Periprosthetic Infections of the Shoulder

Diagnosis and Management

E. Scott Paxton, MD; Andrew Green, MD; Van S. Krueger, MD, MBA


J Am Acad Orthop Surg. 2019;27(21):e935-e944. 

In This Article

Clinical Evaluation

Patient Evaluation

Avariety of patient factors is associated with an increased risk of PJIS and should heighten the clinical suspicion. These include male sex, higher body mass index, and younger age at the time of primary surgery, as well as immune-compromising conditions and medications.[4,13–16] In a large, retrospective cohort study by Richards et al,[17] every 1 year increase in patient age conferred a 5% lower infection rate (95% confidence interval, 2% to 8%) and male patients had a 2.59 times (95% confidence interval, 1.27 to 5.31) greater infection rate than female patients. Additionally, arthroplasty after trauma, reverse shoulder arthroplasty, and a history of non-arthroplasty surgery are the risk factors for PJIS.[17,18]


High-quality plain radiographs over time are required for the proper evaluation of implant loosening and osteolysis. Early loosening is rare and concerning for an infection (Figure 1). The significance of humeral osteolysis is less clear because it is associated with press fit implants and stress shielding. However, infection needs to be considered if progressive bone loss is present with accompanying pain.

Figure 1.

AP radiograph of the right shoulder of a 60-year-old woman 18 months after implantation of a reverse shoulder arthroplasty for a fracture. She had persistent pain and stiffness. There is evidence of loosening with glenoid component migration and lucency about the cemented humeral stem. These components were grossly loose at the time of revision, and cultures were positive for Cutibacterium acnes.

Pottinger et al[19] reported that humeral component loosening and humeral osteolysis had an odds ratio of 3 and 10, respectively, for the risk of a positive C acnes culture. Although nonspecific, other radiographic cues indicative of a PJIS include joint effusion causing inferior subluxation of the humeral component, endosteal scalloping, generalized bone resorption, periosteal reaction, and periprosthetic radiolucency.[20]

Advanced imaging is used on a case-by-case basis in the evaluation of infection. With the advent of metal artifact reduction protocols, CT and MRI can be used to assess the osseous and soft-tissue structures, respectively. A CT arthrogram may be used to evaluate rotator cuff tendon integrity and glenoid component loosening.

Technetium-99m three-phase bone scan imaging is sensitive for arthroplasty failure but is limited in determining the etiology.[20] An isolated, indium-labeled white blood cell (WBC) study has proven to be poor at detecting PJIS caused by low-virulence organisms.[21] For these reasons, nuclear imaging is not routinely used in the evaluation of PJIS.

Serologic Testing

Serum WBC count, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) are standard in the workup of periprosthetic infection. Similar to those in patients undergoing hip and knee arthroplasty, these inflammatory markers have a high specificity and positive predictive value (PPV) in PJIS. However, due to the greater prevalence of low-virulence infections, the sensitivity and negative predictive value (NPV) are significantly lower in PJIS. In a recent systematic review, only 6.8% and 37.6% of patients with a PJIS had an elevated WBC and ESR, respectively.[6] By contrast, CRP was elevated in 62.1% of cases.[6] Because of this low sensitivity, these tests are often normal, especially late after primary surgery. Therefore, it must be understood that infection cannot be excluded based on these tests.

Arthrocentesis/Joint Aspiration

In contrast to the lower extremity, the relevance of aspiration in the evaluation of PJIS is less clear.[22] Shoulder aspiration, even under imaging guidance, rarely yields enough synovial fluid for the analysis and culture. Fluoroscopic-and ultrasonography-guided aspiration ensures intra-articular placement of the aspiration; however, both techniques, especially ultrasonography, can be operator dependent and add additional costs. Furthermore, fluoroscopy exposes the patient to radiation. Patients should be off antibiotics for 2 weeks before aspiration for optimal culture results. In the event enough fluid is obtained, the cultures have a high specificity and PPV but low sensitivity and NPV.[23] Much like normal serologic testing, joint aspiration with a dry tap or with negative cultures does not rule out an infection.

