This international, multi-center study investigated the effect of individual components of surgery on the clinical outcomes of patients treated for fracture-related infection (FRI). All patients with surgically treated FRIs, confirmed by the FRI consensus definition, were included. Data were collected on demographics, time from injury to FRI surgery, soft tissue reconstruction, stabilization and systemic and local anti-microbial therapy. Patients were followed up for a minimum of one year. In total, 433 patients were treated with a mean age of 49.7 years (17–84). The mean follow-up time was 26 months (range 12–72). The eradication of infection was successful in 86.4% of all cases and 86.0% of unhealed infected fractures were healed at the final review. In total, 3.3% required amputation. The outcome was not dependent on age, BMI, the presence of metalwork or time from injury (recurrence rate 16.5% in FRI treated at 1–10 weeks after injury; 13.1% at 11–52 weeks; 12.1% at >52 weeks: p = 0.52). The debridement and retention of a stable implant (DAIR) had a failure rate of 21.4%; implant exchange to a new internal fixation had a failure rate of 12.5%; and conversion to external fixation had a failure rate of 10.3% (adjusted hazard ratio (aHR) DAIR vs. Ext Fix 2.377; 95% C.I. 0.96–5.731). Tibial FRI treated with a free flap was successful in 92.1% of cases and in 80.4% of cases without a free flap (HR 0.38; 95% C.I. 0.14–1.0), while the use of NPWT was associated with higher recurrence rates (HR 3.473; 95% C.I. 1.852–6.512). The implantation of local antibiotics reduced the recurrence from 18.7% to 10.0% (HR 0.48; 95% C.I. 0.29–0.81). The successful treatment of FRI was multi-factorial. These data suggested that treatment decisions should not be based on time from injury alone, as other factors also affected the outcome. Further work to determine the best indications for DAIR, free flap reconstruction and local antibiotics is warranted.

Aims: The number of convolutional neural networks (CNN) available for fracture detection and classification is rapidly increasing. External validation of a CNN on a temporally separate (separated by time) or geographically separate (separated by location) dataset is crucial to assess generalizability of the CNN before application to clinical practice in other institutions. We aimed to answer the following questions: are current CNNs for fracture recognition externally valid?; which methods are applied for external validation (EV)?; and, what are reported performances of the EV sets compared to the internal validation (IV) sets of these CNNs? Methods: The PubMed and Embase databases were systematically searched from January 2010 to October 2020 according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. The type of EV, characteristics of the external dataset, and diagnostic performance characteristics on the IV and EV datasets were collected and compared. Quality assessment was conducted using a seven-item checklist based on a modified Methodologic Index for NOn-Randomized Studies instrument (MINORS). Results: Out of 1,349 studies, 36 reported development of a CNN for fracture detection and/or classification. Of these, only four (11%) reported a form of EV. One study used temporal EV, one conducted both temporal and geographical EV, and two used geographical EV. When comparing the CNN’s performance on the IV set versus the EV set, the following were found: AUCs of 0.967 (IV) versus 0.975 (EV), 0.976 (IV) versus 0.985 to 0.992 (EV), 0.93 to 0.96 (IV) versus 0.80 to 0.89 (EV), and F1-scores of 0.856 to 0.863 (IV) versus 0.757 to 0.840 (EV). Conclusion: The number of externally validated CNNs in orthopaedic trauma for fracture recognition is still scarce. This greatly limits the potential for transfer of these CNNs from the developing institute to another hospital to achieve similar diagnostic performance. We recommend the use of geographical EV and statements such as the Consolidated Standards of Reporting Trials–Artificial Intelligence (CONSORT-AI), the Standard Protocol Items: Recommendations for Interventional Trials–Artificial Intelligence (SPIRIT-AI) and the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis–Machine Learning (TRIPOD-ML) to critically appraise performance of CNNs and improve methodological rigor, quality of future models, and facilitate eventual implementation in clinical practice. Cite this article: Bone Jt Open 2021;2(10):879–885.

