Optimized use and performance of culture for periprosthetic joint infection diagnosis: a comprehensive literature review

SUMMARY

Periprosthetic joint infection (PJI) is a significant complication of arthroplasty. Implicit in the diagnosis of PJI is the presence of organisms in what should be a sterile milieu. Given limits to culture-based PJI diagnosis, optimal utilization of culture in the diagnostic approach to PJI needs clarification. Culture is important in confirming a PJI diagnosis and characterizing the infecting microorganism(s) to inform management. As part of the development of a unified definition of PJI (by a group representing multiple scientific societies), a systematic review was performed to inform the role of culture in the definition of PJI. In addition, a comprehensive review on synovial fluid and periprosthetic tissue collection for culture, and associated culture methodology and results interpretation, was performed to inform best practices. Gaps in the literature were identified to develop research priorities.

INTRODUCTION

The diagnosis of periprosthetic joint infection (PJI) is not always straightforward; diagnostic criteria were initially established in 2011 (1), albeit based on limited evidence. Thereafter, multiple groups built on initial guidelines and incorporated evolving data on diagnostic test performance, leaving clinicians and investigators with multiple PJI diagnostic algorithms (1–4) that were not always concordant. In late 2023, an international panel consisting of representatives from the Musculoskeletal Infection Society (MSIS), the European Society of Clinical Microbiology and Infectious Diseases (ESCMID) Study Group for Implant-Associated Infections, the European Bone and Joint Infection Society (EBJIS), the International Consensus Meeting on Musculoskeletal Infection (ICM), and the Infectious Diseases Society of America (IDSA) met to develop unified criteria and an algorithm for PJI diagnosis. Eight working groups were convened to systematically review the literature on tests that might inform PJI diagnosis for a Unified PJI Definition; each group performed a systematic review using a standardized approach.

The Culture Working Group was tasked with systematically evaluating the literature to evaluate whether synovial fluid, sonicate fluid, and/or periprosthetic tissue culture should be included as diagnostic criteria for PJI. Given that PJI itself is defined by the presence of organisms in what should otherwise be a sterile milieu, culture is implicit in the diagnosis of PJI. It is therefore unsurprising that existing PJI diagnostic criteria include culture (Table 1). Culture not only contributes to defining the presence or absence of infection but also informs optimal antimicrobial management (based on the microorganisms detected and their antimicrobial susceptibility). Therefore, in addition to addressing the intuitive question as to whether culture should be included in the diagnostic criteria for PJI, the literature was surveyed to determine how to optimize synovial fluid and periprosthetic tissue culture performance and interpret results thereof. This manuscript therefore serves two roles. First, the systematic review of the Culture Working Group on behalf of the Unified PJI Definition is presented. Second, the comprehensive review is reported to address questions of optimizing the performance of synovial fluid and periprosthetic tissue culture for PJI diagnosis. Three of the authors (S.N., R.P., J.E.) comprised the Culture Working Group for the Unified PJI Definition and additionally conducted the comprehensive review; the fourth author (J.-J.A.-C.) contributed to the comprehensive review.

TABLE 1.

Position of culture in definitions of periprosthetic joint infection (PJI) across prior diagnostic guidelines

Organization (reference)	Year	Role of culture in PJI diagnosis
Musculoskeletal Infection Society (MSIS) (1)	2011	PJI is confirmed when a pathogen is isolated by culture from at least two separate tissue or fluid specimens obtained from the affected joint;a isolation of a microorganism in one culture of periprosthetic tissue or fluid is a minor criteriona
Infectious Diseases Society of America (IDSA) (2)	2013	Two or more intraoperative cultures or a combination of preoperative aspiration and intraoperative cultures that yield the same organism (indistinguishable based on common laboratory tests, including genus and species identification or results of antimicrobial susceptibility testing) may be considered definitive evidence of PJI; growth of a virulent microorganism (e.g., Staphylococcus aureus) in a single specimen of a tissue biopsy or synovial fluid may also represent PJI; one of multiple tissue cultures or a single aspiration culture that yields an organism that is a common contaminant (e.g., coagulase-negative staphylococci, Cutibacterium acnes) should not necessarily be considered evidence of definite PJI and should be evaluated in the context of other available evidence
International Consensus Meeting (ICM) (3)	2018	Two positive growths of the same organism using standard culture methods confirm PJI;b a single positive culture is a minor criterion assigned two points in a scoring system that determines the likelihood of PJI (infected, inconclusive, non-infected)b
European Bone and Joint Infection Society (EBJIS) (4)	2021	PJI is confirmed with two or more positive specimens with the same microorganism;c if no other standalone criteria are met, a single positive aspiration or intraoperative tissue or fluid culture may be considered along with at least one other suggestive diagnostic study to contribute to a “likely PJI” diagnosisc

^a PJI is also confirmed when there is a sinus tract communicating with the prosthesis. Other minor criteria include elevated erythrocyte sedimentation rate and C-reactive protein; elevated synovial fluid leukocyte count and synovial fluid polymorphonuclear neutrophil (PMN) percentage; presence of purulence in the joint; and >5 neutrophils per high-powered field (HPF) on histology. Four of six minor criteria are required to confirm a PJI diagnosis; thresholds are not clearly delineated.

^b Sinus tract with evidence of communication to the joint or visualization of the prosthesis also confirms PJI. Other factors afforded weight in the scoring system (different thresholds for acute versus chronic infection) include C-reactive protein; D-dimer; erythrocyte sedimentation rate; synovial fluid leukocyte count, positive leukocyte esterase, positive alpha defensin, and PMN percentage; histology; and the presence of intraoperative purulence.

^c PJI is also confirmed when any of the following are found: synovial fluid leukocyte >3,000 cells/µL, PMN percentage >80%, and positive alpha defensin; >50 colony-forming units (CFU) of any organism in sonicate fluid (contingent on the sonication method); ≥5 neutrophils in ≥5 HPF or visible microorganisms on histology. Other factors considered in the diagnostic rubric to contribute to likely infection include clinical features; C-reactive protein; synovial fluid leukocyte count, PMN percentage, and positive alpha defensin; >1 CFU of any organism in sonicate fluid; histology; and white blood cell scintigraphy.

SYSTEMATIC REVIEW

The Working Group was tasked with identifying manuscripts published during or after 2000 in which the performance of culture was evaluated against a reference standard, either any of the established diagnostic criteria (AAOS, MSIS, IDSA, ICM, EBJIS) (1–7) and/or clinical outcome. Importantly, as all the reference PJI diagnostic criteria include culture, only studies that excluded culture as part of the PJI diagnosis reference standard were included, to reliably evaluate the independent diagnostic accuracy of culture. This approach, while rigorous, limited the number of studies available for inclusion.

Methods

A literature search (Fig. 1) was conducted utilizing two databases (PubMed, EMBASE) to identify articles published between 1 January 2000 and 1 February 2024. Prospective and retrospective observational clinical studies, systematic reviews, and meta-analyses were included; case reports and narrative reviews were not included. Only studies evaluating the diagnostic accuracy of culture for PJI diagnosis in adults were included. Working Group members screened over 1,500 studies (Fig. 2); two independent reviewers screened each study for eligibility. Any article for which the reviewers disagreed on inclusion underwent full-text review by all three Working Group members; consensus was reached on all studies. Once studies not relevant to PJI and/or culture were excluded, most remaining studies were excluded because culture was included in the PJI diagnostic reference standard, preventing assessment of the independent value of culture as a test for PJI. Ultimately, four studies met the inclusion criteria and were considered in the systematic review (8–11). Details on the included studies are shown in Tables 2 and 3. Each included article was evaluated for bias (Fig. 3) using the QUADAS-2 tool and given a GRADE (Grading of Recommendations, Assessment, Development, and Evaluations) classification score.

Fig 1.

Search strategy used for the systematic review. PJI, periprosthetic joint infection. DAIR, debridement, antibiotics, and implant retention.

Fig 2.

PRISMA flow diagram for the systematic review. PJI, periprosthetic joint infection.

TABLE 2.

Characteristics of studies included in the systematic review^a

Study (author, year)	PJI reference standard	Number					Type of specimens studied
		Knees	Hips	Shoulders	PJIs	Controls
Minassian AM, 2014 (8)	IDSA 2013	21	50	3	79	243	Periprosthetic tissues
Peel TN, 2016 (9)	IDSA 2013	160	138	49	117	252	Periprosthetic tissues
Park K-H, 2016 (10)	IDSA 2013	21	33	0	54	0	Periprosthetic tissues and sonicate fluid
Yan Q, 2018 (11)	IDSA 2013	140	47	23	104	125	Periprosthetic tissues and sonicate fluid

^a PJI, periprosthetic joint infection; IDSA, Infectious Diseases Society of America.

TABLE 3.

Diagnostic test performance as measured in the studies included in the systematic review^a

Study (author, year)	Index test	Sensitivity			Specificity			PPV	NPV
		Sensitivity	Lower 95% CI	Upper 95% CI	Specificity	Lower 95% CI	Upper 95% CI
Minassian, 2014 (8)	All specimens	83.5	73.5	90.9	97.1	94.2	98.8	90.4	97.6
	All specimens	84.3	75	91.1	97.9	95.3	99.3		97.6
Peel, 2016 (9)	Agars and TG broth combined	62.6	51.7	72.5	98.1	96.1	99.3
	Aerobic and anaerobic bottles combined	92.1	84.9	97	99.7	98.7	100
	Aerobic BCB	90.2	65.5	91.1	96.3	94.8	98.6
	Anaerobic BCB	82	79.4	96.5	97.1	93.7	98.1
	Aerobic agars	59.4	45.3	72.5	99.5	98.3	100
	Anaerobic agars	32.2	20.8	45.7	99.5	98.3	100
	TG broth	74.8	61.5	85.8	99.4	98.1	99.9
	Combination all media	99.1	95.7	100	99.4	94.8	98.7
Park, 2016 (10)	Sonicate fluid	64.8
	Tissue	57.4
	Sonicate fluid− antibiotic+	77.8
	Tissue− antibiotic+	72.2
	Sonicate fluid− antibiotic−	58.3
	Tissue− antibiotic−	50.0
Yan, 2018 (11)	Tissue culture	86.3	78.3	92.4	99.6	97.7	100
	Sonicate fluid	88.7	81	94.3	99.6	97.7	100
	Tissue and sonicate fluid combined	99.1	95.7	100	99.5	97.6	100

^a Studies included in the systematic review met criteria outlined in the process for the Unified PJI Definition. See text and Fig. 2 for details on inclusion. CI, confidence interval; PPV, positive predictive value; NPV, negative predictive value; TG, thioglycolate; BCB, blood culture bottles; LCM, latent class model; IDSA, Infectious Diseases Society of America; MSIS, Musculoskeletal Infection Society.

Fig 3.

Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) evaluation of articles included in the systematic review. Panel A shows a summary of quality assessments according to QUADAS-2. Panel B shows a detailed quality assessment of each included study. Robin Patel was the senior author on the two 2016 manuscripts and the 2018 manuscript.

Results

Four studies reported sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of culture for diagnosis of PJI using IDSA criteria, with culture not included in the reference standard. Two studies documented chronicity of the included cases and were a mix of suspected “early” postoperative, acute hematogenous, and “chronic” PJI. All studies included specimens from hip and knee arthroplasty; three of the four studies also included specimens from shoulder arthroplasty. All included studies evaluated the performance of periprosthetic tissue culture; two also evaluated sonicate fluid culture. No studies meeting the inclusion criteria evaluated synovial fluid culture.

