We reviewed prospectively collected data from patients who underwent partial or complete removal of osteosynthesis implants between July 2012 to May 2021 at a tertiary university hospital. We sent the retrieved implants to our microbiology lab for SFC. Reasons for implant removal include both septic and aseptic causes. We aimed to send all removed osteosynthesis material for SFC, also without suspicion of infection. The minimum diagnostic workup was that all patients received examinations for clinical signs of infection (e.g. wound defects, sinus tracts) and radiological imaging. If the diagnosis was unclear or if there was suspicion of infection, more diagnostic methods were used, such as serum inflammatory markers and blood cultures. Histopathological examination was not part of routine diagnostics. Patients who had fewer than three tissue cultures obtained were excluded from the analysis.

We classified patients based on the consensus FRI criteria defined by the AO foundation and the EBJIS, fused as the FRI group (Metsemakers et al., 2018). Patients were classified as “suggestive FRI” or “confirmed FRI”. Although not included in the consensus definition, we classified patients as “aseptic” if they did not meet any FRI criteria. Patients were classified before and after incorporating SFC results to evaluate if these results led to a new confirmed FRI. We analysed FRI cases based on the time interval between the index surgery and onset of symptoms: early (less than 2 weeks), delayed (2–10 weeks), and late FRI (more than 10 weeks) (Stevenson et al., 2022).

Added diagnostic value was defined as the number of patients for whom SFC was essential for pathogen detection and confirming diagnosis. SFC was considered to be essential for pathogen identification when tissue cultures alone yielded inconclusive results. Additionally, in patients with (preoperative) confirmed FRI by non-microbiological criteria, a positive SFC was considered to have added value when tissue cultures yielded a single positive result.

Patients received empiric antibiotic treatment in the case of high suspicion of infection or sepsis. Based on definite culture results, antibiotic treatment was tailored to targeted therapy or discontinued if negative, a decision made in close consultation with the medical microbiologist.

We aimed to obtain at least three tissue cultures from every patient. All tissue cultures were taken with separate sterile rongeurs and placed in separate sterile containers. The containers were immediately delivered to the microbiology laboratory for further analysis. Tissue samples were divided into 0.1 mL aliquots and placed onto blood agar, chocolate agar, McConkey agar, and Brucella blood agar. The cultures were incubated at 35–37 °C for 14 d. Blood, chocolate, and McConkey plates were incubated in CO₂ 5 %, and the Brucella blood agar plate was incubated anaerobically.

The osteosyntheses components that were removed during surgery were sent to the microbiology laboratory in a sterile hard-polypropylene container. Material was processed within 6 h after implant removal. A sodium chloride solution was added to cover at least 90 % of the implant. The container was placed in an SFC bath (BactoSonic 14.2, Bandelin electronic, Berlin, Germany) for 1 min at 40 kHz. Before and after sonication, the container was vortexed manually for 30 s. Subsequently, 100 µL of SFC fluid was transferred to blood, chocolate, and Brucella agar plates, and 10 mL was inoculated into aerobe and anaerobe blood culture bottles. All media were incubated at 35–37 °C for 14 d, with blood and chocolate agar plates being incubated in 5 % CO₂, and the Brucella agar was incubated anaerobically. In case of growth on the plates, the number of colony-forming units was counted. Microorganisms were classified by using the matrix-assisted laser desorption/ionization time-of-flight analyser (MALDI-ToF) system (BrukerDaltonics). According to FRI criteria, SFC as separate diagnostics cannot confirm infection, and no cut-off is given as a diagnostic threshold. Differentiation between possible contamination or true pathogens was therefore determined together with other diagnostic criteria and consensus during multidisciplinary team meetings.

Microorganisms found by tissue and SFC were analysed. According to FRI consensus criteria, pathogens are confirmed when at least two separate deep tissue and/or implant cultures (including SFC) yield the same microorganism. No diagnostic cut-off value for a positive SFC is defined by these criteria.

A contaminant is a microorganism that is unlikely to be causative for infection and more likely to have been introduced during sample collection, processing, or analysis. Single positive cultures were considered to be suggestive of infection and interpreted alongside clinical findings. In case (highly) virulent pathogens are identified, one positive sample increases suspicion of infection (Depypere et al., 2020). We considered the following to be virulent microorganisms: Staphylococcus aureus, Staphylococcus lugdunensis, Enterococcus spp., beta-hemolytic streptococci, Enterobacterales, and Pseudomonas aeruginosa (Ryan, 2022). Cutibacterium acnes, Coagulase-negative Staphylococci, Corynebacterium spp., anaerobes, and other microorganisms not commonly associated with FRI were considered to be low-virulence microorganisms for the purpose of this study. Micrococcus spp. was considered to be a contaminant. All culture results were discussed in multidisciplinary team meetings, including our microbiology department, to establish diagnosis and treatment plans.

