Aims. To assess how the cost-effectiveness of total hip arthroplasty (THA) and total knee arthroplasty (TKA) varies with age, sex, and preoperative Oxford Hip or Knee Score (OHS/OKS); and to identify the patient groups for whom THA/TKA is cost-effective. Methods. We conducted a cost-effectiveness analysis using a Markov model from a United Kingdom NHS perspective, informed by published analyses of patient-level data. We assessed the cost-effectiveness of THA and TKA in adults with hip or knee osteoarthritis compared with having no arthroplasty surgery during the ten-year time horizon. Results. THA and TKA cost < £7,000 per quality-adjusted life-year (QALY) gained at all preoperative scores below the absolute referral thresholds calculated previously (40 for OHS and 41 for OKS). Furthermore, THA cost < £20,000/QALY for patients with OHS of ≤ 45, while TKA was cost-effective for patients with OKS of ≤ 43, since the small improvements in quality of life outweighed the cost of surgery and any subsequent revisions. Probabilistic and one-way sensitivity analyses demonstrated that there is little uncertainty around the conclusions. Conclusion. If society is willing to pay £20,000 per QALY gained, THA and TKA are cost-effective for nearly all patients who currently undergo surgery, including all patients at and above our calculated absolute referral thresholds. Cite this article: Bone Joint J 2020;102-B(7):950–958

Aims. The aim of this study was to evaluate the performance of metagenomic next-generation sequencing (mNGS) in detecting pathogens from synovial fluid of prosthetic joint infection (PJI) patients. Methods. A group of 75 patients who underwent revision knee or hip arthroplasties were enrolled prospectively. Ten patients with primary arthroplasties were included as negative controls. Synovial fluid was collected for mNGS analysis. Optimal thresholds were determined to distinguish pathogens from background microbes. Synovial fluid, tissue, and sonicate fluid were obtained for culture. Results. A total of 49 PJI and 21 noninfection patients were finally included. Of the 39 culture-positive PJI cases, mNGS results were positive in 37 patients (94.9%), and were consistent with culture results at the genus level in 32 patients (86.5%) and at the species level in 27 patients (73.0%). Metagenomic next-generation sequencing additionally identified 15 pathogens from five culture-positive and all ten culture-negative PJI cases, and even one pathogen from one noninfection patient, while yielding no positive findings in any primary arthroplasty. However, seven pathogens identified by culture were missed by mNGS. The sensitivity of mNGS for diagnosing PJI was 95.9%, which was significantly higher than that of comprehensive culture (79.6%; p = 0.014). The specificity is similar between mNGS and comprehensive culture (95.2% and 95.2%, respectively; p = 1.0). Conclusion. Metagenomic next-generation sequencing can effectively identify pathogens from synovial fluid of PJI patients, and demonstrates high accuracy in diagnosing PJI. Cite this article: Bone Joint Res 2020;9(7):440–449

Aims. This study aimed to determine the diagnostic performance of radiographic criteria to detect aseptic acetabular loosening after revision total hip arthroplasty (THA). Secondary aims were to determine the predictive values of different thresholds of migration and to determine the predictive values of radiolucency criteria. Patients and Methods. Acetabular component migration to re-revision was measured retrospectively using Ein-Bild-Rontgen-Analyse (EBRA-Cup) and manual measurements (Sutherland method) in two groups: Group A, 52 components (48 patients) found not loose at re-revision and Group B, 42 components (36 patients) found loose at re-revision between 1980 and 2015. The presence and extent of radiolucent lines was also assessed. Results. Using EBRA, both proximal translation and sagittal rotation were excellent diagnostic tests for detecting aseptic loosening. The area under the receiver operating characteristic (ROC) curves was 0.94 and 0.93, respectively. The thresholds of 2.5 mm proximal translation or 2° sagittal rotation (EBRA) in combination with radiolucency criteria had a sensitivity of 93% and specificity of 88% to detect aseptic loosening. The sensitivity, specificity, positive predictive value and negative predictive value (NPV) of radiolucency criteria were 41%, 100%, 100% and 68% respectively. Manual measurements of both proximal translation and sagittal rotation were very good diagnostic tests. The area under the ROC curve was 0.86 and 0.92 respectively. However, manual measurements had a decreased specificity compared with EBRA. Radiolucency criteria had a poor sensitivity and NPV of 41% and 68% respectively. Conclusion. This study shows that EBRA and manual migration measurements can be used as accurate diagnostic tools to detect aseptic loosening of cementless acetabular components used at revision THA. Radiolucency criteria should not be used in isolation to exclude loosening of cementless acetabular components used at revision THA given their poor sensitivity and NPV. Cite this article: Bone Joint J 2017;99-B:458–64