Based on the included studies, a positive culture result for the purpose of PJI diagnosis was defined as two positive periprosthetic tissue cultures with a phenotypically indistinguishable microorganism, or one each from synovial fluid and periprosthetic tissue with a phenotypically indistinguishable microorganism, or detection of a microorganism in sonicate fluid above a threshold that maximizes specificity. Specificity of a positive culture using any of these definitions was shown to exceed 96%, informing its role as a standalone test in the diagnosis of PJI. Because none of the included articles considered the diagnostic accuracy of a single positive synovial fluid or periprosthetic tissue culture, the reliability of a single positive synovial fluid or periprosthetic tissue culture could not be evaluated.

Delphi recommendations

For the Unified PJI Definition, the Culture Working Group recommended the following:

1. Two positive cultures with a phenotypically indistinguishable organism from periprosthetic tissue should be a standalone criterion for PJI diagnosis.

2. One positive culture from synovial fluid or sonicate fluid above a threshold that maximizes specificity, and one positive culture from periprosthetic tissue with a phenotypically indistinguishable organism should be a standalone criterion for PJI diagnosis.

3. A single positive synovial fluid, sonicate fluid, or periprosthetic tissue culture should be a diagnostic criterion, but not a standalone criterion, for PJI diagnosis.

4. Negative cultures do not exclude PJI.

Although not specifically evaluated or endorsed as part of the Unified PJI Definition process, intuitively, we assert that one positive culture from synovial fluid and another from sonicate fluid above a threshold that maximizes specificity, with a phenotypically indistinguishable organism, could also be considered a standalone criterion for PJI diagnosis.

COMPREHENSIVE REVIEW

Given that so few articles were evaluated as part of the systematic review and that many of the important questions faced are on optimization of culture performance in suspected PJI cases, a follow-on comprehensive review of the literature on synovial fluid and periprosthetic tissue culture was undertaken. A detailed review of sonication culture was not performed as part of the comprehensive review, as others have recently reviewed this literature (12–17). Articles identified through the systematic review literature search, including those not included in the primary analysis, were assessed. Additional articles from the author’s files and references were also included. For synovial fluid culture, accuracy in different joint types (e.g., hip, knee), reliability of a single positive culture, concordance with periprosthetic tissue culture, and the role of imaging guidance in sampling were considered. For periprosthetic tissue culture, performance in different joint types and the optimal number of periprosthetic tissue cultures to optimize diagnostic accuracy were considered. For both synovial fluid and periprosthetic tissue culture, ideal culture methodology practices (e.g., specimen volume, specimen processing, culture media types, duration of culture incubation) were assessed. Whether the type of organism identified (e.g., virulent pathogen or commensal skin microbiota) impacts the reliability of the result was assessed. Finally, whether antibiotic pre-treatment (preoperative therapeutic antimicrobials and/or intraoperative prophylaxis) impacts culture yield and accuracy, whether there should be an antibiotic holiday prior to culture collection (and if so, for how long), and whether certain culture methodologies perform better in the setting of antibiotic pretreatment were considered. Recommendations on culture interpretation and optimization resulting from this comprehensive review are shown in Tables 4 and 5.

TABLE 4.

Best practices for synovial fluid culture for PJI diagnosis

Process	Best practices
Specimen collection	Withhold antibiotics prior to synovial fluid collection for culture; an antibiotic holiday of at least 14 days is advised Collect synovial fluid for culture using aseptic technique; avoid aspiration through an open wound or sinus tract Note that higher volumes of synovial fluid improve culture yield; the ideal volume to optimize yield is unknown Do not use swabs If needed, collect synovial fluid with imaging-guided aspiration; ultrasound may be preferred to fluoroscopy
Culture processes	Culture synovial fluid in aerobic and anaerobic blood culture bottles, or if not possible, using aerobic and anaerobic agars and an anaerobic broth. Whether or not blood culture bottle supplements enhance culture yield is unclear Do not routinely perform Gram stain or fungal or mycobacterial stains/cultures on synovial fluid For optimal yield, incubate aerobic cultures for 5–7 days and anaerobic cultures for 14 days, noting that many pathogens will be identified earlier Do not routinely subculture negative blood culture broths
Interpretation	An organism that grows in a single culture may or may not represent the true pathogen causing PJI; a single positive culture for a virulent microorganism (e.g., Staphylococcus aureus) is more likely to represent PJI than a single positive culture for a microorganism that could be either a contaminant or pathogen (e.g., coagulase-negative Staphylococcus species, Corynebacterium species, Cutibacterium acnes); caution is warranted when there is no other evidence of PJI, especially with the recovery of C. acnes or other skin microbiota-type microorganisms Negative cultures do not exclude PJI Caution is warranted when interpreting negative cultures collected with lavage

TABLE 5.

Best practices for periprosthetic tissue culture for PJI diagnosis

Process	Best practices
Specimen collection	Withhold antibiotics prior to collection of periprosthetic tissue for culture; an antibiotic holiday of at least 14 days is advised Collect specimens for periprosthetic tissue culture using aseptic technique, ideally, with separate instruments; avoid collecting specimens from open wounds or a sinus tract When culturing in blood culture bottles (preferred), three separate tissue specimens should be collected; when culturing using agar plates and broth, four separate tissue specimens should be collected Do not use swabs
Culture processes	Because of imperfect concordance, collect operative periprosthetic tissue cultures even when preoperative synovial fluid culture has been performed Ideally, culture periprosthetic tissues in aerobic and anaerobic blood culture bottles, or if not possible, using aerobic and anaerobic agars and an anaerobic broth Do not routinely perform Gram stain or fungal or mycobacterial stains/cultures on periprosthetic tissues For optimal yield, incubate aerobic cultures for 5–7 days and anaerobic cultures for 14 days, noting that many pathogens will be identified earlier Do not routinely subculture negative blood culture bottles
Interpretation	Detection of a phenotypically indistinguishable organism in two separate specimens is diagnostic of PJI; the identified organism can usually be considered the cause of PJI An organism that grows in a single culture may or may not represent the true pathogen causing PJI; a single positive culture for a virulent microorganism (e.g., Staphylococcus aureus) is more likely to represent PJI than a single positive culture for a microorganism that could be either a contaminant or pathogen (e.g., coagulase-negative Staphylococcus species, Corynebacterium species, Cutibacterium acnes) Negative cultures do not exclude PJI

SYNOVIAL FLUID CULTURE

Based on the systematic review, the Culture Working Group recommends that synovial fluid culture be obtained preoperatively whenever PJI is considered, noting that a negative synovial fluid culture neither excludes PJI nor obviates surgical management when other evidence of PJI is present.

Fundamental to the Unified PJI Definition recommendation is the understanding that synovial fluid culture is neither perfectly sensitive nor perfectly specific. In a review of the literature from 1983 to 1999 on synovial fluid culture for PJI diagnosis (joint type not specified), sensitivity ranged from 4% to 100% and specificity from 88% to 100%, and it was commented that in “most studies no well-formulated reference standard for the diagnosis of [PJI] was used” (18). A review article published in 2012 by Corvec et al. listed the sensitivity and specificity of synovial fluid culture as 60% to 80% and 97%, respectively (19). In an article by Drago et al., sensitivity was listed as 42% to 90% (20). Enz et al. reported that synovial fluid culture (methods not provided) sensitivity and specificity for diagnosis of hip or knee PJI were 52% and 97%, respectively (21). Watanabe et al. recently reviewed 32 studies on synovial fluid culture sensitivity and specificity for PJI diagnosis, reporting values of 63% (95% confidence interval [CI] 56% to 70%) and 96% (95% CI, 93% to 98%), respectively (12). The limited sensitivity may relate to the imperfect classification of PJI itself, the array of microorganisms that cause PJI—some of which do not grow well in culture, the presence of PJI-causing organisms in sites other than the joint space (such as only “in the bone socket of the prosthesis” [21]), the presence of microbial aggregates in synovial fluid, the duration of infection (which can influence the type of microorganism present and its ease of culture), possibly the anatomical site of PJI (e.g., hip, knee, shoulder), patient exposure to systemic or local antimicrobial agents, whether saline injection and reaspiration were performed, the volume of synovial fluid cultured and/or methodologic variations in synovial fluid processing for culture and associated culture methods (22). The reason for the imperfect specificity is that synovial fluid aspiration is obtained by percutaneous sampling, risking specimen contamination by organisms on the skin surface, and may also be affected by contamination during work-up in the laboratory. That many microorganisms that cause PJI can be detected as pathogens or contaminants complicates the interpretation of positive culture results.

Variations in synovial fluid collection procedures (i.e., preanalytical aspects) and laboratory methods (i.e., analytical aspects) may occur. The former includes whether synovial fluid is aspirated preoperatively and/or intraoperatively (which may influence specificity and risk of skin microbiota contamination), and for preoperative aspiration differences in skin preparation and procedural draping, where the aspiration is performed and by whom, use (or not) of processes that discard the “skin plug,” and whether saline injection and re-aspiration is used for dry taps (see below). Other preanalytical considerations include synovial fluid volume collected, specimen collection containers employed, whether direct (i.e., procedural) inoculation of culture media is performed, and transportation times and conditions. Laboratory (analytical) aspects include synovial fluid volumes used for culture (a decision made in the laboratory based in part on what is submitted), specific solid and liquid (including blood culture bottles) media used and amount of synovial fluid inoculated to each, use of aerobic and anaerobic culture incubation conditions, whether or not terminal broth subcultures are included, and durations of incubation. Ultimately, interpretation as to what constitutes a positive culture also bears consideration. In most studies reviewed, many of these parameters were not clearly outlined, and methods are known to vary from institution to institution. Methods for microbial identification have improved over time, allowing more nuanced questions to be answered. In the past, for example, coagulase-negative staphylococci were typically reported as such, whereas today, specific species (e.g., Staphylococcus epidermidis) are commonly reported. Fundamentally, PJI is a collection of different diseases; for example, Staphylococcus aureus hip PJI could be considered its own entity, and questions about synovial fluid culture performance asked about it are separate from those asked about Cutibacterium acnes shoulder PJI, for example.

A final challenge is determining whether the organism(s) detected is the cause of PJI, not just whether a positive or negative result correlates with the presence or absence of PJI. Most studies focused on the latter, that is, addressing the overall sensitivity and specificity of synovial fluid culture for PJI diagnosis. Equally or more important is whether the detected organism is the cause of PJI, and whether all organisms causing infection have been detected. These questions are rarely addressed, perhaps because doing so is challenging. van Schaik et al. conducted a systematic review of concordance between preoperative synovial fluid culture and intraoperative tissue cultures in patients undergoing hip or knee revision surgery, analyzing seven studies comprising 1,677 patients (23). Depending on how concordance was defined in the studies, concordance rates varied between 45% and 79%, with six studies judged as having an unclear to high risk and one as having a low risk of bias (23). To this end, how best to determine whether what is detected by synovial fluid culture is clinically significant remains an incompletely answered question.

In the process of reviewing the literature in support of the Unified PJI Definition, a large volume of research on synovial fluid culture was reviewed, enabling exploration of questions surrounding synovial fluid culture. Acknowledging that the analytic performance of synovial fluid culture likely varies by anatomic location, specific questions were asked about hip, knee, and shoulder arthroplasty; additionally, interpretation of single positive synovial fluid cultures, organism concordance with operative cultures, methodologic laboratory considerations, and imaging-based aspiration were addressed.

What is the sensitivity and specificity of synovial fluid culture for hip PJI diagnosis?

In 2013, Yee et al. published a review on synovial fluid culture for hip PJI diagnosis, reporting sensitivities and specificities of 50% to 86% and 88% to 97%, respectively (24). In a meta-analysis published in 2013, Qu et al. reported the sensitivity and specificity of synovial fluid culture for hip PJI diagnosis to be 70% and 94%, respectively (25). In a third article published in 2013, Fink et al. reviewed the literature and reported sensitivities and specificities of synovial fluid culture for hip PJI ranging from 12% to 89% and 50% to 100%, respectively (22). Literature reviews address multiple studies that do not necessarily use the same criteria to diagnose PJI, likely contributing to broad ranges in reported results.