Cutoff values for synovial fluid cell count and differential have not been definitively established for shoulder arthroplasty. The primary benefit of aspiration is attained if there is synovial fluid with an elevated WBC and positive cultures as this information may facilitate the decision for definite surgical treatment—one- or two-stage revision. However, in the setting of normal serologic markers and no other risk factors, the utility of a joint aspiration should be left up to the clinical judgment of the surgeon.

Synovial Fluid Analysis

There is a growing experience with the analysis of synovial fluid inflammatory markers to identify PJIS. Frangiamore et al[24] compared shoulders with and without infection undergoing revision shoulder arthroplasty. They found that synovial interleukin (IL)-6 demonstrated a sensitivity, specificity, and positive and negative likelihood ratios of 87%, 90%, and 8.45 and 0.15, respectively.

Most recently, Frangiamore et al[25] performed a broader analysis. Synovial IL-6, granulocyte-macrophage colony-stimulating factor, interferon-γ, IL-1β, IL-2, IL-8, and IL-10 were significantly elevated in revision cases identified as infected. IL-6, tumor necrosis factor-α, and IL-2 as a combination model displayed superior diagnostic ability than any cytokine alone, with a sensitivity of 0.80, a specificity of 0.93, PPV of 0.87, and NPV of 0.89.

α-Defensin is a microbicidal peptide regulated by proinflammatory cytokines which has recently been studied as a marker of infection. Frangiamore et al[26] examined this marker in patients undergoing shoulder arthroplasty and found that α-defensin showed a statistically significant difference between patients with infection (median, 3.2 signal to cutoff ratio) and those without infection (median, 0.21 signal to cutoff ratio) (P = 0.006). They reported an area under the curve, sensitivity, specificity, and positive and negative likelihood ratios of 0.78, 63%, 95%, and 12.1 and 0.38, respectively.

Synovial protein analysis is an emerging diagnostic tool, but the lack of a point of care test limits utility because it usually takes at least 24 hours to process. Future studies and innovation are required.

Prerevision Tissue Sampling

Dilisio et al[23] demonstrated an improved utility of arthroscopic tissue biopsy over glenohumeral joint aspiration in diagnosing PJI in painful arthroplasties before revision. In 19 patients without obvious signs of infection but at least one positive culture at revision surgery, prerevision arthroscopic biopsy yielded 100% sensitivity, specificity, PPV, and NPV, whereas fluoroscopically guided joint aspiration culture yielded a sensitivity of 16.7%, a specificity of 100%, positive predictive value of 100%, and negative predictive value of 58.3%. Tashjian et al[27] also assessed the accuracy of prerevision sampling with similar but less conclusive findings. The decision to perform prerevision tissue biopsy for culture should be weighed against the added risk and costs of an additional procedure.

Obtaining tissue cultures is likely most beneficial in the clinical scenario of a painful arthroplasty without overt signs of infection or in the arthroplasty with presumed aseptic loosening. This potentially allows the surgeon to identify a cause of the pain and also helps plan eventual treatment by providing a preoperative diagnosis if revision is to be performed.

Intraoperative Evaluation

In cases of an obvious infection, wound cultures are essential to identify the pathogen and to direct antimicrobial therapy. In cases without an obvious infection, intraoperative sampling is instrumental in establishing the diagnosis of PJI and guiding further treatment. The latter cases have been termed failed arthroplasties with positive cultures or unexpected positive cultures (UPCs).[8,32] In these cases, the preoperative evaluation has not clearly identified an infection; serum WBC, ESR, and CRP within normal limits; and normal aspiration and imaging.

Despite the obvious importance of the intraoperative assessment, the role and utility of routine cultures in all revisions remain unclear. If the surgeon has identified another obvious reason for failure (eg, typically component malposition, rotator cuff tearing, or fracture) and has elected to pursue revision surgery, controversy exists regarding the necessity of routine tissue cultures during all revision surgery and also regarding their interpretation. It is uncertain in this setting whether a positive culture is indicative of an infection, a contaminant, or a commensal to deep tissue.[10] In addition, consideration must be given to the fact that several studies have noted positive cultures of control specimens[12] (see further discussion later). The decision to obtain tissue cultures in this setting is left to the discretion of the surgeon.

