Aim of study (background): Probability of survival (Ps) is calculated presently by using the TRISS methodology for trauma patients. This utilises physiological scoring parameters, injury scoring system and the patient age. The physiological parameters need to be recorded for determining the RTS, but this data is often missed. The aim of our study was to assess how the essential variables correlate with the Ps and if any other variables contribute significantly to the Ps. Depending on the correlation of any new variables, is it time to change the formula for calculating Ps?. Material and methods: A retrospective study was carried out involving 678 trauma patients. The ISS and RTS scores were calculated from the trauma charts. The relationship between ISS, RTS, age and GCS with Ps was assessed using the correlation and regression analysis and then the affect of gender on Ps was assessed using a T test. Results: ISS of trauma victims had a mean of 10.22. The results showed a strong negative correlation between ISS and Ps with an r value of −0.633 (p< 0.005). GCS correlated strongly with Ps, with an r value of 0.733 (p< 0.005). In the regression analysis; ISS and GCS showed a strong correlation with Ps. RTS made the weakest contribution to Ps, followed by age. GCS made the strongest unique contribution. There also no significant difference in the mean scores of Ps for males and females (p< 0.005). Conclusion: The results indicated significantly strong correlations between ISS and GCS with Ps. There was a poor correlation between RTS and Ps. This is helpful for the patients in whom RTS scores cannot be calculated, as GCS can be used in place of RTS. There may be a need for a new system to calculate Ps using GCS and gender

Aim of study (background): Probability of survival (Ps) is calculated presently by using the TRISS methodology. This utilises physiological scoring parameters, injury scoring system and age. The physiological parameters need to be recorded for determining the RTS, but are frequently missed. The aim of my study was to assess if any other variables contribute significantly to the Probability of survival, and if they do is it time to change the variables used to calculate the Probability of survival?. Material and methods: A retrospective study was carried out from January’01 to August’03. The relationship between Injury Severity Score (ISS), Revised Trauma Score (RTS), age and Glasgow Coma Score (GCS) with Probability of survival was assessed using the correlation and regression analysis and then the affect of gender on probability of survival was assessed. Results: ISS had a mean of 21.69 (range, 2–50). The results showed a strong negative correlation between ISS and Ps with an r value of −0.692 (p< 0.005). GCS correlated strongly with Ps, with an r value of 0.457 (p< 0.005). In the regression analysis; ISS, RTS, age and GCS showed a strong correlation with Ps. RTS made the strongest unique contribution to Ps, followed by age, ISS and then GCS. There also was a significant difference in the mean scores of Ps for males and females (p< 0.005). Conclusion: The results indicated significantly strong correlations between GCS and Ps. This is helpful for the patients in whom RTS scores cannot be calculated, as GCS can be used in place of RTS. Results also showed that gender affects Ps and hence could be used in calculations. There may be a need for a new system to calculate Ps using GCS and gender

Introduction. The Kaplan Meier estimator is widely used in orthopedics. In situations where another event prevents the occurrence of the event of interest, the Kaplan Meier estimator is not appropriate and a competing risks model has to be applied. We questioned how much bias is introduced by erroneous use of the Kaplan Meier estimator instead of a competing risks model in a hip revision surgery cohort. Methods. In our previously published cohort study, 62 acetabular revisions (58 patients) were performed between January 1979 and March 1986. Twenty to twenty-five years after surgery, no patients were lost to follow-up. Thirteen patients underwent revision surgery. During the 20 to 25 years follow-up, 30 patients (33 acetabular revisions) died of causes unrelated to their hip surgery. Results. In the data set analyzed, the Kaplan Meier method overestimates the probability of implant failure by 6.7%, 13.8%,26.8%,48.6% at 5, 10, 15 and 20 years respectively. Discussion. We have performed two different analyses for a hip revision surgery cohort and discussed the use of a competing risks model. Ignoring competing risks leads to biased estimations of the probability of having future revision surgery. Therefore we recommended the use of a competing risks model whenever there are competing risks present

Objectives. The Kaplan-Meier estimation is widely used in orthopedics to calculate the probability of revision surgery. Using data from a long-term follow-up study, we aimed to assess the amount of bias introduced by the Kaplan-Meier estimator in a competing risk setting. Methods. We describe both the Kaplan-Meier estimator and the competing risk model, and explain why the competing risk model is a more appropriate approach to estimate the probability of revision surgery when patients die in a hip revision surgery cohort. In our study, a total of 62 acetabular revisions were performed. After a mean of 25 years, no patients were lost to follow-up, 13 patients had undergone revision surgery and 33 patients died of causes unrelated to their hip. Results. The Kaplan-Meier estimator overestimates the probability of revision surgery in our example by 3%, 11%, 28%, 32% and 60% at five, ten, 15, 20 and 25 years, respectively. As the cumulative incidence of the competing event increases over time, as does the amount of bias. Conclusions. Ignoring competing risks leads to biased estimations of the probability of revision surgery. In order to guide choosing the appropriate statistical analysis in future clinical studies, we propose a flowchart