Aims. The diagnosis of developmental dysplasia of the hip (DDH) is challenging owing to extensive variation in paediatric pelvic anatomy. Artificial intelligence (AI) may represent an effective diagnostic tool for DDH. Here, we aimed to develop an anteroposterior pelvic radiograph deep learning system for diagnosing DDH in children and analyze the feasibility of its application. Methods. In total, 10,219 anteroposterior pelvic radiographs were retrospectively collected from April 2014 to December 2018. Clinicians labelled each radiograph using a uniform standard method. Radiographs were grouped according to age and into ‘dislocation’ (dislocation and subluxation) and ‘non-dislocation’ (normal cases and those with dysplasia of the acetabulum) groups based on clinical diagnosis. The deep learning system was trained and optimized using 9,081 radiographs; 1,138 test radiographs were then used to compare the diagnoses made by deep learning system and clinicians. The accuracy of the deep learning system was determined using a receiver operating characteristic curve, and the consistency of acetabular index measurements was evaluated using Bland-Altman plots. Results. In all, 1,138 patients (242 males; 896 females; mean age 1.5 years (SD 1.79; 0 to 10) were included in this study. The area under the receiver operating characteristic curve, sensitivity, and specificity of the deep learning system for diagnosing hip dislocation were 0.975, 276/289 (95.5%), and 1,978/1,987 (99.5%), respectively. Compared with clinical diagnoses, the Bland-Altman 95% limits of agreement for acetabular index, as determined by the deep learning system from the radiographs of non-dislocated and dislocated hips, were -3.27° - 2.94° and -7.36° - 5.36°, respectively (p < 0.001). Conclusion. The deep learning system was highly consistent, more convenient, and more effective for diagnosing DDH compared with clinician-led diagnoses. Deep learning systems should be considered for analysis of anteroposterior pelvic radiographs when diagnosing DDH. The deep learning system will improve the current artificially complicated screening referral process. Cite this article: Bone Joint J 2020;102-B(11):1574–1581

Aims. The routine use of dual-mobility (DM) acetabular components in total hip arthroplasty (THA) may not be cost-effective, but an increasing number of patients undergoing THA have a coexisting spinal disorder, which increases the risk of postoperative instability, and these patients may benefit from DM articulations. This study seeks to examine the cost-effectiveness of DM components as an alternative to standard articulations in these patients. Patients and Methods. A decision analysis model was used to evaluate the cost-effectiveness of using DM components in patients who would be at high risk for dislocation within one year of THA. Direct and indirect costs of dislocation, incremental costs of using DM components, quality-adjusted life-year (QALY) values, and the probabilities of dislocation were derived from published data. The incremental cost-effectiveness ratio (ICER) was established with a willingness-to-pay threshold of $100 000/QALY. Sensitivity analysis was used to examine the impact of variation. Results. In the base case, patients with a spinal deformity were modelled to have an 8% probability of dislocation following primary THA based on published clinical ranges. Sensitivity analysis revealed that, at its current average price ($1000), DM is cost-effective if it reduces the probability of dislocation to 0.9%. The threshold cost at which DM ceased being cost-effective was $1180, while the ICER associated with a DM THA was $71 000 per QALY. Conclusion. These results indicate that under specific clinical and economic thresholds, DM components are a cost-effective form of treatment for patients with spinal deformity who are at high risk of dislocation after THA. Cite this article: Bone Joint J 2018;100-B:1297–1302

