu S, Neradilek MB, et al. Prognostic factors for bacterial cultures positive for Propionibacterium acnes and other organisms in a large series of revision shoulder arthroplasties performed for stiffness, pain, or loosening. J Bone Joint Surg Am. 2012;94(22):2075-83. doi:10.2106/JBJS.K.00861

15.       Lucas RM, Hsu JE, Whitney IJ, et al. Loose glenoid components in revision shoulder arthroplasty: is there an association with positive cultures? J Shoulder Elbow Surg. 2016;25(8):1371-5. doi:10.1016/j.jse.2015.12.026

16.       Matsen FA 3rd, Butler-Wu S, Carofino BC, et al. Origin of Propionibacterium in surgical wounds and evidence-based approach for culturing Propionibacterium from surgical sites. J Bone Joint Surg Am. 2013;95(23):e181. doi:10.2106/JBJS.L.01733

17.       Ricchetti ET, Collins AP, Khazzam MS, et al. Evaluation of the International Consensus Meeting (ICM) diagnostic categories for shoulder periprosthetic joint infection: Cutibacterium acnes predominates even in definite infections. J Shoulder Elbow Surg. 2026;35(10):2362-71. doi:10.1016/j.jse.2026.04.053

18.       Collins AP, Whitson AJ, Budge MD, et al. Can standard preoperative serum laboratory tests predict bacterial presence at the time of revision shoulder arthroplasty? J Shoulder Elbow Surg. 2026;35(10):2372-81. doi:10.1016/j.jse.2026.04.036

19.       Pupaibool J, Tarabichi S, Shahi A, et al. 2025 ICM: serological diagnosis of surgical site infection (SSI)/periprosthetic joint infection (PJI). J Arthroplasty. 2025;41(1 Suppl 1):S248-60. doi:10.1016/j.arth.2025.10.102

20.      Collins AP, Whitson AJ, Duquin TR, et al. Defining optimal synovial aspirate thresholds and utility of frozen sections in predicting bacterial presence at the time of revision shoulder arthroplasty. J Shoulder Elbow Surg. 2026;35(11):2607-17. doi:10.1016/j.jse.2026.05.005

21.       Unter Ecker N, Koniker A, Gehrke T, et al. What is the diagnostic accuracy of alpha-defensin and leukocyte esterase test in periprosthetic shoulder infection? Clin Orthop Relat Res. 2019;477(7):1712-8. doi:10.1097/CORR.0000000000000762

22.      Streck LE, Forster J, von Hertzberg-Boelch SP, et al. The role of synovial fluid aspiration in shoulder joint infections. BMC Musculoskelet Disord. 2022;23(1):390. doi:10.1186/s12891-022-05285-x

23.      Lapner P, Nam D, Cheema A, et al. Diagnostic accuracy of preoperative percutaneous synovial biopsy and aspirate compared with open biopsy for prosthetic shoulder infections. J Shoulder Elbow Surg. 2025;34(2):441-8. doi:10.1016/j.jse.2024.08.016

24.      Mortazavi SMJ, Abdelazeem A, Abdelnasser MK, et al. 2025 ICM: diagnosis of surgical site infection (SSI)/periprosthetic joint infection (PJI) with samples from the wound and clinical signs. J Arthroplasty. 2025;41(1 Suppl 1):S243-7. doi:10.1016/j.arth.2025.10.107

25.      Falstie-Jensen T, Daugaard H, Søballe K, et al. Labeled white blood cell/bone marrow single-photon emission computed tomography with computed tomography fails in diagnosing chronic periprosthetic shoulder joint infection. J Shoulder Elbow Surg. 2019;28(6):1040-8. doi:10.1016/j.jse.2018.10.024

26.      Mook WR, Klement MR, Green CL, et al. The incidence of Propionibacterium acnes in open shoulder surgery: a controlled diagnostic study. J Bone Joint Surg Am. 2015;97(12):957-63. doi:10.2106/JBJS.N.00784