Synovial Fluid Aspiration

Similar to preoperative arthrocentesis, the glenohumeral joint can be aspirated to obtain synovial fluid for analysis intraoperatively. Once the deltopectoral interval has been developed, the joint can be aspirated directly through the anterior capsule. The fluid can be analyzed for WBC, inflammatory markers (eg, IL-6 and α-defensin), and culture.

Frozen Section/Histopathology

Intraoperative samples for frozen section histopathology may provide additional information to assist in the diagnosis of PJI. Mirra et al[28] described the criterion for periprosthetic infection diagnosis as more than five neutrophils in five or more high-power microscopic fields. Unfortunately, this criterion was developed for the hip and knee and does not seem to be applicable to low-virulence organisms, such as C acnes. Less than 10% of patients with positive C acnes cultures have acute inflammation on intraoperative histology.[29,30] Grosso et al[31] proposed a shoulder-specific threshold for the diagnosis of PJIS as 10 or more total polymorphonuclear neutrophils in five high-power fields. This optimized threshold demonstrated a sensitivity of 72% for infection with 100% specificity. Owing to the lack of a clear cutoff and operator variability, the use of frozen sectioning is institution dependent and requires active coordination between the surgeon and pathologist. If histologic signs of acute inflammation are identified, a decision may be made to proceed with staged treatment or to aid in the interpretation of intraoperative culture results.

Tissue Culture

The results of intraoperative tissue cultures are considered the "benchmark" to establish the diagnosis of PJIS. Swab cultures have a lower yield than tissue culture, especially for isolating C acnes, which can be intracellular and encased in a biofilm. At least four different tissue samples should be obtained at revision surgery.[32] Using sterile instruments, the tissue should be sampled adjacent to the implants along with any other tissue exhibiting inflammatory changes. Although sonication of the explanted prosthesis may liberate biofilm-formative bacteria, the evidence regarding this technology in the shoulder suggests no additional benefit.[33]

One of the most challenging aspects of revision shoulder arthroplasty is determining the significance of a positive intraoperative culture. C acnes was historically considered a contaminant and continues to be a source of contamination in cultures of all types. Mook et al[34] reported that negative control specimens (eg, sterile gauze) had a 13% positive culture rate. Hudek et al[10] identified positive C acnes cultures in 36.4%of first-time shoulder surgery cases. Wong et al[12] reported the 33 of 94 shoulders (38%) undergoing primary arthroplasty had at least 1 positive culture, with 17 patients (19%) having ≥2 positive cultures. C acnes was the most common organism (67%).

Cultures of shoulder tissue should be held longer than the laboratory standard of 3 to 5 days to identify C. acnes. They should be held for at least 14 days on aerobic, anaerobic, and broth media.[32] Recent literature suggests that inoculation of tissue samples into blood culture bottles rather than on agar plates or in broth may yield faster and more sensitive results for PJIs.[35]

Although Pottinger et al[19] suggested holding cultures for up to 28 days, this may increase the possibility of false-positive results. Frangiamore et al[36] classified patients with positive tissue cultures as "probable true-positive" or "probable contaminant" and showed that the cultures of the former became positive in a shorter period, 5 days (range, 4 to 6 days) versus 9 days (range, 6 to 12 days) (P = 0.002), and "probable true-positives" grew more positive cultures (P = 0.001) and a higher percentage of positive cultures (P < 0.001).

Based on the hip and knee literature, perioperative antibiotics do not need to be withheld until intraoperative samples have been obtained.[37]

Polymerase Chain Reaction–Restriction Fragment Length Polymorphism

Investigators are developing diagnostics tests specific for C acnes. Holmes et al[38] developed a rapid test (less than 24 hours) for detecting C acnes in surgical biopsy specimens with as few as 10 bacterial cells using a polymerase chain reaction–restriction fragment length polymorphism assay. The goal is to develop a "point of care" assay, which can be used during surgery to make a rapid diagnosis.

Diagnostic Criteria

The Musculoskeletal Infection Society published criteria for hip and knee PJI (Table 2).[39] Hsu et al[40] demonstrated the lack of a standardized classification system for PJIS in a recent systematic review. Six of 22 studies (27%) did not include an explicit statement describing how an infection was defined, and 11 studies (55%) used an author-defined combination of clinical symptoms, laboratory tests, radiographic characteristics, findings on aspiration, and culture results. The lack of agreement on diagnostic criteria for a PJIS remains. Grosso et al[31] proposed an alternative set of criteria specifically for PJIS (Table 3).