Aims. We quantitatively compared the 3D bone density distributions on CT scans performed on scaphoid waist fractures subacutely that went on to union or nonunion, and assessed whether 2D CT evaluations correlate with 3D bone density evaluations. Methods. We constructed 3D models from 17 scaphoid waist fracture CTs performed between four to 18 weeks after fracture that did not unite (nonunion group), 17 age-matched scaphoid waist fracture CTs that healed (union group), and 17 age-matched control CTs without injury (control group). We measured the 3D bone density for the distal and proximal fragments relative to the triquetrum bone density and compared findings among the three groups. We then performed bone density measurements using 2D CT and evaluated the correlation with 3D bone densities. We identified the optimal cutoff with diagnostic values of the 2D method to predict nonunion with receiver operating characteristic (ROC) curves. Results. In the nonunion group, both the distal (100.2%) and proximal (126.6%) fragments had a significantly higher bone density compared to the union (distal: 85.7%; proximal: 108.3%) or control groups (distal: 91.6%; proximal: 109.1%) using the 3D bone density measurement, which were statistically significant for all comparisons. 2D measurements were highly correlated to 3D bone density measurements (Spearman’s correlation coefficient (R) = 0.85 to 0.95). Using 2D measurements, ROC curve analysis revealed the optimal cutoffs of 90.8% and 116.3% for distal and proximal fragments. This led to a sensitivity of 1.00 if either cutoff is met and a specificity of 0.82 when both cutoffs are met. Conclusion. Using 3D modelling software, nonunions were found to exhibit bone density increases in both the distal and proximal fragments in CTs performed between four to 18 weeks after fracture during the course of treatment. 2D bone density measurements using standard CT scans correlate well with 3D models. In patients with scaphoid fractures, CT bone density measurements may be useful in predicting the likelihood of nonunion. Cite this article: Bone Joint J 2020;102-B(9):1200–1209

Aims. Whether to perform hybrid surgery (HS) in contrast to anterior cervical discectomy and fusion (ACDF) when treating patients with multilevel cervical disc degeneration remains a controversial subject. To resolve this we have undertaken a meta-analysis comparing the outcomes from HS with ACDF in this condition. Methods. Seven databases were searched for studies of HS and ACDF from inception of the study to 1 September 2019. Both random-effects and fixed-effects models were used to evaluate the overall effect of the C2-C7 range of motion (ROM), ROM of superior/inferior adjacent levels, adjacent segment degeneration (ASD), heterotopic ossification (HO), complications, neck disability index (NDI) score, visual analogue scale (VAS) score, Japanese Orthopaedic Association (JOA) score, Odom’s criteria, blood loss, and operating and hospitalization time. To obtain more credible results contour-enhanced funnel plots, Egger’s and Begg’s tests, meta-regression, and sensitivity analyses were performed. Results. In total, 17 studies involving 861 patients were included in the analysis. HS was found to be superior to ACDF in maintaining C2-C7 ROM and ROM of superior/inferior adjacent levels, but HS did not reduce the incidence of associated level ASD. Also, HS did not cause a higher rate of HO than ACDF. The frequency of complications was similar between the two techniques. HS failed to achieve more favourable outcomes than ACDF using the NDI, VAS, JOA, and Odom’s scores. HS did not show any more advantages in operating or hospitalization time but did show reduction in blood loss. Conclusion. Although HS maintained cervical kinetics, it failed to reduce the incidence of ASD. This finding differs from previous reports. Moreover, patients did not show more benefits from HS with respect to symptom improvement, prevention of complications, and clinical outcomes. Cite this article: Bone Joint J 2020;102-B(8):981–996