27.      Namdari S, Nicholson T, Parvizi J. Cutibacterium acnes is isolated from air swabs: time to doubt the value of traditional cultures in shoulder surgery? Arch Bone Jt Surg. 2020;8(4):506-10. doi:10.22038/abjs.2020.40642.2095

28.      Wong JC, Schoch BS, Lee BK, et al. Culture positivity in primary total shoulder arthroplasty. J Shoulder Elbow Surg. 2018;27(8):1422-8. doi:10.1016/j.jse.2018.05.024

29.      Ahsan ZS, Somerson JS, Matsen FA 3rd. Characterizing the Propionibacterium load in revision shoulder arthroplasty: a study of 137 culture-positive cases. J Bone Joint Surg Am. 2017;99(2):150-4. doi:10.2106/JBJS.16.00422

30.      Ji B, Yang C, Aimaiti A, et al. Effective treatment using a single-stage revision with non-contact low frequency ultrasonic debridement in the treatment of periprosthetic joint infection: a prospective single-arm study. Bone Joint J. 2025;107-B(6 Supple B):9-14. doi:10.1302/0301-620X.107B6.BJJ-2024-0848.R2

31.       Frangiamore SJ, Saleh A, Grosso MJ, et al. Early versus late culture growth of Propionibacterium acnes in revision shoulder arthroplasty. J Bone Joint Surg Am. 2015;97(14):1149-58. doi:10.2106/JBJS.N.00881

32.      Hsu JE, Gorbaty JD, Whitney IJ, Matsen FA 3rd. Single-stage revision is effective for failed shoulder arthroplasty with positive cultures for Propionibacterium. J Bone Joint Surg Am. 2016;98(24):2047-51. doi:10.2106/JBJS.16.00149

33.      Yao JJ, Jurgensmeier K, Whitson AJ, et al. Oral and IV antibiotic administration after single-stage revision shoulder arthroplasty. J Bone Joint Surg Am. 2022;104(5):421-9. doi:10.2106/JBJS.20.02263

34.      Falstie-Jensen T, Lange J, Daugaard H, et al. Unexpected positive cultures after revision shoulder arthroplasty: does it affect outcome? J Shoulder Elbow Surg. 2021;30(6):1299-308. doi:10.1016/j.jse.2020.12.014

35.      Shenoy DA, Parsons KE, Henriquez AR, et al. Occult infection risk with stem retention in revision anatomic to reverse shoulder arthroplasty. J Shoulder Elbow Surg. 2026. Epub ahead of print. doi:10.1016/j.jse.2026.10.007

36.      Kelly JD 2nd, Hobgood ER. Positive culture rate in revision shoulder arthroplasty. Clin Orthop Relat Res. 2009;467(9):2343-8. doi:10.1007/s11999-009-0875-x

37.      Akgün D, Wiethölter M, Maziak N, et al. Two-stage exchange arthroplasty for periprosthetic shoulder infection is associated with high rate of failure to reimplant and mortality. J Clin Med. 2021;10(21):5186. doi:10.3390/jcm10215186

38.      Belay ES, Danilkowicz RM, Bullock GS, et al. Single-stage versus two-stage revision for shoulder periprosthetic joint infection: a systematic review and meta-analysis. J Shoulder Elbow Surg. 2020;29(12):2476-86. doi:10.1016/j.jse.2020.05.034

39.      Bdeir M, Lerchl A, Hetjens S, et al. One- vs. two-stage revision for periprosthetic shoulder infections: a systematic review and meta-analysis. Antibiotics (Basel). 2024;13(5):440. doi:10.3390/antibiotics13050440

40.      Fehring TK, Otero JE, Fehring KA, et al. One-stage versus two-stage exchange arthroplasty for periprosthetic joint infection. J Bone Joint Surg Am. 2026;108(14):1070-82. doi:10.2106/JBJS.25.00713

41.       Hansen E, Ji B, Dietz MJ, et al. 2025 ICM: one-stage exchange. J Arthroplasty. 2025;41(1 Suppl 1):S326-40. doi:10.1016/j.arth.2025.10.099