Fink et al. reported on 100 hip arthroplasty aspirations performed under sterile conditions in the operating room with the patient under general anesthesia under image intensifier control, with 10 mL of normal saline injected and aspirated for dry taps (see below). Fluid was injected into BD BACTEC-PEDS-PLUS/F-Medium (Becton Dickinson) blood culture bottles, treated with Fastidious Organism Supplement (FOS) (Becton Dickinson), and incubated using the BD BACTEC 9050 automatic blood culture system (Becton Dickinson) for 14 days; synovial fluid culture had a sensitivity of 64% (95% CI, ±14%), and specificity of 96% (95% CI, ±5%) (22). In this study, “diagnosis obtained from the revision surgery samples was regarded as the definitive result with respect to” PJI. Barker et al. reported a retrospective cohort study of patients whose prosthetic hips were aspirated under fluoroscopic guidance, with PJI “confirmed by biopsy,” and reported a sensitivity of 59% (CI, 35%–82%) and specificity of 94% (CI, 89%–98%) (26). Mean aspiration volume for infected and non-infected joints was 6 mL (2–36 mL) and 11 mL (1–200 mL), respectively (26). When unable to obtain synovial fluid, joints were injected with saline and reaspirated; this was considered not to yield increased falsely positive results in this study (26).

What is the sensitivity and specificity of synovial fluid culture for knee PJI diagnosis?

In a 2013 review by Yee et al. on synovial fluid culture for knee PJI diagnosis, sensitivities and specificities of 65% to 72% and 95% to 96%, respectively, were reported (24). In the 2013 meta-analysis by Qu et al., the sensitivity and specificity of synovial fluid culture for PJI diagnosis were 78% and 96% for knee arthroplasty (25). As for hip arthroplasties, these literature reviews address multiple studies that themselves may not use the same criteria to diagnose PJI.

What is the sensitivity and specificity of synovial fluid culture for shoulder PJI diagnosis?

Shoulder aspiration can be challenging in terms of recovering enough fluid for culture (27), with attempted aspirations frequently yielding dry taps (28). The predictive value of preoperative joint aspiration for shoulder PJI diagnosis is incompletely defined, leading some to suggest the existence of “limited evidence to support routine preoperative aspiration during the workup of a suspected shoulder PJI” (28). In a study of 35 patients with shoulder PJI classified by IDSA criteria, requiring a minimum volume of 1 mL, with synovial fluid inoculated into aerobic (BacT/ALERT FA Plus) and anaerobic (BacT/ALERT FN Plus) blood culture bottles and incubated for 14 days, pre-operative aspiration did not provide required minimum fluid volume in 11/35 cases and only 42% of the remaining 24 cases had positive cultures (29). Zanna et al. recently reported the sensitivity of synovial fluid culture for microbial detection in 50 patients with shoulder PJI defined by modified MSIS criteria (7, 30). Specimens were inoculated into thioglycolate broth and incubated for 14 days, after which (or at the time of detection of bacterial growth), subculture to Columbia blood plates was performed and incubated under aerobic and anaerobic conditions (30). The sensitivity of synovial fluid culture was 56% for shoulder PJI. Preoperative cultures yielding C. acnes demonstrated sensitivity, specificity, PPV, and NPV lower than 80% (30). Those yielding Gram-negative pathogens demonstrated the highest sensitivity and specificity (both 100%), with polymicrobial infections showing the lowest sensitivity (30). Overall, Zanna et al. concluded that preoperative synovial fluid aspiration for shoulder PJI poorly predicts intraoperative culture results, albeit with more favorable concordance for monomicrobial preoperative cultures, particularly those yielding Gram-negative bacteria and S. aureus (30).

Are single positive synovial fluid cultures diagnostic of PJI for certain organism types (and if so, which organisms)?

According to the Unified PJI Definition, a single positive preoperative synovial fluid culture, especially for a potential contaminant, is not diagnostic of PJI; in the absence of other criteria for PJI, this should not inform the need for surgery. According to Trebse and Roskar, “One positive synovial fluid culture is insufficient to diagnose PJI” but if, “the second culture is positive for the same organism, the specificity is high enough to confirm the diagnosis” and the “combination of one positive synovial fluid culture and a positive cell count or differential is also diagnostic” (31). That said, detection of S. aureus is likely to represent a pathogen (32).

Some have recommended repeat joint aspiration when the workup of PJI reveals conflicting data (33). Accordingly, Hassebrock et al. evaluated the correlation between two fluoroscopically guided hip aspirations in 60 patients classified by MSIS criteria with less than 90 days between aspirations and no intervening surgery or antibiotics (33). Culture methods were not detailed. The first aspiration yielded positive cultures in 23 patients, negative cultures in 30, and seven in whom fluid volume was insufficient to perform culture; repeat aspiration changed culture results in 25 cases (42%) (33). Two of 30 initially culture-negative aspirations were positive on the second aspiration (33). When a culture was initially negative with low clinical suspicion of infection, no second synovial fluid culture was positive (33). The single patient with a low clinical suspicion of infection with two positive cultures was reaspirated due to an unexpected positive first culture, which grew C. acnes, with the repeat aspiration revealing S. epidermidis, also considered a contaminant (33). The authors concluded that repeat aspiration is useful in patients with conflicting clinical data and a history or high clinical suspicion of PJI (33).

A single positive synovial fluid culture for S. aureus is more likely to represent PJI than a single positive culture for a microorganism that could be either a contaminant or a pathogen (e.g., coagulase-negative Staphylococcus species, Corynebacterium species, Cutibacterium species). Single positive synovial fluid cultures are generally insufficient to establish a diagnosis of PJI by themselves; detection of a microorganism that could be a contaminant or a pathogen in a single synovial fluid (or periprosthetic tissue) culture may or may not represent the pathogen causing PJI in cases of PJI. Therefore, tissues should be collected for aerobic and anaerobic culture, even when synovial fluid culture is positive, to confirm the microbiology; this will also maximize the chance of identifying polymicrobial PJI (15, 32).

What is the concordance between microorganisms identified through preoperative synovial fluid sampling and intraoperative cultures?

While the presence of microorganisms in an otherwise sterile location is fundamental to PJI diagnosis, knowing the infecting microorganism’s identity and antimicrobial susceptibility is needed for idealized PJI management. While most studies focus on the sensitivity and specificity of synovial fluid culture for the diagnosis of PJI, clinicians managing PJI are equally interested in understanding whether the detected microorganism(s) are the true and only cause of PJI.

Multiple studies have attempted to identify the concordance between preoperative synovial fluid and intraoperative (synovial fluid and periprosthetic tissue) cultures. The concordance literature is challenged by a lack of consistent definitions and a lack of standard comparator definitions (denominators). Studies reviewed considered that if the same organism grew in preoperative synovial fluid culture and intraoperative culture with no new organisms recovered, this would be considered a “concordant” result. However, growth of additional organisms in operative specimens (polymicrobial infection) was variably considered; some studies included these as concordant if the same organism grew in operative cultures regardless of additional growth, while others considered these “partially concordant” or “discordant.” Most studies considered only identification of the organism (at the genus or species level) in determining concordance, while some also considered antimicrobial susceptibility of the organism, classifying only those that had both an identical species and antimicrobial susceptibility as concordant. Concordance was not uniformly calculated using the same denominator—some investigators measured the portion of concordant specimens against only those with PJI who also had positive intraoperative cultures, and others against all patients with PJI, with yet others, including all patients undergoing aspiration with or without confirmed PJI. Some included negative preoperative synovial fluid cultures in the numerator in determining overall concordance, whereas others included only those in which a microorganism was identified. Handling of “false positive” preoperative synovial fluid cultures was not uniform—in some studies, preoperative synovial fluid cultures were considered falsely positive if the organism did not grow in operative culture regardless of the presence of PJI; others only considered these to be falsely positive in the absence of PJI. Some, but not all, studies considered antibiotic pretreatment prior to culture collection; several excluded patients who had received antibiotics in the 14 days prior to specimen collection. Importantly, in the absence of other confirmatory information, interpretation of the accuracy of a positive preoperative synovial fluid culture when operative cultures are negative (and in the absence of another explanation, such as intercurrent antibiotic exposure) can be challenging. Some studies excluded this scenario by evaluating concordance through the lens of only those who had positive operative tissue cultures. For these reasons, understanding true “concordance” is challenging.

For the purposes of this review, the perspective of the clinician was considered in defining concordance, specifically, ‘When an organism is identified preoperatively in a patient ultimately diagnosed with PJI, is the organism identified preoperatively the “correct” and the only pathogen causing a patient’s infection?’ For this, studies that considered only those that had positive preoperative aspiration cultures and were ultimately diagnosed with PJI (or from which these data points could be obtained through the publication) were considered. Using this definition, overall concordance in patients with PJI who had positive preoperative cultures with the same organism (and with no additional organisms identified) varied between 44% and 84%, compared to operative cultures (32, 34–39). Identification of additional microorganisms in operative cultures happened with variable frequency, occurring 3%–29% of the time (30, 32, 34–41). When considering causes of discordance, growth of additional microorganisms in culture was more common than identification of a different microorganism in culture, in most studies (32, 34, 35, 37, 38, 40, 41), although negative operative cultures were more common than both in studies that also included patients with negative operative cultures (32).

Studies evaluating concordance in general included too few patients to assess predictors of concordance. Most studies reported concordance in patients with infected hip or knee arthroplasty; two (30, 37) considered only patients with shoulder PJI, in whom concordance was 62% and 80%. Three studies compared concordance between different joint types and found no difference in concordance between hips, knees, and/or shoulders (32, 35, 42). Concordance by microorganism was difficult to ascertain and not significant, although in most studies, concordance was numerically lower with Staphylococcus species and Cutibacterium species than with Gram-negative bacteria (30, 42). Interestingly, two studies suggested that concordance was higher when less time elapsed between preoperative aspiration and operative specimen collection for culture (32, 36). In a study by Boyle et al., of 363 patients with hip or knee PJI who had positive preoperative aspiration cultures, the mean number of days from synovial fluid aspiration to intraoperative tissue culture collection was longer in the discordant group (14 ± 17.6 versus 20.0 ± 20.4 days; P = 0.016).

In summary, concordance between organisms identified in preoperative synovial fluid culture and intraoperative cultures is imperfect; even when the same microorganism is identified in operative cultures, a significant minority of patients with PJI will have additional organisms identified. The organism identified preoperatively may not be the correct or only pathogen; clinicians should not rely exclusively on preoperative culture in determining management strategies.

How should synovial fluid culture be performed for PJI diagnosis?

This question focuses on laboratory-based aspects of synovial fluid culture, excluding preanalytical considerations. Synovial fluid volumes submitted to the laboratory are often insufficient for all the requested tests. Minimum volumes required for culture vary from laboratory to laboratory, and specimen volume submitted to culture (and distribution to the media used) is rarely documented on an individual basis. Rockov et al. reviewed 294 revision hip and knee arthroplasty procedures and identified 45 cases (15%) with discordant synovial fluid and operative culture results (43). The media inoculated was dependent on the volume of synovial fluid sent to the microbiology laboratory. Volumes ≤2 mL were limited to broth cultures with drops of specimen placed on solid media. Larger volumes allowed for additional placement of fluid into blood culture bottles. The mean aspiration volume for concordant cases was higher than for discordant cases (19.1 versus 10.2 mL, P = 0.02) (43). The proportion of anaerobic bacteria and fungi was greater among discordant compared to concordant cases [52% for discordant cases versus 8% for concordant cases (P < 0.001)] (43). The optimal cutoff value for predicting concordant cultures was 3.5 mL for “typical” bacteria and 12.5 mL for anaerobic bacteria and fungi (43). Aspiration cultures are more likely to correlate with intraoperative cultures with higher aspiration volumes, and the optimal aspiration volume is higher for anaerobic bacteria and fungi (43). Providers should be cautious when a small volume of aspirated fluid yields a negative culture, particularly when an anaerobic bacterium is suspected, or when there is a high pre-test suspicion of PJI (43). Ideally, minimum synovial fluid volumes for culture should be established, and any testing performed using less than minimal volumes should be clearly communicated in laboratory reports. That said, there is unfortunately no established minimum synovial volume for culture; across studies, the amount of synovial fluid required for culture per institutional policy has varied (33, 44). Hassebrock et al. established 10 mL as a standard required volume for culture (33). Establishing such volumes is important to guide proceduralists in obtaining specimens to collect necessary volumes for culture and all other tests being requested, including those that may be added on downstream.