Aims. This study aimed to investigate the role of quantitative histological analysis in the diagnosis of fracture-related infection (FRI). Patients and Methods. The clinical features, microbiology culture results, and histological analysis in 156 surgically treated nonunions were used to stratify the likelihood of associated infection. There were 64 confirmed infected nonunions (one or more confirmatory criteria: pus, sinus, and bacterial growth in two or more samples), 66 aseptic nonunions (no confirmatory criteria), and 26 possibly infected nonunions (pathogen identified from a single specimen and no confirmatory criteria). The histological inflammatory response was assessed by average neutrophil polymorph (NPs) counts per high-power field (HPF) and compared with the established diagnosis. Results. Assuming a cut-off of over five neutrophils per high-power field to diagnose septic nonunion, there was 80% sensitivity and 100% specificity (accuracy 90%). Using a cut-off of no neutrophils seen in any high-power field to diagnose aseptic nonunion, there was a sensitivity of 85% and a specificity of 98% (accuracy 92%). Conclusion. Histology can be used in a bimodal fashion as a diagnostic test for FRI. The presence of more than five NPs/HPF had a positive predictive value for infected nonunion of 100%, while the complete absence of any NPs is almost always indicative of an aseptic nonunion (positive predictive value of 98%). Cite this article: Bone Joint J 2018;100-B:966–72

Aim. The aim of this study was to compare the cost-effectiveness of intramedullary nail fixation and ‘locking’ plate fixation in the treatment of extra-articular fractures of the distal tibia. Patients and Methods. An economic evaluation was conducted from the perspective of the United Kingdom National Health Service (NHS) and personal social services (PSS), based on evidence from the Fixation of Distal Tibia Fractures (UK FixDT) multicentre parallel trial. Data from 321 patients were available for analysis. Costs were collected prospectively over the 12-month follow-up period using trial case report forms and participant-completed questionnaires. Cost-effectiveness was reported in terms of incremental cost per quality adjusted life year (QALY) gained, and net monetary benefit. Sensitivity analyses were conducted to test the robustness of cost-effectiveness estimates. Results. Mean NHS and PSS costs were significantly lower for patients treated with an intramedullary nail than for those treated with a locking plate (-£970, 95% confidence interval (CI) -1685 to -256; p = 0.05). There was a small increase in QALYs gained in the nail fixation group (0.01, 95% CI -0.03 to 0.06; p = 0.52). The probability of cost-effectiveness for nail fixation exceeded 90% at cost-effectiveness thresholds as low as £15 000 per additional QALY. The cost-effectiveness results remained robust to several sensitivity analyses. Conclusion. This trial-based economic evaluation suggests that nail fixation is a cost-effective alternative to locking plate fixation. Cite this article: Bone Joint J 2018;100-B:624–33

Aims. The primary aim was to assess the independent influence of coronavirus disease (COVID-19) on 30-day mortality for patients with a hip fracture. The secondary aims were to determine whether: 1) there were clinical predictors of COVID-19 status; and 2) whether social lockdown influenced the incidence and epidemiology of hip fractures. Methods. A national multicentre retrospective study was conducted of all patients presenting to six trauma centres or units with a hip fracture over a 46-day period (23 days pre- and 23 days post-lockdown). Patient demographics, type of residence, place of injury, presentation blood tests, Nottingham Hip Fracture Score, time to surgery, operation, American Society of Anesthesiologists (ASA) grade, anaesthetic, length of stay, COVID-19 status, and 30-day mortality were recorded. Results. Of 317 patients with acute hip fracture, 27 (8.5%) had a positive COVID-19 test. Only seven (26%) had suggestive symptoms on admission. COVID-19-positive patients had a significantly lower 30-day survival compared to those without COVID-19 (64.5%, 95% confidence interval (CI) 45.7 to 83.3 vs 91.7%, 95% CI 88.2 to 94.8; p < 0.001). COVID-19 was independently associated with increased 30-day mortality risk adjusting for: 1) age, sex, type of residence (hazard ratio (HR) 2.93; p = 0.008); 2) Nottingham Hip Fracture Score (HR 3.52; p = 0.001); and 3) ASA (HR 3.45; p = 0.004). Presentation platelet count predicted subsequent COVID-19 status; a value of < 217 × 10. 9. /l was associated with 68% area under the curve (95% CI 58 to 77; p = 0.002) and a sensitivity and specificity of 63%. A similar number of patients presented with hip fracture in the 23 days pre-lockdown (n = 160) and 23 days post-lockdown (n = 157) with no significant (all p ≥ 0.130) difference in patient demographics, residence, place of injury, Nottingham Hip Fracture Score, time to surgery, ASA, or management. Conclusion. COVID-19 was independently associated with an increased 30-day mortality rate for patients with a hip fracture. Notably, most patients with hip fracture and COVID-19 lacked suggestive symptoms at presentation. Platelet count was an indicator of risk of COVID-19 infection. These findings have implications for the management of hip fractures, in particular the need for COVID-19 testing. Cite this article: Bone Joint J 2020;102-B(9):1219–1228