42.      Elhence A, Böhler C, Kolhoff F, et al. 2025 ICM: two-stage. J Arthroplasty. 2025;41(1 Suppl 1):S385-98. doi:10.1016/j.arth.2025.10.095

43.      Li HK, Rombach I, Zambellas R, et al. Oral versus intravenous antibiotics for bone and joint infection. N Engl J Med. 2019;380(5):425-36. doi:10.1056/NEJMoa1710926

44.      Yao JJ, Jurgensmeier K, Woodhead BM, et al. The use and adverse effects of oral and intravenous antibiotic administration for suspected infection after revision shoulder arthroplasty. J Bone Joint Surg Am. 2020;102(11):961-70. doi:10.2106/JBJS.19.00846

45.      Mortazavi SMJ, Palmer A, Yildiz F, et al. 2025 ICM: anemia, blood-saving, and thromboprophylaxis. J Arthroplasty. 2025;41(1 Suppl 1):S19-29. doi:10.1016/j.arth.2025.10.082

46.      Marigi EM, Bartels DW, Yoon JH, et al. Antibiotic prophylaxis with cefazolin is associated with lower shoulder periprosthetic joint infection rates than non-cefazolin alternatives. J Bone Joint Surg Am. 2022;104(10):872-80. doi:10.2106/JBJS.21.00445

47.      Schmitt MW, Chenault PK, Samuel LT, et al. The effect of operative time on surgical-site infection following total shoulder arthroplasty. J Shoulder Elbow Surg. 2023;32(11):2371-5. doi:10.1016/j.jse.2023.05.010

48.      Hsu JE, Whitson AJ, Woodhead BM, et al. Randomized controlled trial of chlorhexidine wash versus benzoyl peroxide soap for home surgical preparation: neither is effective in removing Cutibacterium from the skin of shoulder arthroplasty patients. Int Orthop. 2020;44(7):1325-9. doi:10.1007/s00264-020-04594-x

49.      Hsu JE, Whitson AJ, Van Dyke R, et al. Dynamics of Cutibacterium repopulation onto the skin surface of the shoulder after chlorhexidine application. Int Orthop. 2023;47(6):1511-5. doi:10.1007/s00264-023-05786-x

50.      Chalmers PN, Beck L, Stertz I, Tashjian RZ. Hydrogen peroxide skin preparation reduces Cutibacterium acnes in shoulder arthroplasty: a prospective, blinded, controlled trial. J Shoulder Elbow Surg. 2019;28(8):1554-61. doi:10.1016/j.jse.2019.03.038

51.       Brown NM, Cipriano CA, Moric M, et al. Dilute betadine lavage before closure for the prevention of acute postoperative deep periprosthetic joint infection. J Arthroplasty. 2012;27(1):27-30. doi:10.1016/j.arth.2011.03.034

52.      Miquel J, Huang TB, Athwal GS, et al. Vancomycin is effective in preventing Cutibacterium acnes growth in a mimetic shoulder arthroplasty. J Shoulder Elbow Surg. 2022;31(1):159-64. doi:10.1016/j.jse.2021.06.015

53.      Hatch MD, Daniels SD, Glerum KM, Higgins LD. The cost effectiveness of vancomycin for preventing infections after shoulder arthroplasty: a break-even analysis. J Shoulder Elbow Surg. 2017;26(3):472-7. doi:10.1016/j.jse.2016.07.071

54.      Matsen FA 3rd, Whitson A, Hsu JE. Preoperative skin cultures predict periprosthetic infections in revised shoulder arthroplasties. JBJS Open Access. 2020;5(4):e20.00095. doi:10.2106/JBJS.OA.20.00095

55.      Manning L, Slullitel PA, Buttaro MA, et al. 2025 ICM: monitoring and declaring success. J Arthroplasty. 2025;41(1 Suppl 1):S437-42. doi:10.1016/j.arth.2025.10.104