Bacterial aggregates can limit culture sensitivity, especially if small volumes are cultured. If plate cultures are exclusively used for synovial fluid culture, pre-treatment of hip and knee PJI synovial fluid with dithiothreitol may increase bacterial counts and microbiological yield (20); exclusive use of plate culture is not, however, recommended.

In 2013, at the ICM, delegates suggested that cultures be incubated for 5 to 14 days; in 2018, the same group recommended that cultures be incubated for 5 to 7 days, and when PJI with low-virulence organisms is suspected, or if preoperative cultures are negative but there is a high clinical suspicion of infection, for up to 21 days. While a 5-day incubation is sufficient for isolation of aerobic bacteria, some anaerobic bacteria, such as C. acnes, need prolonged (e.g., 14-day) anaerobic incubation for recovery, using either an anaerobic broth or ideally (see below) an anaerobic blood culture bottle (15, 19).

Ideally, synovial fluid should be cultured in aerobic and anaerobic blood culture bottles (15). This allows detection of fastidious or slow-growing organisms, detection of more pathogens and fewer contaminants, and improved sensitivity (24, 45). In 2001, Hughes et al. reported that synovial fluid (not limited to prosthetic joint-associated fluid) inoculated into pediatric blood culture bottles detected more pathogens and fewer contaminants (possibly from less manual handling) than conventional methods (46). In an early small study of operative inoculation of synovial fluid from hip or knee arthroplasties into blood culture bottles (BACTEC 9240, aerobic and anaerobic, 6-day incubation), 92% (23/25) sensitivity and 100% (0/9) specificity were reported (47). Jordan et al. reported a retrospective case series of 219 patients undergoing revision hip or knee surgery, in whom synovial fluid was intraoperatively aspirated and cultured in aerobic and anaerobic blood culture bottles, as well as submitted in a sterile universal container (48). Detailed culture methods were not described. Synovial fluid culture in blood culture bottles was associated with a sensitivity and specificity for PJI diagnosis (defined using MSIS criteria) of 85% and 100%, respectively, whereas synovial fluid submitted in a sterile container yielded a sensitivity and specificity of 26% and 99%, respectively (48). Kuo et al. reported on synovial fluid from 77 patients (41 hips and 39 knees) who met ICM criteria for PJI, whose cultures were performed by inoculation into BD BACTEC Plus Aerobic blood culture bottles (BD Diagnostics Systems) and incubated using an automated blood culture system (BACTEC FX; BD Diagnostics Systems) and a conventional method (referred to as “swabs”); the two methods yielded positive results in 80% and 59% of cases (P = 0.002) (49). Barker et al. performed synovial fluid cultures from hip arthroplasties using blood culture bottles (Becton Dickinson, incubated for 10–14 days) or collected in “universal containers” and plated to aerobic and anaerobic plates with 3- and 5-day incubations, respectively, and enrichment broth incubated for 5 days. False positive and false negative rates were numerically higher with the latter, leading the authors to recommend blood culture bottles for cultures (26). Font-Vizcarra et al. studied 87 patients with knee or hip PJI and 63 with aseptic loosening (defined using “clinical, radiographic and histologic parameters”); synovial fluid was aspirated and 1–3 mL inoculated into each of an aerobic and anaerobic blood culture bottle (BACTEC 9240 system; BD Diagnostic Systems) and incubated on a BACTEC 9240 system for 5 days (50). Synovial fluid culture had a sensitivity and specificity of 86% and 100%, respectively. Sensitivity was higher (91%) for infections in the first 6 weeks after arthroplasty than for those presenting thereafter (79%) (50). Despite the high specificity reported, another study using the same microbiologic methods and by the same group raises the possibility of falsely positive results; 10% of synovial fluid specimens collected from 402 elective primary total hip arthroplasties yielded positive cultures (with the most frequently isolated microorganism being coagulase-negative staphylococci [32/41]) (51). It is not recommended to blindly subculture negative blood culture broths at the conclusion of incubation (or at any time along the way). The optimal choice of blood culture medium and whether supplementation (e.g., FOS) is needed remains undefined (45). BacT/ALERT has a United States regulatory claim for culturing body fluids. Ideally, purpose-designed/validated aerobic and anaerobic culture bottles for PJI culture would be welcomed.

If blood culture bottles are not used, anaerobic enrichment broths are recommended for inclusion in the culture process. Whether such broths should be blindly subcultured to solid media “at the end of incubation, even if the broth is transparent” is controversial (but recommended by some [19]); this practice is, however, tedious for the laboratory and risks introduction of contaminants.

Renz et al. reported on synovial fluid culture sensitivity for Cutibacterium PJI (52). One milliliter of synovial fluid was inoculated into a pediatric blood culture bottle (BACTEC PedsPlus/F) and incubated for 14 days, and 0.1 mL was inoculated on aerobic and anaerobic sheep blood agar plates and incubated 7 days aerobically and 14 days anaerobically, with the remaining fluid inoculated into thioglycolate broth. Sensitivity of synovial fluid culture for Cutibacterium PJI diagnosis (including 30 hip, 19 shoulder, 12 knee, and one elbow prosthesis) was 49% (52). In this study, PJI was defined as growth of Cutibacterium species in at least two periprosthetic tissue cultures or with sonication of the removed implant (>50 cfu/mL); if Cutibacterium species grew only in synovial fluid, at least one non-microbiological criterion was needed to confirm infection. The sensitivity of synovial fluid culture is lowered with administration of antecedent antibiotics, with the odds of having a culture-negative specimen increasing several-fold if the patient has received antibiotics within the preceding 3 months (24). Antibiotics should be stopped for a minimum of 2 weeks before joint aspiration (24), although the ideal antibiotic-free window is not well-defined.

Li et al. studied 286 hip or knee aspirations performed without imaging guidance, where, if >1 mL of synovial fluid was obtained, fluid was directly cultured, but if ≤1 mL was aspirated, 10 mL of saline solution was injected and the joint reaspirated (53). Fluids were injected into BacT/ALERT FA FAN (Fastidious Antimicrobial Neutralization) (bioMérieux) and BacT/ALERT PF Pediatric FAN (bioMérieux) bottles and incubated for 14 days, with PJI diagnosis made using modified MSIS criteria (53). Saline lavage and re-aspiration were used in 82 cases (47 PJI cases and 35 non-PJI cases), and direct aspiration in 204 cases (99 PJI cases and 105 non-PJI cases) (53). The saline solution lavage rate was 15% (21 of 140), and 42% (61 of 146) for knees and hips, respectively (53). Sensitivity and specificity were 80% (95% CI, 72%–86%) and 96% (95% CI, 91%–98%), including 77% and 99%, for non-lavage cases, and 85% and 86%, for dry tap cases, respectively (53). Interestingly, in false-positive cases (mostly coagulase-negative staphylococci), bacteria were only identified in 1 of 2 bottles, whereas in true-positive cases, 96% of specimens yielding coagulase-negative staphylococci (48 of 50) were positive in both bottles (53). Heckmann et al. evaluated the effect of saline lavage on synovial fluid culture in patients undergoing hip or knee revision; intraoperatively, prior to arthrotomy, the maximum amount of fluid possible was aspirated (to simulate a dry tap, pre-lavage) followed by injection of 20 mL of normal saline and re-aspiration (post-lavage) (54). Among the 78 patients with data available for analysis, 16 patients met modified MSIS criteria for PJI; of ten positive pre-lavage fluid cultures, only six remained positive post-lavage (all for the same microorganism) (54). These results, together with those of Christensen et al. (55), suggest that when considering potential PJI, scrutiny should be applied when interpreting negative cultures in those who require saline lavage for synovial fluid collection.

In summary, idealized volumes for synovial fluid culture are not defined, but should be in future studies. Aerobic and anaerobic blood culture bottles should be used for culture (without terminal subculture), the duration of anaerobic culture incubation should be extended (e.g., 14 days) in cases where C. acnes may be a pathogen, and overall, methods are more established for hip and knee PJI than for PJI involving other joint types, including shoulder PJI.

Should synovial fluid collected for culture be aspirated with imaging guidance?

Imaging-guided aspiration (e.g., using ultrasound, fluoroscopy, computed tomography) is more typically used for suspected hip and shoulder than knee PJI. In one-third to one-half of hip aspirations, for example, little to no fluid is aspirated (defined as <0.5 or <1 mL, depending on the study), referred to as a “dry tap” (55). Sterile saline may be injected into the joint space and reaspirated to attempt to obtain fluid (55). Rates of dry taps may be higher with non-imaging-guided aspirations and in patients with high body mass index (55).

Partridge et al. performed a retrospective analysis of 580 hip and knee synovial fluid aspirations in patients with moderate to high risk of infection, who subsequently proceeded to revision arthroplasty (40). PJI was defined as being present if ≥5 specimens were sent to the laboratory and ≥3 grew the same microorganism with the same antimicrobial susceptibility profile, or when only three or four tissue specimens were submitted and ≥2 grew the same microorganism with the same antimicrobial susceptibility profile. Fluid was aspirated under fluoroscopic guidance primarily in 313 (54%) cases and after saline injection and re-aspiration in 267 (46%) (40). Synovial fluid was inoculated into aerobic and anaerobic BACTEC blood culture bottles and incubated on the BD BACTEC 9240 instrument for 7 days (40). If culture bottles showed no growth at 7 days, terminal subcultures to non-selective media were performed (40). If less than 5 mL of fluid was available, only an aerobic bottle was inoculated (40). Overall sensitivity and specificity were 84% (95% CI, 78% to 89%) and 85% (95% CI, 81% to 88%), respectively (40). Sensitivity and specificity of saline injection and re-aspiration after dry tap were 87% (95% CI, 79% to 92%) and 79% (95% CI, 72% to 84%), compared with 81% (95% CI, 71% to 88%) and 90% (95% CI, 85% to 93%) for direct aspiration (40). In six cases, synovial fluid was positive for what was considered an “incorrect organism,” and in 18 cases, multiple organisms were isolated from tissues but only one from synovial fluid (40).

Christensen et al. reviewed patients with suspected hip PJI defined using MSIS criteria who underwent imaging-guided hip aspiration and compared the sensitivity and specificity of synovial fluid culture between those with dry and successful taps (55). The culture methodology was not described. A greater proportion of patients with successful compared to unsuccessful taps met MSIS PJI criteria preoperatively (30% versus 8%, P < 0.001). There was no difference in concordance between preoperative and intraoperative culture between the groups (68% versus 66%, P = 0.71), although the dry tap group had a higher rate of negative preoperative cultures followed by positive intraoperative cultures (86% versus 41%, P = 0.047) (34). Among the most common discordant cultures in the successful tap cohort were S. aureus detected preoperatively but not intraoperatively, while among the most common discordant cultures in the dry tap cohort were negative S. aureus preoperatively but positive intraoperatively.