Aims. Prosthetic joint infection (PJI) remains a major clinical challenge. Neutrophil CD64 index, Fc-gamma receptor 1 (FcγR1), plays an important role in mediating inflammation of bacterial infections and therefore could be a valuable biomarker for PJI. The aim of this study is to compare the neutrophil CD64 index in synovial and blood diagnostic ability with the standard clinical tests for discrimination PJI and aseptic implant failure. Methods. A total of 50 patients undergoing revision hip and knee arthroplasty were enrolled into a prospective study. According to Musculoskeletal Infection Society (MSIS) criteria, 25 patients were classified as infected and 25 as not infected. In all patients, neutrophil CD64 index and percentage of polymorphonuclear neutrophils (PMN%) in synovial fluid, serum CRP, ESR, and serum CD64 index levels were measured preoperatively. Receiver operating characteristic (ROC) curves and the area under the curve (AUC) were analyzed for each biomarker. Results. Serum CD64 index showed no significant difference between the two groups (p = 0.091). Synovial fluid CD64 index and PMN% discriminated good differentiation between groups of PJI and aseptic failure with AUC of 0.946 (95% confidence interval (CI) 0.842 to 0.990) and 0.938 (95% CI 0.832 to 0.987) separately. The optimal threshold value of synovial CD64 index for the diagnosis of PJI was 0.85, with a sensitivity of 92.00%, a specificity of 96.00%, and diagnostic odds ratio (DOR) of 227.11. Conclusion. The present study demonstrates that CD64 index in synovial fluid could be a promising laboratory marker for screening PJI. The cut-off values of 0.85 for synovial CD64 index has the potential to distinguish aseptic failure from PJI. Cite this article: Bone Joint J 2020;102-B(4):463–469

Aims. The aim of this study was to evaluate the diagnostic value of preoperative serum CRP, white blood cell count (WBC), percentage of neutrophils (%N), and neutrophil to lymphocyte ratio (NLR) when using the fracture-related infection (FRI) consensus definition. Methods. A cohort of 106 patients having surgery for suspected septic nonunion after failed fracture fixation were studied. Blood samples were collected preoperatively, and the concentration of serum CRP, WBC, and differential cell count were analyzed. The areas under the curve (AUCs) of diagnostic tests were compared using the z-test. Regression trees were constructed and internally cross-validated to derive a simple diagnostic decision tree. Results. Using the FRI consensus definition, 46 patients (43%) were identified as infected. Sensitivity, specificity, and AUC of CRP were 67% (95% confidence interval (CI) 52% to 80%), 61% (95% CI 47% to 74%), and 0.64 (95% CI 0.54 to 0.74); of WBC count were 17% (95% CI 9% to 31%), 95% (95% CI 86% to 99%), and 0.57 (95% CI 0.50 to 0.62); of %N 13% (95% CI 6% to 26%), 87% (95% CI 76% to 93%), and 0.50 (95% CI 0.43 to 0.56); and of NLR 28% (95% CI 17% to 43%), 80% (95% CI 68% to 88%), and 0.54 (95% CI 0.46 to 0.63), respectively. A better performance of serum CRP was shown in comparison to the leucocyte count (p = 0.006), %N (p < 0.001), and NLR (p = 0.001). A statistically lower serum CRP level was shown in patients with an infection caused by a low virulence microorganism in comparison to high virulence bacteria (p = 0.008). We found that a simple decision tree approach using only low serum neutrophils (< 3.615 × 10. 9. /l) and low CRP (< 2.45 mg/l) may allow better identification of aseptic cases. Conclusion. The evaluated serum inflammatory markers showed limited diagnostic value in the preoperative diagnosis of FRI when using the uniform FRI Consensus Definition. Therefore, they should remain as suggestive criteria in diagnosing FRI. Although CRP showed a higher performance in comparison to the other serum markers, it is insufficiently accurate to diagnose a septic nonunion, especially when caused by low virulence microorganisms. Cite this article: Bone Joint J 2020;102-B(7):904–911