Avariety of measures is used to prevent PJIS during index shoulder arthroplasty.[41] Nasal swab screening for methicillin-resistant Staphylococcus aureus (MRSA) colonization is a standard practice in patients undergoing hip and knee arthroplasty and may be undertaken before shoulder arthroplasty, although more research is needed to confirm its efficacy and cost effectiveness. If the culture is positive, decolonization is performed.

The use of chlorhexidine gluconate (CHG) cloths the night before and morning of surgery significantly reduces the culture rate of CNS; however, C acnes culture rates are not affected.[41] Alternatives to CHG include topical benzoyl peroxide 5% gel with or without clindamycin 1.2%.[41]

The surgical site should be prepared with an alcohol-based CHG solution which is more effective at eliminating overall bacteria burden and CNS than other commonly used solutions, such as an iodophor and alcohol preparation or an iodine scrub and paint preparation.[42] Cefazolin alone is used as prophylactic antibiotics within 1 hour of the start of surgery. In patients with penicillin allergy or high-risk/culture positive for MRSA, weight-based vancomycin is used instead or in addition to cefazolin. Antibiotics are usually continued for 24 hours postoperatively but are not required. Clindamycin may also be considered if there is a penicillin allergy.

Gloves should be changed after draping the patient. The skin knife blade should be discarded because skin preparation and IV antibiotics do not sterilize the deeper dermal layer.[43] Changing gloves before touching the implant is suggested because instruments and gloves have been cultured demonstrating contamination. Although Nowinski et al[44] reported a decreased infection ratewith antibiotic loaded cement, no high-level evidence exists to support this practice.

During the closure, 1 g of vancomycin powder may be placed in the surgical wound because it reduces surgical site infections.[41] In addition, Hatch et al[45] demonstrated that vancomycin is a cost-effective measure for the prevention of surgical site infections in patients undergoing shoulder arthroplasty. Alternatively, an injection of gentamicin at the end of the procedure significantly reduced the rate of PJIS in one retrospective study.[46] The data supporting the use of topical or injected antibiotics are limited and should not be a substitute for thorough irrigation of the surgical site before meticulous closure.


Surgery with adjuvant antibiotic therapy is the mainstay for the management of a PJIS. Nonsurgical management with chronic antibiotic suppression can be considered for patients who are unable to safely undergo surgical intervention or when the revision surgery entails more morbidity than the infection.

Débridement With Implant Retention

Irrigation and débridement (I&D) with retention of the implants is reserved for the management of selected acute PJI. Modular components should be removed and replaced. In patients with reverse shoulder arthroplasty, the polyethylene component may be exchanged during the procedure, although no specific evidence exists to support this. Literature on débridement with implant retention is limited, with no consensus on the definition of "acute." Dennison et al[47] reported on 10 cases of acute postoperative and delayed-onset acute hematogenous infections treated with I&D with component retention with a 70% long-term retention rate and moderate functional and pain outcomes. A systematic review of surgical treatments showed a persistent infection rate of 29.6%, the highest of all groups analyzed, and recommended against component retention.[48] The benefit is less morbidity and the possibility of superior functionality if the infection clears.

Single-stage Revision Versus Two-stage Revision

In two-stage revision management, the first stage involves removal of all implants along with a thorough I&D of the joint, periarticular tissue and removal of all cement, followed by placement of an antibiotic-impregnated cement spacer (Figure 2). At the second stage, a non-cemented component may be desirable secondary to the concerns of recurrent infection and need for further revision surgery, but stability is paramount and cementation (with or without antibiotics) may be required for solid fixation. The patient is treated with systemic IV antibiotics generally for at least 6 weeks, occasionally followed by oral antibiotics.

Figure 2.

AP (A) and lateral (B) radiographs of the left shoulder of an 88-year-old woman with an anatomic total shoulder arthroplasty with periprosthetic joint infection of the shoulder. There is lucency around the glenoid bone cement. The patient underwent two-stage revision. AP (C) and lateral (D) radiographs of the same patient after placement of antibiotic spacer. AP (E) and lateral (F) radiographs of the same patient after second-stage revision to reverse shoulder arthroplasty.