Use of ultrasound potentially reduces the need for special procedure rooms, exposure of patients and staff to radiation, and expense, and may result in greater patient satisfaction with less pain compared to fluoroscopy (56, 57). Duck et al. reported 349 ultrasound-guided hip aspirations, reporting mean and median aspiration fluid volumes of 17 and 4 mL, respectively (not including the 9% requiring lavage), an overall culture contamination rate of 2%, matching aspiration and surgical culture results in 81% of cases, and 100% and 91% specificity for cultures obtained with lavage and bloody aspirates, respectively; in this study, PJI was classified using ICM criteria (56).

In summary, caution is warranted when interpreting negative cultures collected with saline lavage. Collection of synovial fluid with imaging-guided aspiration appears to be acceptable, with some evidence for ultrasound being preferred to fluoroscopy; since imaging is often used when there is a high probability of a dry tap, caveats noted in the prior section about applying scrutiny when interpreting negative cultures in patients who require saline lavage for synovial fluid collection should be applied.

PERIPROSTHETIC TISSUE CULTURE

Culture of periprosthetic tissue is key for the microbiological diagnosis of PJI. PJI diagnostic guidelines from different societies and consensus groups have all included periprosthetic tissue cultures as major criteria (Table 1) (1, 3–6, 58). In the new Unified PJI Definition, culture of synovial fluid, periprosthetic tissue, and/or fluid derived from sonication of explanted devices (i.e., sonicate fluid) are included—two positive cultures from periprosthetic tissue for a phenotypically indistinguishable organism, or one each from periprosthetic tissue and either synovial fluid or sonicate fluid, are standalone criteria for PJI diagnosis. Implicit in the definition of PJI and substantiated by years of research, positive periprosthetic tissue cultures, according to pre-established criteria (59), are considered diagnostic of PJI. However, what constitutes a positive culture and how best to sample and process periprosthetic tissues for culture require additional consideration. What should be considered a positive periprosthetic tissue culture? How should periprosthetic tissue cultures be performed? Is it necessary to include liquid media, or are solid media sufficient? Should tissue culture in blood culture bottles be performed? What is the optimal methodology for clinical microbiology laboratories to process tissues and perform and workup culture? These and many other questions have been an active research topic since Atkins et al. described the value of periprosthetic tissue culture for PJI diagnosis (59).

What is the sensitivity and specificity of periprosthetic tissue culture, and does this differ according to the affected joint?

For hip prostheses, Spangehl et al. reported a sensitivity, specificity, PPV, and NPV of 94%, 97%, 77%, and 99%, respectively, for periprosthetic tissue cultures from 178 hip prostheses (of which 35 were infected according to author-defined criteria) (60). Müller et al. demonstrated sensitivity, specificity, PPV, and NPV values of 78%, 92%, 96%, and 63%, respectively, among 50 patients (37 with PJI, classified using author-defined criteria) (61).

Mikkelsen et al. evaluated 120 knee arthroplasty revisions among 118 patients (26 with PJI classified using author-defined criteria, of which 10 were “early onset” [within 4 weeks] and 16 “late onset”) (62). The sensitivity of periprosthetic tissue cultures (using Kamme and Lindberg interpretative criteria [63]) was 46%, specificity 100%, PPV 100%, and NPV 81% (62). Baré et al. obtained similar results, with a sensitivity of 53%, specificity of 94%, PPV of 85%, and NPV of 83% for intraoperative tissue culture in an evaluation of 79 PJIs classified using Spangehl interpretive criteria (60) among 295 patients with knee arthroplasties (64).

Most studies reported results of hip and knee prostheses together. Aggarwal et al. reported a sensitivity of 93%, specificity of 98%, PPV of 93%, and NPV of 98% for tissue cultures among 117 cases (74 hip and 43 knee; 30 infected by MSIS criteria) (65). This study also demonstrated a low sensitivity and specificity of surgically obtained swabs for PJI diagnosis (65).

Many studies have evaluated the sensitivity and specificity of periprosthetic tissue biopsy cultures compared to sonication cultures, and in these studies, results for conventional periprosthetic biopsy cultures are reported (13, 31, 66). As an example, Trampuz et al., using author-defined criteria, reported a sensitivity of tissue cultures of 61% and specificity of 99% among 331 patients undergoing implant explantation (207 knees and 124 hips; 79 infected) (67).

Data on the performance of tissue culture in diagnosing shoulder PJI are distinct from data of hips and knees, in part due to differences in microbial etiology of shoulder PJI compared to hip and knee PJI, and because of the challenge in interpreting the detection of C. acnes (68, 69) or coagulase-negative Staphylococcus species in such cases (70). In the study by Zanna et al., only two culture-negative cases (sensitivity 96%) were found among 50 infected patients (defined by 2018 ICM criteria) (30). Diagnostic approaches, including the number of cultures needed to diagnose PJI, are less well-defined for shoulder compared to hip and knee PJI. Patel et al. assessed sensitivity and specificity in shoulder PJI based on ICM criteria (71); in this study, the highest specificity was obtained for a single positive culture yielding a virulent organism or two positive cultures with an identical low virulence organism (100% in both cases), although with lower sensitivity (20% and 50%, respectively). In this study, a single positive culture with a low virulence organism had a sensitivity of 55%, but specificity was only 83% (71).

Several studies involving different joint types have also evaluated the usefulness of preoperative biopsy cultures for PJI diagnosis. Ottink et al., using IDSA criteria for the definition of infection, compared preoperative and intraoperative biopsy cultures; intraoperative tissue biopsy (in 29 patients) had a sensitivity of 82%, specificity of 100%, PPV 100%, and NPV 90%, while preoperative biopsy cultures (16 patients) were less accurate, demonstrating 33% sensitivity, 85% specificity, 33% PPV, and 85% NPV (72). Eisler et al. studied preoperative ultrasound-guided capsule biopsy cultures, reporting a sensitivity of 67% and specificity of 68% among 72 hip prostheses (nine infected according to author-defined criteria) (73). In this study, biopsies were taken from patients with suspected septic loosening and considered definitively infected if two out of two cultures yielded the same bacteria (73). Finally, a more recent study by Simon et al. reported a sensitivity of 80% and specificity of 80% for culture specimens obtained using open incisional biopsy in 32 patients (20 infections defined with a combination of MSIS and author-defined criteria) (74). Fuerst et al. analyzed the role of preoperative biopsies in 69 non-infected and 17 infected knee arthroplasties (using criteria described by Atkins et al.) (59), reporting a sensitivity of 100%, specificity of 95%, PPV of 87%, and NPV of 100% (75). Niemann et al. evaluated open incisional biopsy cultures among 38 patients with confirmed PJI defined by MSIS criteria (10 hips, 28 knees). Compared to surgically obtained specimens, open incisional biopsy was 75% sensitive and 67% specific with a PPV of 60% and NPV of 80% for hip PJI. For knees, the same parameters were 63%, 95%, 83%, and 87%, respectively (76). Open biopsy cultures were also evaluated by Klaber et al. in 126 patients (48 infected according to modified MSIS criteria); sensitivity and specificity were 69% and 89%, respectively (77). Prujin et al. evaluated biopsies obtained preoperatively from shoulders compared to surgically obtained tissue specimens from 37 patients (12 undergoing arthroscopic biopsies and 25 mini-open biopsies) (78). In this study, PJI was confirmed when two or more cultures grew the same low-virulence organism, or at least one culture yielded a virulent organism (78). Arthroscopically obtained biopsies had a sensitivity of 60% and specificity of 86%, while mini-open biopsies had a sensitivity and specificity of 67% and 86%, respectively, when compared to conventionally obtained biopsies during surgery (78). Akgün et al. (79) retrospectively analyzed the utility of arthroscopic biopsies for the detection of PJI in painful shoulder arthroplasties without objective signs of infection. They concluded that arthroscopically obtained tissue biopsies for culture offered high sensitivity (80%) and specificity (94%) in shoulder PJI diagnosis when at least two cultures positive for the same microorganism were considered as infection.

In summary, hip and knee periprosthetic tissue cultures have a high specificity, suggesting that a positive result (as defined numerically by two positive cultures with identical microorganisms) can be considered diagnostic of infection in most cases. However, sensitivity is more variable, and therefore, a negative tissue culture does not exclude PJI. The threshold for positivity and diagnostic performance of tissue culture for shoulder PJI is not as well defined as for hip and knee PJI.

Does the type of preoperative tissue specimen impact sensitivity and specificity?

Muñoz-Mahamud et al. studied the relevance of specimen type for microbiological evaluation in 123 hip revisions and 175 knee revisions, of which 28 were infected according to IDSA and MSIS criteria (80). In each case, six specimens were collected for culture during arthrotomy and implant removal, including two synovial fluid specimens, two tissue specimens from the synovial-like periarticular membrane (referred to as the neosynovium), and two tissue specimens from the prosthesis–bone interface. Periprosthetic tissue specimens were cultivated on both aerobic and anaerobic agar media and in thioglycolate fluid medium enriched with vitamin K and hemin for 10 days. Of the 28 infected cases, microorganisms were detected from synovial fluid in 20, and from solid tissue in a different 20 cases. The yield of periprosthetic membrane and neosynovium cultures in diagnosing PJI did not differ significantly, though the yield was numerically highest when both the membrane and neosynovium were cultured. The authors concluded that yield is highest when all three culture types are performed (80).

How should periprosthetic tissue culture be processed to optimize culture performance for PJI diagnosis?

Periprosthetic tissue culture methodology impacts diagnostic performance, although not all aspects of culture methodology have been well studied. The first step in microbiological processing of periprosthetic tissues is homogenization. Different homogenization techniques have been analyzed across several studies. Suren et al. compared manual homogenization with a semi-automated method using sterile stainless steel beads in a system that mills samples in sealed bags using magnetic fields (UltraTurrax, Axon Lab AG) (81). In this study, 38 joint specimens from 35 patients (20 hips, 18 knees) were evaluated, with 27 patients diagnosed with PJI by MSIS criteria. Hip tissue cultures had a sensitivity and specificity of 81% and 100%, respectively, using MSIS criteria as the reference standard (81). For knees, sensitivity and specificity were 27% and 57%, respectively (81). The authors concluded that this method of homogenization was not superior to conventional methodology. Rieber et al. also compared two semi-automated homogenization methods: the UltraTurrax method and a method using ceramic bead mills (Precellys Evolution homogenizer, Berin Technologies) in 80 patients (40 with PJI according to ICM criteria) and found that the two methods performed similarly (82). Sebastian et al. compared dithiothreitol treatment of periprosthetic tissues and retrieved implants versus treatment with normal saline and sonication in 73 cases (34 PJI cases diagnosed using MSIS criteria); dithiothreitol treatment had similar sensitivity (67%) and specificity (100%) to conventional treatment (59% and 100%, respectively) (83). This same system was evaluated by Fang et al. together with tissue mechanical homogenization, tissue manual milling, tissue sonication, and direct tissue culture using specimens from 46 patients (28 with PJI according to both the MSIS and IDSA criteria) (84). Sensitivity was 82% for mechanical homogenization, 79% for dithiothreitol treatment, 54% for manual milling, 71% for sonication, and 46% for direct culture; small sample size limited the power to determine differences (84). Jazmati et al. described a protocol using bead beating to homogenize tissue biopsies and including blood culture bottles as a culture medium, again with good results (85). They used microbiological criteria to diagnose PJI (with the reference being isolation of a typical pathogen in one specimen or a potential skin commensal in two or more specimens) and obtained a sensitivity and specificity of 91% and 98%, respectively, when using blood culture bottles and 86% and 96%, respectively, when using thioglycolate enrichment broth (85). Homogenization with steel bead mills was also used by Bémer et al. in a study that evaluated different aspects of microbiology for conventional diagnosis of PJI classified using IDSA criteria (86). This study included 215 patients with confirmed PJI; 89% had a positive microbiological result. This percentage decreased to 77% among 70 patients who had received preoperative antibiotics (86). In this study, the rate of culture positivity varied from 77% to 92% depending on the sample. Interestingly, tissue in contact with the implant yielded optimal results (92% positivity). However, no strict analysis of the homogenization method was performed. Peel et al. (9) reported tissue homogenization using a Seward Stomacher 80 Biomaster (Seward Inc.), with results described in the next section. These studies demonstrate that, although a method that obtains good tissue homogenization must be used for tissue processing, the preferred method (or whether one is better than another) is not well defined.