Aims. This study aims to determine the proportion of patients with end-stage knee osteoarthritis (OA) possibly suitable for partial (PKA) or combined partial knee arthroplasty (CPKA) according to patterns of full-thickness cartilage loss and anterior cruciate ligament (ACL) status. Methods. A cross-sectional analysis of 300 consecutive patients (mean age 69 years (SD 9.5, 44 to 91), mean body mass index (BMI) 30.6 (SD 5.5, 20 to 53), 178 female (59.3%)) undergoing total knee arthroplasty (TKA) for Kellgren-Lawrence grade ≥ 3 knee OA was conducted. The point of maximal tibial bone loss on preoperative lateral radiographs was determined as a percentage of the tibial diameter. At surgery, Lachman’s test and ACL status were recorded. The presence of full-thickness cartilage loss within 16 articular surface regions (two patella, eight femoral, six tibial) was recorded. Results. According to articular cartilage loss and ACL status, 195/293 (67%) were suitable for PKA or CPKA: medial unicompartmental knee arthroplasty (UKA) 97/293 (33%); lateral UKA 25 (9%); medial bicompartmental arthroplasty 31 (11%); lateral bicompartmental arthroplasty 12 (4%); bicondylar-UKA 23 (8%); and patellofemoral arthroplasty (PFA) seven (2%). The ACL was intact in 166 (55%), frayed in 82 (27%), disrupted in 12 (4%), and absent in 33 (11%). Lachman testing was specific (97%) but poorly sensitive (38%) for disrupted/absent ACLs. The point of maximal tibial bone loss showed good interclass correlation (ICC 0.797, 0.73 to 0.85 95% confidence interval (CI); p < 0.001) and was more posterior when the ACL was absent. Maximum tibial bone loss occurring at > 55% of the anterior to posterior distance predicted ACL absence with 93% sensitivity and 91% specificity (area under the curve 0.97 (0.94 to 0.99 95% CI; p < 0.001). Conclusion. ACL status can be reliably determined from a lateral radiograph using the location of maximal tibial bone loss. According to regions of cartilage loss and ACL status, two-thirds of patients with end-stage knee OA could potentially be treated with PKA or CPKA. Cite this article: Bone Joint J 2020;102-B(6):716–726

Aim. This paper describes the methods applied to assess the cost-effectiveness of cemented versus uncemented hemiarthroplasty among hip fracture patients in the World Hip Trauma Evaluation Five (WHiTE5) trial. Methods. A within-trial cost-utility analysis (CUA) will be conducted at four months postinjury from a health system (National Health Service and personal social services) perspective. Resource use pertaining to healthcare utilization (i.e. inpatient care, physiotherapy, social care, and home adaptations), and utility measures (quality-adjusted life years) will be collected at one and four months (primary outcome endpoint) postinjury; only treatment of complications will be captured at 12 months. Sensitivity analysis will be conducted to assess the robustness of the results. Conclusion. The planned analysis strategy described here records our intent to conduct a within-trial CUA alongside the WHiTE5 trial

Aims. The purpose of this study was to validate the diagnosis of periprosthetic joint infection (PJI) in the Danish Hip Arthroplasty Register (DHR). Patients and Methods. We identified a cohort of patients from the DHR who had undergone primary total hip arthroplasty (THA) since 1 January 2005 and followed them until first-time revision, death, emigration or until 31 December 2012. Revision for PJI, as registered in the DHR, was validated against a benchmark which included information from microbiology databases, prescription registers, clinical biochemistry registers and clinical records. We estimated the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for PJI in the DHR alone and in the DHR when combined with microbiology databases. Results. In total, 1382 of the 37 826 primary THAs in the DHR were registered as having been revised for any cause once 26 patients with errors in registration had been excluded: 232 of these were for PJI. For this group, the sensitivity was 67%, specificity 95%, PPV 77%, and NPV 92%. Combining the data from the DHR with those from microbiology databases increased the sensitivity to 90% and also improved specificity (100%), PPV (98%) and NPV (98%). Conclusion. Only two thirds of revisions for PJI were captured in the DHR and only 77% of the PJI reported to the DHR could be confirmed to be infected. . Take home message: combining the data from the DHR with those from microbiology databases substantially improved the validity of the diagnosis of PJI and should enable future register-based studies. Cite this article: Bone Joint J 2016;98-B:320–5