We summarized patient characteristics using descriptive statistics. Percentages were used for presenting categorical variables. Descriptive comparisons between the categorical variables were performed using Pearson's χ2 test or Fisher's exact test. The sensitivity, specificity, negative predictive value, and positive predictive value of the diagnostic tools were calculated using 2×2 contingency tables. We used FRI consensus criteria as a reference standard to determine these values (Metsemakers et al., 2018). We calculated the combined sensitivity and specificity of SFC and tissue culture by defining a positive result as the isolation of identical organisms from tissue cultures or concordant growth of identical organisms in a single tissue culture and SFC. Factors associated with FRI were assessed for the relationship between positive SFC by using univariate analysis and multivariate logistic regression analysis. Variables were selected based on clinical relevance of implant-related factors for inclusion in multivariate analysis. A p value below 0.05 for a two-sided test was considered to be statistically significant. We performed the statistical analysis using IBM SPSS Statistics (version 28.0.1.0).

We analysed 96 patients with three or more tissue cultures and SFC. Postoperative diagnosis based on FRI criteria (without SFC), included 14 aseptic cases, 35 suggestive FRIs, and 47 confirmed FRIs (Table 1).

After including SFC results, four aseptic cases changed diagnosis to suggestive FRI, and three cases changed from suggestive FRI to confirmed FRI (Fig. 1). The three cases that changed to confirmed FRI included two late-onset FRIs and one early-onset FRI.

In 37 confirmed FRIs (74 %), the diagnosis was initially confirmed using non-microbiological criteria (Fig. 1). In three of these cases (8 %), SFC was needed to achieve microbiological confirmation together with a single positive tissue culture.

In 18 / 50 confirmed FRIs (36 %), five or more tissue cultures were obtained. In these cases, SFC was not essential for confirming FRI.

The sensitivity of tissue culture was 74 %, and the specificity was 98 % (Table 2). The sensitivity of SFC was 84 %, and the specificity was 83 %. Tissue culture had lower sensitivity compared to SFC (74 % vs. 84 %, with p<0.002). Tissue cultures had higher specificity compared to SFC cultures, but this difference was not statistically significant (98 % vs. 83 %, p<0.174). The sensitivity of tissue cultures alone was significantly lower than tissue cultures combined with SFC (74 % vs. 88 %, with p<0.001). Specificity of tissue cultures alone was significantly higher than the combination of tissue cultures with SFC (98 % vs. 83 %, with p<0.001).

The most common identified microorganisms were Staphylococcus aureus, Cutibacterium acnes, and Staphylococcus epidermidis (CoNS) (Table 3). Based on preoperative diagnosis, SFC was more frequently positive in all diagnostic groups. Tissue cultures were positive in 45 cases: 1 / 15 aseptic, 16 / 48 suggestive FRIs, and 28 / 33 confirmed FRIs. SFC was positive in 54 cases: 5 / 15 aseptic, 20 / 48 suggestive FRIs, and 29 / 33 confirmed FRIs. Based on postoperative diagnosis, tissue cultures yielded polymicrobial results in 22 / 50 confirmed FRIs (44 %), and SFC yielded results in 20 / 50 confirmed FRIs (40 %). SFC yielded isolated positive results in four aseptic cases, four suggestive FRI cases, and three confirmed FRI cases.

A total of 12 cases received antibiotics between 1 and 14 d before removal of hardware (2 suggestive FRIs, 10 confirmed FRIs). Tissue cultures were positive in 8 / 12 cases (67 %), and SFC was positive in 10 / 12 cases (83 %). This included two cases where SFC was positive with negative tissue cultures and three cases with positive SFC and one positive tissue culture.

In one suggestive FRI, SFC confirmed FRI together with tissue culture by identifying a virulent pathogen (Pseudomonas aeruginosa). In one confirmed FRI, SFC was isolated as positive and identified Staphylococcus aureus with negative tissue cultures.

A total of 51 cases were treated as FRIs with antibiotic therapy, including 49 / 50 confirmed FRIs and 2 / 36 suggestive FRIs. In one confirmed FRI, no antibiotic therapy was given, and an observational (watchful waiting) management approach was implemented. Cases that were reclassified from aseptic to suggestive FRI did not receive antibiotic therapy.

In 37 / 50 confirmed FRI cases, tissue cultures confirmed infection by identifying identical pathogens in at least two specimens. In 21 / 50 (42 %), SFC would not have altered antibiotic therapy as the pathogens were concordantly identified by tissue culture. In 13 / 37 cases, SFC additionally detected a virulent pathogen not identified by tissue cultures (all Enterococcus spp.), which influenced the antibiotic therapy regimen. In 10 / 13 cases, SFC identified additional low-virulent pathogens not detected by tissue culture, which could have influenced antibiotic therapy in eight cases (16 %). Micrococcus luteus was additionally detected in two cases and was considered to be a contaminant.

SFC was positive in 54 cases (Table 4). Of these, six occurred in delayed FRIs (implant removal 2–10 weeks after surgery), and 48 occurred in late FRIs (>10 weeks). Although not statistically significant, SFC was more frequently positive in patients with a history of open fractures (9 / 15 cases; p=0.750). In cases with implant loosening, SFC was positive in 15 / 20 cases (p=0.057). Non-union was significantly associated with negative SFC results (24 / 39; p=0.004). Among patients with preoperative antibiotic therapy, SFC was significantly associated with a positive result (p=0.031).

In binary logistic regression analysis, non-union was independently associated with lower odds of positive SFC ([OR] 0.3, 95 % CI: 0.1–0.7, with p=0.005) (Table 5). Implant loosening was associated with higher odds of positive SFC ([OR] 3.8, 95 % CI 1.1–12.4, with p=0.034).