What are the optimal media types for periprosthetic tissue cultures for the identification of organisms in suspected PJI?

After homogenization, samples are inoculated into or onto media and incubated for growth. Both aerobic and anaerobic media should be used. Anaerobic media are essential for isolation of C. acnes, a pathogen of special interest in shoulder PJI (68, 69, 87–97), but also relevant in PJI affecting other joint types (89, 92, 95). Moreover, other anaerobic bacteria need anaerobic media for isolation (98, 99). In this context, Bémer et al. demonstrated the need for both aerobic and anaerobic media for optimal processing of tissue specimens from patients suspected of PJI (86). In this study, a previous analysis showed that anaerobic Schaedler broth was positive in 75% of samples, compared to the anaerobic blood agar medium, which was positive in 56%. Among patients with PJI caused by aerobic microorganisms, the pediatric blood culture bottle had the highest positivity (83%), followed by the blood agar plate (71%). For PJI caused by anaerobic bacteria, the medium yielding the highest positivity was Schaedler anaerobic broth (75%), followed by the pediatric blood culture bottle (54%). The best combination of media for a 100% agreement of confirmed PJI diagnosis was both liquid media (pediatric blood culture bottle and Schaedler enrichment broth) with blood agar or chocolate agar (86). Omitting one medium would have resulted in a decrease in agreement for aerobic bacteria when the pediatric blood culture bottle was removed, or for anaerobic bacteria when Schaedler anaerobic broth was removed (86). Importantly, an anaerobic blood culture bottle was not included in this study.

Several studies have analyzed the value of liquid media for PJI diagnosis. Shannon et al. compared thioglycolate broth to CDC anaerobic sheep blood agar for isolating C. acnes from 14 shoulder arthroplasties (88). In this study, the liquid was more likely to be positive than the solid medium (P < 0.00001) (88). Jordan et al. evaluated an enrichment process using a preincubation of biopsies in brain heart infusion (BHI) broth for 24 hours followed by inoculation onto solid media, and compared this to standard processing directly in agar plates (not further described in the article) in specimens from 217 revision surgeries (31 infected according to MSIS criteria) (100); among the infected cases, there was increased detection of coagulase-negative staphylococci (15 using standard agar plates, 28 using the enrichment process), which was also found among non-infected cases (one in standard culture, 61 in enriched culture). The addition of BHI broth improved sensitivity (25% for standard culture versus 45% for enriched culture), but with decreased specificity (98% for standard culture versus 59% for enriched culture), and PPV (90% for standard culture versus 42% for enriched culture). This study raised questions about the interpretation of positive results from liquid media alone (100). Hughes et al. also compared different culture media, with cooked meat enrichment broth being the best after blood culture bottles (101). Smith et al. (102) evaluated the utility of broth-only cultures using thioglycolate broth. In this study, PJI was diagnosed using MSIS criteria (which do not distinguish between positive cultures obtained on solid media and in broth only). Broth-only culture sensitivity, specificity, PPV, and NPV were 19%, 88%, 13%, and 92%, respectively. In a study by Peel et al. (9), the authors evaluated the performance of different media, including aerobic and anaerobic agars, thioglycolate broth, and aerobic and anaerobic blood culture bottles, finding higher sensitivities for enrichment broth (75% using Bayesian latent class modeling for PJI diagnosis or 33% using IDSA criteria for PJI diagnosis) compared to that reported by Smith et al. (102). Peel et al. also showed higher specificities (99% for Bayesian modeling; 100% for IDSA criteria). In both studies, broths were only subcultured when they appeared cloudy, but the criteria used for the definition of PJI were different (102, 103), and other unreported differences could affect the results of both studies.

The use of aerobic and anaerobic blood culture bottles to enrich culture growth, especially when performed with automated or semi-automated commercial systems, has been well-studied. Minassian et al. (8) evaluated the usefulness of the BACTEC system for periprosthetic tissue cultures in 79 cases of PJI (50 hips and 21 knees, diagnosed using IDSA criteria) and 243 controls. The authors evaluated two criteria for PJI microbiological diagnosis: first, the growth of an identical microorganism in at least two specimens, and second, the growth of an organism in only one specimen when the organism is highly pathogenic (e.g., S. aureus, Pseudomonas aeruginosa). Using the first criterion, sensitivity and specificity were 83% and 97%, respectively, while if the second criterion was also used, sensitivity increased to 84% and specificity remained at 98% (8). Peel et al. published a series of articles (9, 104) evaluating different aspects of the use of aerobic and anaerobic BACTEC blood culture bottles for periprosthetic tissue culture for PJI diagnosis. In the first article (9), the BACTEC system was compared with the use of solid media and thioglycolate broth, both singly and in combination. Periprosthetic tissue culture in blood culture bottles had a sensitivity and specificity of 92% and 100%, respectively, whereas standard plate and broth cultures had a sensitivity and specificity of 63% and 98%, respectively (9). Their next study evaluated the number of specimens needed for diagnosis using the BACTEC system (9); the authors established that three specimens were optimal when blood culture bottles were used, while four were optimal when standard methodology was used (104). There was no benefit to the use of five or more specimens.

Three additional studies used a different automated blood culture system (BacT/Alert, bioMérieux), with similar results. Sanabria et al. demonstrated improved sensitivity of blood culture bottles compared with the conventional approach (79% versus 76%, respectively) using 158 periprosthetic tissue specimens from hips, knees, elbows, ankles, and shoulders, from 62 patients with suspicion of PJI (no diagnostic criteria were specified); in addition, the BacT/Alert system recovered organisms faster than the classical approach, with more than 80% of isolates detected in fewer than 20 hours using the BacT/Alert system, compared to 5 days needed to detect a similar percentage by the conventional method (105). Birdsall et al. (106) also noted a faster time to organism recovery and improved sensitivity with the use of blood culture bottles; sensitivity was 93% and specificity 88% with the use of blood culture bottles compared to 60% and 90%, respectively, for solid media; PJI was defined according to both MSIS and IDSA criteria in this study. Duployez et al. (107) also compared BacT/Alert with solid media, with blood culture bottles being positive in 149 patients with PJI defined according to IDSA criteria, of whom 58 had negative solid media cultures. Finally, a study that compared BACTEC and BacT/Alert systems by Jeverica et al. showed no differences between them (BACTEC Lytic versus Bac SN) for C. acnes isolation (108). Li et al. (109) performed a meta-analysis of four studies (1,071 patients) and concluded that culturing periprosthetic tissues in blood culture bottles improves performance compared to conventional methods that use solid media and/or enrichment broths, with an overall sensitivity of 70% and specificity of 97%. Based on these studies, the use of automated blood culture systems for performing periprosthetic tissue cultures is recommended in suspected PJI.

In addition to aerobic and anaerobic culture media, the need to employ special media for mycobacteria and/or fungi to improve culture performance in PJI diagnosis has been studied. While many microorganisms can be the cause of PJI, the relative importance of microorganisms is different, with some being more frequently identified than others (110, 111). Mycobacterial PJI is uncommon and mainly the topic of case reports or short series; further, rapidly growing mycobacteria can grow in or on standard bacteriological media (112–114). Fungal PJI is usually caused by Candida species, which can grow in or on commonly used bacteriological media (115). The use of specific media for mycobacteria and/or fungi was analyzed in several studies (116–119), and all showed similar results. The routine use of specific mycobacterial or fungal media is not justified because these infections are uncommon; they should be used only in cases with high suspicion of a fungal or mycobacterial etiology.

In summary, all patients undergoing surgery for suspected PJI should have periprosthetic tissues collected for aerobic and anaerobic cultures. Periprosthetic tissue cultures should ideally be performed in blood culture bottles, or if not possible, using aerobic and anaerobic agars and an anaerobic broth. The routine inclusion of fungal or mycobacterial culture media is not recommended.

What is the optimal duration of incubation for periprosthetic tissue cultures?

The duration of culture incubation to maximize sensitivity has been evaluated. In the study by Atkins et al., specimens were inoculated onto chocolate agar plates (for incubation at 37°C in a CO₂-enriched atmosphere), blood agar plates (for incubation in CO₂ and an anaerobic atmosphere at 37°C), and Robertson’s cooked meat broth for a duration of 7 days (5 days for broth) (59). These incubation times were considered standard until the work by Schäfer et al. (120) in 2008, who studied extended culture incubation to 14 days. In this study, the authors also expanded on the number of culture media, using tryptic soy 5% sheep blood agar, chocolate agar, McConkey II agar (for aerobic culture), and Schaedler agar with vitamin K1 and 5% sheep blood agar (for anaerobic culture), with BHI and Schaedler broths, for aerobic and anaerobic cultures, respectively (120). Among cultures that ultimately grew microorganisms, 74% were positive by 7 days, increasing to 100% by 13 days. Interestingly, contaminants were detected later than pathogens, although 52% of contaminants were still detected in the first 7 days (120). In this study, the authors noted that some species were detected more frequently than others with prolonged incubation, especially C. acnes, Gram-positive aerobic rods, and Finegoldia species. Other studies have demonstrated a higher yield with prolonged culture incubation, especially for anaerobes (98, 121). This is particularly relevant for the detection of C. acnes (122). Several studies support a 10- or 14-day incubation as the standard for microbiological diagnosis to increase sensitivity (45, 69, 97, 123–129), especially for anaerobic bacteria (129, 130) and also for chronic infections, but not for acute PJI (127, 131). This finding has not been uniform. Moreover, studies differ as to the length of incubation necessary to recover C. acnes (89, 129, 132, 133), with some suggesting that a short incubation (7 days) may be adequate for most specimens, including those yielding C. acnes (88, 99, 108, 134–137). Tarabichi et al. (138) evaluated 536 patients diagnosed with PJI using the 2018 ICM criteria and reported that Gram-negative bacteria grew faster than Gram-positive bacteria (P < 0.001), with C. acnes and other less common species (especially anaerobes) growing more slowly. Finally, Morreel et al. studied whether organisms isolated from patients with infection of different durations grew after different incubation durations (139). The authors evaluated 68 “early acute,” 52 “late acute,” and 67 “chronic” infections, according to EBJIS criteria (4). Specimens from patients with “acute” infections were positive within 5 days in more than 97% of cases, while a 14-day incubation period was needed to yield positive results in a similar proportion of patients with “chronic” infections. Reported another way, 7 days of incubation led to a positive result in 99% of “early acute,” 100% of “late acute,” and 88% of “late chronic” infections. Interestingly, there was no difference between contaminants and true pathogens by time-to-detection (139).

Certain situations provide stronger justification for extended culture incubation (e.g., potential C. acnes shoulder arthroplasty infection). Not well-addressed in existing data is whether solid media should continue to undergo incubation after pathogens are identified earlier. This may, of course, depend on the nature of the early growth (e.g., a single colony of microorganism that may be a contaminant or a true pathogen versus a plate laden with S. aureus colonies) and suspicion for polymicrobial infection. Considering all these data, a 5–7-day incubation period is recommended for aerobic tissue cultures (solid media and/or broth), and a 14-day incubation period for anaerobic tissue cultures (solid media and/or broth) from patients with suspected PJI.

What is the optimal number of tissue cultures with phenotypically indistinguishable organisms to confirm PJI?