Aims. Natural Language Processing (NLP) offers an automated method to extract data from unstructured free text fields for arthroplasty registry participation. Our objective was to investigate how accurately NLP can be used to extract structured clinical data from unstructured clinical notes when compared with manual data extraction. Methods. A group of 1,000 randomly selected clinical and hospital notes from eight different surgeons were collected for patients undergoing primary arthroplasty between 2012 and 2018. In all, 19 preoperative, 17 operative, and two postoperative variables of interest were manually extracted from these notes. A NLP algorithm was created to automatically extract these variables from a training sample of these notes, and the algorithm was tested on a random test sample of notes. Performance of the NLP algorithm was measured in Statistical Analysis System (SAS) by calculating the accuracy of the variables collected, the ability of the algorithm to collect the correct information when it was indeed in the note (sensitivity), and the ability of the algorithm to not collect a certain data element when it was not in the note (specificity). Results. The NLP algorithm performed well at extracting variables from unstructured data in our random test dataset (accuracy = 96.3%, sensitivity = 95.2%, and specificity = 97.4%). It performed better at extracting data that were in a structured, templated format such as range of movement (ROM) (accuracy = 98%) and implant brand (accuracy = 98%) than data that were entered with variation depending on the author of the note such as the presence of deep-vein thrombosis (DVT) (accuracy = 90%). Conclusion. The NLP algorithm used in this study was able to identify a subset of variables from randomly selected unstructured notes in arthroplasty with an accuracy above 90%. For some variables, such as objective exam data, the accuracy was very high. Our findings suggest that automated algorithms using NLP can help orthopaedic practices retrospectively collect information for registries and quality improvement (QI) efforts. Cite this article: Bone Joint J 2020;102-B(7 Supple B):99–104

Aims. To validate the Sydney Hamstring Origin Rupture Evaluation (SHORE), a hamstring-specific clinical assessment tool to evaluate patient outcomes following surgical treatment. Methods. A prospective study of 70 unilateral hamstring surgical repairs, with a mean age of 47.3 years (15 to 73). Patients completed the SHORE preoperatively and at six months post-surgery, and then completed both the SHORE and Perth Hamstring Assessment Tool (PHAT) at three years post-surgery. The SHORE questionnaire was validated through the evaluation of its psychometric properties, including; internal consistency, reproducibility, reliability, sensitivity to change, and ceiling effect. Construct validity was assessed using Pearson’s correlation analysis to examine the strength of association between the SHORE and the PHAT. Results. The SHORE demonstrated an excellent completion rate (100%), high internal consistency (Cronbach’s alpha 0.78), and good reproducibility (intraclass correlation coefficient (ICC) 0.82). The SHORE had a high correlation with the validated PHAT score. It was more sensitive in detecting clinical change compared to the PHAT. A ceiling effect was not present in the SHORE at six months; however, a ceiling effect was identified in both scores at three years post-surgery. Conclusion. This study has validated the SHORE patient reported outcome measure (PROM) as a short, practical, reliable, valid, and responsive tool that can be used to assess symptom and function following hamstring injury and surgical repair. Cite this article: Bone Joint J 2020;102-B(3):388–393