During the second stage, a new prosthesis is implanted. Zhang et al[49] performed an interval open biopsy before definitive implantation of a new final prosthesis to ensure infection eradication. Eighteen patients with PJIS demonstrated a persistent infection rate of 22% in all patients and 38%in those with a C acnes infection.

In some cases, patients may be satisfied with the antibiotic spacer and forgo the second stage. Grubhofer et al[50] found that 14 of 38 patients (37%) treated in a two-stage manner did not elect to have the second stage. In a systematic review, Nelson et al[6] found infection eradication rates for antibiotic spacer retention, one-stage revision, and two-stage revision of 90.3%, 91.7%, and 93.8%, respectively. Single-stage revision produced the highest mean Constant score.

Two-stage revision is the most common method used owing to the precedent set forth in the lower extremity.[6] However, in the shoulder, surgeons have found success in treating PJIS with a single-stage revision. In this case, all components are explanted, a thorough I&D is performed, and a new prosthetic implant is placed. The patient is placed on antibiotics specific to intraoperative culture sensitivities. Failure rates, defined as undergoing another surgery on the affected shoulder, have been reported as 4.0% to 8.3% for single-stage revision compared with 6% to 14% for two-stage revision, and functional outcomes appear to be comparable or in favor of single-stage revision.[6,48] Complication rates appear to be higher in two-stage revision cases. With a single-stage revision, the patient does not have to undergo the morbidity of an additional surgery or the dysfunction of living with an antibiotic spacer.[51]

More research is needed to determine when each approach is appropriate, but promising results have been observed for single-stage revision. However, comparative analysis of single- and two-stage revisions is fraught with the limitations of selection bias.

Unexpected Positive Cultures

There is little consensus on how to treat UPCs. Hsu et al[52] reported on 55 unmatched, single-stage revision shoulders without a preoperative diagnosis of the infection. Twenty-seven of the shoulders had ≥2 positive intraoperative cultures for C acnes (culture-positive group) and 28 had no or only one positive culture (the control group). Patients in the culture-positive group received 6 weeks of IV antibiotic, followed by at least 6 months of oral antibiotics. The functional outcomes and the rate of subsequent procedures for persistent pain or stiffness were comparable, if not better, for the culture-positive group, although 14 of the culture-positive group reported adverse effects to antibiotics. Furthermore, Padegimas et al[53] described a cohort of 117 revision shoulder arthroplasties performed for presumed noninfectious reasons, where 23.9% had at least one positive culture from revision surgery, of which 75% grew either C acnes or CNS, or both. The group with UPCs had a revision surgery rate comparable with that of the group without UPCs (7.1% versus 20.2%; P = 0.109). Further research is required to determine the relevance of UPCs.

Salvage Procedures

In elderly/low-demand or frail patients, resection arthroplasty is an option. Nelson et al[6] reported the rate of eradication of infection as 93.3%. The obvious downside is the limited function. Amputation is reserved for severe, life-threatening cases when the infection is uncontrollable.


C acnes infection can be treated with several intravenous antibiotics including penicillins, amoxicillin/clavulanic acid, ceftriaxone, and vancomycin. IV antibiotics are usually continued for 6 weeks. In a single-stage revision, the IV course is usually followed by a period of oral agents. In the case of UPCs, there is some evidence to suggest that patients may not need antibiotics. This remains a controversial subject. Grosso et al[54] demonstrated 5.9% recurrence of the infection (1 of 17) in cases of UPCs that were not treated with antibiotics after revision surgery. All but one of the cultures in the study were C acnes or CNS. The specific number of cultures taken for each patient was not reported. This study included patients with only one positive culture and no clinical signs of infection.

Oral rifampin may be added to help eradicate the biofilm. Antibiotic treatment regimens vary by institution and surgeon. Infectious disease specialists are commonly consulted. However, given the unique nature of periprosthetic infections about the shoulder, especially withC acnes, the shoulder surgeon may be more familiar than the general infectious disease physician and a close collaboration is needed. More investigation is required to determine for whom antibiotics are necessary and, if so, what specific agents and for what duration.