Atkins et al. (59) performed a prospective study to establish a criterion for the microbiological diagnosis of PJI. This study evaluated culture performance from 297 patients (1,206 specimens), among whom 41 were infected according to the presence of acute inflammatory cells in histological specimens. Tissues were disrupted by vigorous manual agitation with sterile glass beads in sterile diluent and inoculated in chocolate agar (for aerobic incubation) and two plates of blood agar (one for aerobic and one for anaerobic incubation) and Robertson’s cooked meat broth. All plates were incubated for 7 days, and broths were subcultured after 5 days (or when they became turbid). Using a mathematical model, they suggested that the optimal number of cultures needed to yield phenotypically indistinguishable organisms in revision arthroplasty would be three or more independent positive specimens from a set of five or six specimens (sensitivity, 65%; specificity, 100%). Likewise, a previously cited study by Mikkelsen et al. (62) reported that the detection of three or more independent positive specimens in infected knee arthroplasties was 46% sensitive and 100% specific. DeHaan et al. (123) studied the number of cultures necessary to confirm PJI after institutional adoption of a standard culture algorithm (five or more tissue cultures held 10 days) and found that three or more positive cultures for the same skin microbiota bacterium (defined as coagulase-negative staphylococci, corynebacteria, Cutibacterium species) were needed to be considered PJI. A prospective study by Marín et al. (140) analyzing 176 specimens from 40 patients with PJI (defined according to criteria published by Del Pozo and Patel [141] and Trampuz and Zimmerli [142]) and 321 specimens from 82 non-infected patients found that two positive cultures yielded a sensitivity of 96% and specificity of 82%.

Given that the threshold of two positive cultures yielding indistinguishable organisms improves sensitivity without a substantial change in specificity, it has been used by many authors (143, 144) and recommended in PJI consensus documents (1, 2, 4–6). We therefore endorse that two or more positive specimens with the same organism can both confirm the presence of infection and that the isolated organism is the cause of the disease.

Are single positive periprosthetic tissue cultures diagnostic of PJI for certain organism types (and if so, which organisms)?

Although ideally, two cultures for the same microorganism are useful to define culture positivity, clinicians are often left with a situation where only one of multiple cultures is positive; understanding whether the organism is a pathogen or not can be challenging. Many organisms causing PJI can equally well be isolated as contaminants. In the study by Atkins et al., only a small number of isolated microorganisms were classically pathogenic, such as S. aureus (59). Minassian et al. evaluated differential interpretation of positive cultures comparing a standard definition (two positive cultures with the same microorganism) and less stringent criteria (one positive culture for a pathogenic organism, such as S. aureus, Escherichia coli, or Candida albicans) (8). In this study, inclusion of less stringent criteria maintained sensitivity at 84% without losing specificity (97% and 98%, respectively). While most authors state that a single positive culture should be considered a contaminant, especially when the infecting microorganism has low virulence, this guidance may differ by joint type and bacterium isolated. In the study by DeHaan et al. (123), the authors suggest that growth in one or more cultures of any virulent organism (e.g., S. aureus) should be considered PJI. Balato et al. (145) conducted a retrospective study of 31 knee PJIs (defined by ICM criteria) and 136 aseptic total knee arthroplasty failures, with five or more tissue specimens obtained from each patient. The percentage with a single positive culture was the same for patients with aseptic loosening and PJI.

Therefore, we posit that a single positive culture may represent a true pathogen or a contaminant. For the diagnosis of PJI, it is useful as a diagnostic criterion, but alone, it is insufficient to confirm infection. A single positive culture for a virulent organism (such as S. aureus) may be more likely to represent true PJI. The single positive culture needs to be considered with other diagnostic criteria for PJI, and clinical judgment used to manage treatment decisions.

What is the optimal total number of cultures needed to yield two cultures with phenotypically indistinguishable organisms (when cultures are positive)?

The early model proposed by Atkins et al. suggested that the optimal number of tissue specimens collected (to obtain three phenotypically indistinguishable organisms) was five or six (59). Bémer et al. (using conventional media with a pediatric blood culture bottle for culture) (86) conducted a prospective multicenter study to determine the minimal number of specimens and culture media required for accurate diagnosis of PJI among 264 suspected PJI cases (of which 215 were confirmed according to IDSA criteria). Bacteriological and PJI diagnostic criteria were modeled using a random selection of two, three, or four specimens and compared with those obtained using the standard five specimens recommended. The authors concluded that four specimens were sufficient for diagnosing PJI. This finding was also supported by other authors. Gandhi et al. (146) evaluated 113 consecutive patients with infected total hip and total knee arthroplasties and correlated the type of culture specimen and number of specimens taken during surgery to the likelihood of a positive culture result using solid aerobic and anaerobic media incubated for 14 days. They concluded that a minimum of four tissue cultures is necessary for hip and knee PJI diagnosis. They highlighted that increasing the number of specimens increases sensitivity but reduces specificity. Peel et al. evaluated the number of specimens needed for PJI diagnosis using the BACTEC system (104); the authors established that three specimens were optimal when blood culture bottles were used, while four were optimal when the standard methodology was used (104). There was no benefit to the collection of five or more specimens. Kheir et al. (129) reviewed 711 patients (329 hips and 382 knees) whose tissues were cultured using plate and broth media and concluded that the optimal number of cultures and growth duration depended on the type of organism. They suggested that when the organism is unknown at the time of surgery, five specimens should be obtained and held for at least 8 days, since this number provides the greatest yield of positive cultures for diagnosing PJI. In the previously cited article by Trampuz et al., the sensitivity of tissue culture increased from 50% to 73% as the number of specimens collected increased from two to five or more. Abdel et al. (147) reviewed 23 studies with a total of 4,321 patients undergoing revision hip and knee arthroplasty. The mean number of culture specimens taken across cohorts included in the studies was four (range 2–8); three or more intraoperative specimens yielded a higher NPV than one or two specimens (85%) to rule out infection without limiting the PPV (78%) to confirm infection (sensitivity, 62%; specificity, 93%).

Based on the above data and acknowledging that the number of cultures needed to optimize diagnostic yield is also contingent upon methodology, we recommend that three tissue specimens be collected during surgery for aerobic and anaerobic culture if tissue cultures are processed in blood culture bottles; four tissue specimens should be collected if using aerobic and anaerobic agar and an anaerobic broth culture.

What is the optimal total number of cultures needed to confirm shoulder PJI?

The number of positive cultures needed to define shoulder PJI is not well established. While the literature considers two phenotypically indistinguishable organisms as indicative of true infection, whether this is always the case for phenotypically homogeneous microorganisms, such as C. acnes, is unknown. The interpretation of two or more isolates of C. acnes as “identical” can be difficult because phenotypic characteristics of this organism are similar across strains, with few characteristics (hemolysis, resistance markers) able to differentiate unrelated strains. Even these differences are difficult to assess in many cases. Bumgarner et al. used single-locus sequence typing to establish the identity between strains isolated from deep tissue biopsies and demonstrated that clonality between isolates cannot be assumed based on phenotypic tests alone (93).

Due to the absence of reliable diagnostic tests, C. acnes shoulder PJI remains difficult to diagnose (52); several authors have assessed the optimal number of cultures needed to confirm infection in shoulder PJI. Ahmadi et al. (148) performed a retrospective analysis of 537 shoulder arthroplasties from 429 patients who underwent revision shoulder arthroplasty. Cultures were positive in 169 of the 537 surgeries (32%). Among these 169 cases, 64% were solely positive for C. acnes. The authors concluded that four or five specimens may not be sufficient to diagnose some shoulder PJIs. Ahsan et al. (91) studied 137 revision shoulder arthroplasties from which a minimum of four specimens had been submitted for culture, and at least one was positive for Cutibacterium species. Their investigation concluded that Cutibacterium species are unevenly distributed within culture-positive revised shoulders, with soft-tissue and explant specimens having higher rates of culture positivity than joint fluid. Consequently, specimen number and source (explant, soft tissue, fluid) may influence culture results for revision shoulder arthroplasties. Ribera et al. used the criterion of two or more positive cultures for a potential contaminant or isolation of a potential pathogen from at least one specimen for the interpretation of cultures in aseptic loosening surgeries (149). However, in this study, the isolation of a microorganism from only one tissue specimen was confirmed in several cases by results obtained using implant sonication (149), suggesting that a single positive tissue culture may still reflect true infection. The 2018 ICM shoulder criteria used a scoring system and assigned point values to cultures based on the number of cultures positive for a phenotypically indistinguishable organism and whether the organism was a pathogen versus a commensal organism (28). Patel et al. attempted to validate the shoulder PJI definition of the second ICM (71); in their study, the highest specificity was obtained for a single positive culture of a virulent organism or two positive cultures with an identical low virulence organism (100% in both cases), although with lower sensitivity (20% and 50%, respectively). A single positive culture with a low virulence organism was more sensitive (55%), but specificity was only 82.6% (71).

For shoulder PJI, the number of positive cultures for a phenotypically indistinguishable organism needed to confirm infection is unknown. In the absence of definitive data, we recommend the same criteria be utilized for the diagnosis of shoulder PJI as for other prosthetic joints. Interpretation of the results of tissue cultures should consider clinical and other diagnostic data.

IMPACT OF PRIOR ANTIMICROBIALS ON THE PERFORMANCE OF CULTURE

The growth and identification of organisms in culture depend on both the viability of infecting microorganisms in submitted specimens and the ability of the microbiology laboratory techniques used to cultivate the organism to the level of detectability. Despite viable organisms in the human host, PJI may yield negative cultures due to the phenotypic state of the microorganisms in submitted specimens. PJI is characterized by the presence of biofilms adherent to the implant surface. Sessile microorganisms can be difficult to culture as they are in a metabolically quiescent state; free-flowing planktonic organisms are more easily culturable using standard methodologies. Periprosthetic tissue specimens may fail to represent the implant surface where biofilms are present; synovial fluid cultures may be similarly affected. In addition, microbial aggregates (especially in synovial fluid) may decrease culture sensitivity. Antimicrobial pretreatment may reduce the yield of culture by reducing or eliminating microbial viability either through causing cell death or by interfering with growth and metabolic activity. In one series, 53% of patients with culture-negative PJI based on author-defined criteria had received antimicrobial therapy in the 3 months prior to the diagnosis (150). Therefore, in this review, the relevance of antimicrobial pretreatment prior to culture collection for PJI was assessed. Two separate types of antimicrobial pre-treatment were considered: the use of intraoperative antimicrobials as prophylaxis prior to operative culture collection, and the therapeutic use of antimicrobials prior to pre-operative synovial fluid aspiration or surgical culture collection.

Do pre- or intraoperative (prophylactic) antimicrobials reduce the yield of culture for the detection and identification of organisms causing PJI?

Antimicrobial prophylaxis reduces the risk of surgical site infection and is included in guidelines for the prevention of infection in arthroplasty (151, 152). Antimicrobial administration is timed such that bactericidal concentrations of the agents are present in tissues at the time of incision. It has been postulated that surgical antimicrobial prophylaxis might reduce the viability of recovered planktonic organisms and thereby reduce the yield of culture in PJI (153).