Aims. α-defensin is a biomarker which has been described as having a high degree of accuracy in the diagnosis of periprosthetic joint infection (PJI). Current meta-analyses are based on the α-defensin laboratory-based immunoassay rather than the quick on-table lateral flow test kit. This study is the first meta-analysis to compare the accuracy of the α-defensin laboratory-based immunoassay and the lateral flow test kit for the diagnosis of PJI. Materials and Methods. A systematic review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Inclusion criteria were all clinical studies where the diagnosis of PJI was uncertain. All studies selected used the Musculoskeletal Infection Society (MSIS) or modified MSIS criteria. Two independent reviewers reviewed the studies and extracted data. A meta-analysis of results was carried out: pooled sensitivity, specificity, positive and negative likelihood ratio, heterogeneity and areas under curves are reported. Results. Ten studies (759 patients) were included. Of these, seven studies (640 patients) evaluated the laboratory-based α-defensin immunoassay and three (119 patients) the lateral flow test. The pooled sensitivity and specificity of the qualitative α-defensin laboratory immunoassay was 0.953 (95% confidence interval (CI) 0.87 to 0.984) and 0.965 (95% CI 0.943 to 0.979) respectively. The pooled positive likelihood ratio (PLR) and negative likelihood ratio (NLR) were 34.86 (95% CI 19.34 to 62.85) and 0.02 (95% CI 0.00 to 0.11). The pooled sensitivity and specificity of the lateral flow test were 0.774 (95% CI 0.637 to 0.870) and 0.913 (95% CI 0.828 to 0.958), respectively. The pooled PLR and NLR were 8.675 (95% CI 4.229 to 17.794) and 0.248 (95% CI 0.147 to 0.418), respectively. Conclusion. The pooled sensitivity and specificity of the lateral flow test were lower than those of the α-defensin laboratory-based immunoassay test. Hence, care must be taken with interpretation of the lateral flow test when relying on its results for the intra-operative diagnosis of PJI. Cite this article: Bone Joint J 2018;100-B:66–72

Aims. The aim of this study was to evaluate the ability of a machine-learning algorithm to diagnose prosthetic loosening from preoperative radiographs and to investigate the inputs that might improve its performance. Methods. A group of 697 patients underwent a first-time revision of a total hip (THA) or total knee arthroplasty (TKA) at our institution between 2012 and 2018. Preoperative anteroposterior (AP) and lateral radiographs, and historical and comorbidity information were collected from their electronic records. Each patient was defined as having loose or fixed components based on the operation notes. We trained a series of convolutional neural network (CNN) models to predict a diagnosis of loosening at the time of surgery from the preoperative radiographs. We then added historical data about the patients to the best performing model to create a final model and tested it on an independent dataset. Results. The convolutional neural network we built performed well when detecting loosening from radiographs alone. The first model built de novo with only the radiological image as input had an accuracy of 70%. The final model, which was built by fine-tuning a publicly available model named DenseNet, combining the AP and lateral radiographs, and incorporating information from the patient’s history, had an accuracy, sensitivity, and specificity of 88.3%, 70.2%, and 95.6% on the independent test dataset. It performed better for cases of revision THA with an accuracy of 90.1%, than for cases of revision TKA with an accuracy of 85.8%. Conclusion. This study showed that machine learning can detect prosthetic loosening from radiographs. Its accuracy is enhanced when using highly trained public algorithms, and when adding clinical data to the algorithm. While this algorithm may not be sufficient in its present state of development as a standalone metric of loosening, it is currently a useful augment for clinical decision making. Cite this article: Bone Joint J 2020;102-B(6 Supple A):101–106

Aims. This pilot study aimed to evaluate prospectively the use of inlet radiographs of the hip as an alternative method of the assessment of reduction after the surgical treatment of developmental dysplasia of the hip (DDH). Patients and Methods. The children in this study underwent surgery between January 2013 and January 2015. All had inlet radiographs and CT scans post-operatively. Data were analysed by determining inter-observer reliability and intra-observer reproducibility, using the kappa value (K). Differences were settled by discussion between the two observers until a consensus was reached. The sensitivity and specificity of the radiographic and CT results were compared. A total of 26 radiographs were obtained from 23 children, with a mean age of 2.38 years (one to five). Results. Similar high levels of intra- and inter-observer agreement were observed (K = 0.834, 95% confidence interval (CI)). There was a high agreement between the radiographic and CT results (K = 0.834, 5% CI), with excellent sensitivity and a specificity of 95.5%. Conclusion. These results suggest that inlet radiographs may be a reliable method of assessing the reduction of the hip after the surgical treatment of DDH. Cite this article: Bone Joint J 2017;99-B:697–701