In recent literature, the fragility index (FI) has been used to evaluate the robustness of statistically significant findings of dichotomous outcomes. This metric is defined as the minimum number of outcome events to flip study conclusions from significant to nonsignificant. Orthopaedics literature is frequently found to be fragile with a median FI of 2 in 150 RCTs across spine, hand, sports medicine, trauma and orthopaedic oncology studies. While many papers discuss limitations of FI, we aimed to further characterize it by introducing the Fragility Likelihood (FL), a new metric that allows us to consider the probability of the event to occur and to calculate the likelihood of this fragility to be reached. We systematically reviewed all randomized controlled trials in the Journal of Bone and Joint Surgery (Am) over 10 years. The FL was calculated with the following formula: A x B x C x 100% (A= FI; B = probability of the event in the group with the smallest number of events; C= probability of the non-event in the group with the highest number of events). A smaller FL demonstrates more robust results and conversely, a larger FL illustrates a higher likelihood of fragility being reached and more fragile the findings. The median FI for the statistically significant outcomes was 2 (Mean: 3.8; Range 0-23). The median FL for the statistically significant outcomes was 11% (Mean: 22%, Range: 2%-73%). This means that the probability of reaching non-significance is only 11% when considering the probability of the event to occur. When comparing studies with the same FI we found the FL to range from 3% to 43%. This illustrates the large differences in robustness between trials with equal FI when the likelihood of the event was taken into consideration. As orthopaedic studies are frequently reported as fragile, we found that by calculating the FL, studies may be more robust than previously assumed based off FI alone. By using the FL in conjunction with FI and p-values will provide additional insight into the robustness of the reported outcomes. Our results indicate that by calculating the FL, study conclusions are stronger than what the FI alone predicts. Although conducting RCTs in surgery can be challenging, we must endeavor to critically evaluate our results so we can answer important orthopaedic questions with certainty

Aims. Iliopsoas impingement occurs in 4% to 30% of patients after undergoing total hip arthroplasty (THA). Despite a relatively high incidence, there are few attempts at modelling impingement between the iliopsoas and acetabular component, and no attempts at modelling this in a representative cohort of subjects. The purpose of this study was to develop a novel computational model for quantifying the impingement between the iliopsoas and acetabular component and validate its utility in a case-controlled investigation. Methods. This was a retrospective cohort study of patients who underwent THA surgery that included 23 symptomatic patients diagnosed with iliopsoas tendonitis, and 23 patients not diagnosed with iliopsoas tendonitis. All patients received postoperative CT imaging, postoperative standing radiography, and had minimum six months’ follow-up. 3D models of each patient’s prosthetic and bony anatomy were generated, landmarked, and simulated in a novel iliopsoas impingement detection model in supine and standing pelvic positions. Logistic regression models were implemented to determine if the probability of pain could be significantly predicted. Receiver operating characteristic curves were generated to determine the model’s sensitivity, specificity, and area under the curve (AUC). Results. Highly significant differences between the symptomatic and asymptomatic cohorts were observed for iliopsoas impingement. Logistic regression models determined that the impingement values significantly predicted the probability of groin pain. The simulation had a sensitivity of 74%, specificity of 100%, and an AUC of 0.86. Conclusion. We developed a computational model that can quantify iliopsoas impingement and verified its accuracy in a case-controlled investigation. This tool has the potential to be used preoperatively, to guide decisions about optimal cup placement, and postoperatively, to assist in the diagnosis of iliopsoas tendonitis. Cite this article: Bone Jt Open 2023;4(1):3–12

Aims. The aim of this trial was to assess the cost-effectiveness of a soft bandage and immediate discharge, compared with rigid immobilization, in children aged four to 15 years with a torus fracture of the distal radius. Methods. A within-trial economic evaluation was conducted from the UK NHS and personal social services (PSS) perspective, as well as a broader societal point of view. Health resources and quality of life (the youth version of the EuroQol five-dimension questionnaire (EQ-5D-Y)) data were collected, as part of the Forearm Recovery in Children Evaluation (FORCE) multicentre randomized controlled trial over a six-week period, using trial case report forms and patient-completed questionnaires. Costs and health gains (quality-adjusted life years (QALYs)) were estimated for the two trial treatment groups. Regression was used to estimate the probability of the new treatment being cost-effective at a range of ‘willingness-to-pay’ thresholds, which reflect a range of costs per QALY at which governments are typically prepared to reimburse for treatment. Results. The offer of a soft bandage significantly reduced cost per patient (saving £12.55 (95% confidence interval (CI) -£5.30 to £19.80)) while QALYs were similar (QALY difference between groups: 0.0013 (95% CI -0.0004 to 0.003)). The high probability (95%) that offering a bandage is a cost-effective option was consistent when examining the data in a range of sensitivity analyses. Conclusion. In addition to the known clinical equivalence, this study found that the offer of a bandage reduced cost compared with rigid immobilization among children with a torus fracture of the distal radius. While the cost saving was small for each patient, the high frequency of these injuries indicates a significant saving across the healthcare system. Cite this article: Bone Joint J 2024;106-B(6):623–630