Six studies that addressed the impact of preoperative antimicrobial prophylaxis on culture yield during revision arthroplasty were identified. Three were conducted in individuals who had both confirmed PJI prior to surgery and a known infecting organism. Burnett et al. identified 26 knee PJIs according to author-defined criteria in which patients did not receive antimicrobial therapy in the 4 weeks prior to surgery (154). In the operating room, percutaneous aspiration was performed, followed by infusion of cefazolin and vancomycin. Tourniquets were applied, and after arthrotomy, additional cultures (three swabs, one synovial tissue culture) were collected between 18 and 33 minutes after antibiotic administration. Identical organisms were identified in the operative aspiration cultures and tissue cultures in all 26 patients (154), suggesting no impact of antimicrobial prophylaxis. Ghanem et al. reported a retrospective review of 171 patients with knee PJI (classified by author-defined criteria) in which prior aspiration cultures were positive (34). Operative cultures were negative in 9/72 (13%) who received surgical prophylaxis, and in 8/99 (8%) who did not receive surgical prophylaxis prior to culture collection (P = 0.3), though non-optimized prophylaxis may have reduced the impact of antimicrobial prophylaxis (34). Tetreault et al. studied 65 patients with either hip or knee PJI according to MSIS criteria who had a known infecting organism and had not received antimicrobials in the 2 weeks prior to surgery (35). Patients were randomized to receive prophylactic antibiotics either before incision or after intraoperative cultures were collected. Concordance between preoperative and operative cultures was similar in patients randomized to antibiotic prophylaxis prior to incision and after culture collection (28/34 [82%] concordant versus 25/31 [81%] concordant) (35).

Three studies considered patients with confirmed PJI in whom the microorganism was not always known prior to surgery. In the first, 40 patients undergoing surgery for suspected PJI involving hips and knees served as their own controls. Each had three tissue cultures collected after arthrotomy; thereafter, cefazolin was administered, after which three additional cultures were collected (155). Tissue concentrations of cefazolin were confirmed to be >2 µg/mL at the time of the second set of culture collection. There was no difference in culture yield between the two sets of cultures (OR 0.99; 95% CI 0.4–2.48), and no difference in the number of colony-forming units per gram of submitted tissue. In another study, 28 patients with PJI defined by criteria published by Zimmerli et al. (156) and involving hips, knees, or shoulders were randomized to receive surgical antimicrobial prophylaxis before or after cultures were collected; cultures were positive in 10/14 in both groups (157). In the final study, 425 patients underwent revision knee arthroplasty, of whom 29 were ultimately diagnosed with PJI according to MSIS criteria (158). 284 underwent surgery when the standard protocol was withholding antibiotic prophylaxis until surgical cultures were collected, and 141 after the protocol was changed to antimicrobial administration prior to incision. There was no difference in the number of positive cultures or types of organisms recovered (158).

In a meta-analysis including these studies, Wouthuyzen-Bakker et al. demonstrated a difference in culture yield of 7% in patients with PJI depending on when preoperative antimicrobials were administered (145/165 [88%] among those receiving prophylaxis prior to culture collection and 344/362 [95%] among those whose cultures were collected before antimicrobial administration; P = 0.004) (159). The difference in culture yield in the setting of only “chronic” PJI was lower (78/89 [88%] and 52/57 [91%]; P = 0.59) (159).

It is important to note that in some of these studies, either antimicrobial prophylaxis and/or culture processing (e.g., use of blood culture bottles for synovial fluid incubation, collection of an ideal number of tissue specimens, adequate duration of culture incubation) was not optimized. With these limitations, the impact of surgical antimicrobial prophylaxis on culture yield in PJI is low. Surgeons should therefore not withhold antimicrobials in patients undergoing revision surgery for PJI when the organism is known. When PJI is suspected, but the organism has not been identified prior to surgery (and/or when a commensal organism is identified via percutaneous aspiration, which might represent contamination), the risks of withholding antibiotics should be weighed against the small increase in culture yield on a case-by-case basis.

Do preoperative (therapeutic) antimicrobials reduce the yield of culture for the detection and identification of organisms causing PJI?

Patients presenting with signs and symptoms of PJI are often treated with antimicrobials before a diagnosis is confirmed; multiple studies have identified prior antimicrobial treatment as a risk factor for culture-negative PJI. Two studies evaluated the impact of antimicrobial therapy on preoperative synovial fluid aspiration. Barrack et al. studied 69 patients undergoing aspiration for suspected knee PJI, of whom 20 were diagnosed by pre-MSIS criteria (160); although antibiotics were held for 2 weeks prior to aspiration, prior antibiotic use was still associated with false-negative aspiration cultures. Shahi et al. studied 161 patients with knee PJI diagnosed by MSIS criteria (161). Cultures were negative in 26% of the 53 who had received antimicrobial therapy within 2 weeks of the aspiration compared to 13% in the 108 who had not received recent antimicrobials (RR 2.0; 95% CI 1.05–3.9).

Six studies evaluated the impact of prior antibiotic exposure on operative culture yield. Spangehl et al. studied 202 patients undergoing revision THA, among whom 35 were diagnosed with PJI by author-defined criteria; operative cultures were positive in 8/17 (47%) who received antibiotics within 2 weeks prior to surgery and 17/18 (94%) of those who had not received antimicrobials in that time frame (162). In another study including 26 patients with knee PJI by author-defined criteria, 12 had positive operative cultures; there was a trend toward lower culture yield in those who received antibiotics prior to surgery, with the lowest yield in those who remained on antibiotics at surgery (62). Malekzadeh et al. performed a retrospective case-control study, including 135 patients with culture-negative PJI (classified by author-defined criteria) of hips and knees and 135 matched controls (163). Patients receiving antimicrobials within 3 months prior to operative culture collection were more likely to have culture-negative infection (matched OR 4.11; 95% CI 2.33–7.25). In a retrospective cohort study, 40 patients with culture-negative PJI were compared to 135 with culture-positive PJI (both defined by MSIS criteria) (164); patients with culture-negative infection were more likely to have received antimicrobials for the same-joint PJI than those who had positive cultures at the time of surgery (68% versus 42%, P = 0.005).

As with the studies on antimicrobial prophylaxis above, some of these studies were older, and microbiological methods were not optimized. In the previously described study by Trampuz et al., any administration of antibiotics within 2 weeks prior to culture collection was associated with a higher false-negative rate in 79 patients with PJI (67). In those whose last exposure was >14 days prior to surgical culture collection, cultures were positive in 77%; culture yield was lower in those who had received antibiotics within 4–14 days of surgery (48%) and within 0–3 days of surgery (41%) (P < 0.001 for trend) (67). The yield of sonicate fluid culture was higher than periprosthetic tissue cultures when antimicrobials had been administered within the 14 days prior to surgery. Scorzolini et al. also included sonication in their protocol when evaluating 41 patients with PJI (classified using author-defined criteria) (165). Periprosthetic tissue culture yield was lower in those whose last antibiotic exposure was within 15 days (17%) versus those who had been off antimicrobials at least 15 days (57%) (P = 0.01), but this difference was not seen when sonicate fluid culture was utilized.

Should antibiotics be held prior to culture collection? If so, what is the optimal duration for antimicrobials to be held so that culture yield is not impacted?

Although it is evident from available literature that preoperative antimicrobial therapy impacts culture yield in PJI, the duration of the antimicrobial “holiday” prior to culture collection, after which the effect is no longer relevant, is unknown. Most studies pre-defined antimicrobial exposure as a categorical variable, defining antibiotic exposure as within 2 weeks prior to culture collection (160–162, 165). The true impact, however, may persist beyond 2 weeks; Malekzadah et al. defined antimicrobial exposure as within 3 months of surgical culture collection (163). In this study, the median duration of antimicrobial therapy was also longer for those who ultimately had culture-negative versus positive infections (35 versus 18 days). The authors were unable to establish an antimicrobial-free time period that would optimize the likelihood of culture positivity (163). In the study performed by Barrack et al., four patients with PJI whose aspiration cultures were negative despite a 2-week antimicrobial therapy holiday underwent repeat aspiration after a 4-week antimicrobial holiday and had positive cultures (160). Based on the available literature, the authors suggest that whenever feasible, antimicrobial therapy should be withheld at least 14 days prior to culture collection; it is uncertain whether longer time intervals may improve culture yield, or whether the duration of the antimicrobial holiday should be longer in those whose pre-holiday antimicrobial duration is longer.

Do certain culture methods perform better than others in the setting of prior antibiotic exposure?

Culture for the identification of organisms causing PJI requires that viable pathogens be collected in pre-operative aspiration and/or surgical tissue and/or implant cultures, that those organisms remain viable during the process of transport to the laboratory, and that the laboratory optimally processes specimens to recover organism growth. Some culture negativity may be due to organisms with fastidious growth characteristics, independent of prior antimicrobial exposure. Prior antimicrobial exposure is most likely to impact the inclusion of viable organisms within the specimen submitted for culture.

Some but not all studies suggest that sonication of implants to liberate organisms may improve the yield of culture in patients with PJI and prior antimicrobial administration. Sonication is hypothesized to dislodge sessile organisms from explanted prosthesis and resuscitate them such that they will revert to a phenotypic growth state (153). Sonicate fluid culture was less impacted by prior antimicrobial therapy than conventional periprosthetic tissue culture in two studies (165), but no difference was seen in a third study (166).

Molecular assays are less impacted by antimicrobial administration compared to traditional periprosthetic cultures, as these do not require the viability of organisms but target microbial DNA. However, molecular assays are still affected by antimicrobial agents if the burden of microbial nucleic acids is affected by the therapy. While studies suggest improved performance of molecular methods in identifying organisms in the setting of prior antimicrobial exposure, this is outside the scope of this review.

FUTURE RESEARCH AND GUIDELINE NEEDS

During this review, important gaps were noted that leave the understanding as to how best to obtain, process, and interpret cultures in the evaluation of suspected PJI imperfect. For synovial fluid culture, data are insufficient to identify an optimal fluid volume, and the optimal duration of synovial fluid culture incubation is incompletely defined. Although synovial fluid culture in aerobic and anaerobic blood culture bottles is recommended, there are several types of blood culture bottles available, and it is unclear whether they are interchangeable for synovial fluid culture. Furthermore, some investigators add supplements to blood culture bottles (to theoretically enhance the growth of certain microorganisms and possibly mitigate the absence of blood); whether supplementation is needed is unknown. Ideally, purpose-designed aerobic and anaerobic bottles for culturing synovial fluid from the site of prosthetic joints would be welcomed. Given limited concordance between synovial fluid and operative cultures, it remains unknown whether a microorganism obtained through percutaneous sampling, particularly if part of commensal skin microbiota, is the pathogenic organism when operative tissue cultures are negative in the absence of antibiotic exposure.

For tissue cultures, further work is needed to understand how many positive cultures of phenotypically indistinguishable organisms are needed to confirm shoulder PJI, and whether, for C. acnes, the standard of phenotypically indistinguishable is sufficient. As with synovial fluid, the development of blood culture bottles specifically designed for processing tissue specimens is of great interest.

While the recommendation to discontinue antibiotics for 2 weeks prior to culture collection is justified by the literature, whether a longer period may be preferred (and whether that is contingent on the duration of prior antibiotic exposure and/or the type of joint) is unknown.

Finally, most of the available literature reflects data from hip and knee arthroplasty. With the growth of arthroplasty in other joints, additional study is needed to know whether the same criteria can be applied to these other joints (especially shoulder, ankle, and elbow arthroplasty).

CONCLUSION

In the decades since the advent of arthroplasty and the recognition of PJI, culture technologies specific to PJI have been developed. The presence of organisms in synovial fluid and/or periprosthetic tissue is implicit in the presence of PJI, yet despite advances, the performance of culture in confirming PJI and identifying the implicated pathogens remains imperfect. Culture performance can be improved by withholding antimicrobials prior to specimen collection, collecting a sufficient volume of fluid and number of tissue specimens, processing aerobic and anaerobic cultures in blood culture bottles, obtaining at least four periprosthetic tissue specimens (three if processing in blood culture bottles), and incubating aerobic cultures for 5–7 days and anaerobic cultures for 14 days (Fig. 4). Culture interpretation is challenging, especially when a single culture is positive for a commensal skin organism which may or may not represent the pathogen in PJI. We highlight important areas for research to continue to optimize culture performance in suspected PJI.

Fig 4.

Overview of culture-based diagnosis of periprosthetic joint infection.