Background: Radiographic measurements of initial displacement of tibial plateau fractures and of postoperative reduction are used to determine treatment strategy and prognosis. We assessed the association between radiographic measurements and the risk of conversion to total knee arthroplasty (TKA) at the time of follow-up. Methods: A total of 862 patients surgically treated for tibial plateau fractures between 2003 and 2018 were eligible for this multicenter cross-sectional study. Patients were approached for follow-up, and 477 (55%) responded. The initial gap and step-off were measured on the preoperative computed tomography (CT) scans of the responders. Condylar widening, residual incongruity, and coronal and sagittal alignment were measured on postoperative radiographs. Critical cutoff values for gap and step-off were determined using receiver operating characteristic curves. Postoperative reduction measurements were categorized as adequate or inadequate on the basis of cutoff values in international guidelines. Multivariable analysis was performed to assess the association between each radiographic measurement and conversion to TKA. Results: Sixty-seven (14%) of the patients had conversion to TKA after a mean follow-up of 6.5 ± 4.1 years. Assessment of the preoperative CT scans revealed that a gap of >8.5 mm (hazard ratio [HR] = 2.6, p < 0.001) and step-off of >6.0 mm (HR = 3.0, p < 0.001) were independently associated with conversion to TKA. Assessment of the postoperative radiographs demonstrated that residual incongruity of 2 to 4 mm was not associated with increased risk of TKA compared with adequate fracture reduction of 4 mm resulted in increased risk of TKA. Coronal (HR = 1.6, p = 0.05) and sagittal malalignment (HR = 3.7 p < 0.001) of the tibia were strongly associated with conversion to TKA. Conclusions: Substantial preoperative fracture displacement was a strong predictor of conversion to TKA. Postoperative gaps or step-offs of >4 mm as well as inadequate alignment of the tibia were strongly associated with an increased risk of TKA. Level of Evidence: Therapeutic Level III. See Instructions for Authors for a complete description of levels of evidence.

The overall incidence of acetabular fractures is estimated as 5 to 8 per 100 000 people per year, which accounts for approximately 60 000 injured individuals annually in Europe.^1,2 A pelvic injury can have major consequences for physical functioning, participating in social activities, and the ability to work. Acetabular fracture treatment consists of either nonoperative treatment (56% of patients), open reduction and internal fixation (38% of patients) or primary total hip arthroplasty (THA; 6% of patients).³

Fracture-related infections (FRIs) are classically considered to be early (0–2 weeks), delayed (3–10 weeks) or late (>10 weeks) based on hypothesized differences in causative pathogens and biofilm formation. Treatment strategies often reflect this classification, with debridement, antimicrobial therapy and implant retention (DAIR) preferentially reserved for early FRI. This study examined pathogens isolated from FRI to confirm or refute these hypothesized differences in causative pathogens over time. Cases of FRI managed surgically at three centres between 2015–2019 and followed up for at least one year were included. Data were analysed regarding patient demographics, time from injury and pathogens isolated. Patients who underwent DAIR were also analysed separately. In total, 433 FRIs were studied, including 51 early cases (median time from injury of 2 weeks, interquartile range (IQR) of 1–2 weeks), 82 delayed cases (median time from injury of 5 weeks, IQR of 4–8 weeks) and 300 late cases (median time from injury of 112 weeks, IQR of 40–737 weeks). The type of infection was associated with time since injury; early or delayed FRI are most likely to be polymicrobial, whereas late FRIs are more likely to be culture-negative, or monomicrobial. Staphylococcus aureus was the most commonly isolated pathogen at all time points; however, we found no evidence that the type of pathogens isolated in early, delayed or late infections were different (p = 0.2). More specifically, we found no evidence for more virulent pathogens (S. aureus, Gram-negative aerobic bacilli) in early infections and less virulent pathogens (such as coagulase negative staphylococci) in late infections. In summary, decisions on FRI treatment should not assume microbiological differences related to time since injury. From a microbiological perspective, the relevance of classifying FRI by time since injury remains unclear.