Aims. To compare the cost-utility of removable brace compared with cast in the management of adult patients with ankle fracture. Methods. A within-trial economic evaluation conducted from the UK NHS and personnel social services (PSS) perspective. Health resources and quality-of-life data were collected as part of the Ankle Injury Rehabilitation (AIR) multicentre, randomized controlled trial over a 12-month period using trial case report forms and patient-completed questionnaires. Cost-utility analysis was estimated in terms of the incremental cost per quality adjusted life year (QALY) gained. Estimate uncertainty was explored by bootstrapping, visualized on the incremental cost-effectiveness ratio plane. Net monetary benefit and probability of cost-effectiveness were evaluated at a range of willingness-to-pay thresholds and visualized graphically. Results. The incremental cost and QALYs of using brace over a 12-month period were £46.73 (95% confidence interval (CI) £-9 to £147) and 0.0141 (95% CI -0.005 to 0.033), respectively. The cost per QALY gained was £3,318. The probability of brace being cost-effective at a £30,000 per QALY willingness-to-pay threshold was 88%. The results remained robust to a range of sensitivity analyses. Conclusion. This within-trial economic evaluation found that it is probable that using a removable brace provides good value to the NHS when compared to cast, in the management of adults with ankle fracture. Cite this article: Bone Jt Open 2022;3(6):455–462

Successful estimation of postoperative PROMs prior to a joint replacement surgery is important in deciding the best treatment option for a patient. However, estimation of the outcome is associated with substantial noise around individual prediction. Here, we test whether a classifier neural network can be used to simultaneously estimate postoperative PROMs and uncertainty better than current methods. We perform Oxford hip score (OHS) estimation using data collected by the NJR from 249,634 hip replacement surgeries performed from 2009 to 2018. The root mean square error (RMSE) of the various methods are compared to the standard deviation of outcome change distribution to measure the proportion of the total outcome variability that the model can capture. The area under the curve (AUC) for the probability of the change score being above a certain threshold was also plotted. The proposed classifier NN had a better or equivalent RMSE than all other currently used models. The threshold AUC shows similar results for all methods close to a change score of 20 but demonstrates better accuracy of the classifier neural network close to 0 change and greater than 30 change, showing that the full probability distribution performed by the classifier neural network resulted in a significant improvement in estimating the upper and lower quantiles of the change score probability distribution. Consequently, probabilistic estimation as performed by the classifier NN is the most adequate approach to this problem, since the final score has an important component of uncertainty. This study shows the importance of uncertainty estimation to accompany postoperative PROMs prediction and presents a clinically-meaningful method for personalised outcome that includes such uncertainty estimation

Aims. A higher failure rate has been reported in haematogenous periprosthetic joint infection (PJI) compared to non-haematogenous PJI. The reason for this difference is unknown. We investigated the outcome of haematogenous and non-haematogenous PJI to analyze the risk factors for failure in both groups of patients. Methods. Episodes of knee or hip PJI (defined by the European Bone and Joint Infection Society criteria) treated at our institution between January 2015 and October 2020 were included in a retrospective PJI cohort. Episodes with a follow-up of > one year were stratified by route of infection into haematogenous and non-haematogenous PJI. Probability of failure-free survival was estimated using the Kaplan-Meier method, and compared between groups using log-rank test. Univariate and multivariate analysis was applied to assess risk factors for failure. Results. A total of 305 PJI episodes (174 hips, 131 knees) were allocated to the haematogenous (n = 146) or the non-haematogenous group (n = 159). Among monomicrobial infections, Staphylococcus aureus was the dominant pathogen in haematogenous PJI (76/140, 54%) and coagulase-negative staphylococci in non-haematogenous PJI (57/133, 43%). In both groups, multi-stage exchange (n = 55 (38%) in haematogenous and n = 73 (46%) in non-haematogenous PJI) and prosthesis retention (n = 70 (48%) in haematogenous and n = 48 (30%) in non-haematogenous PJI) were the most common surgical strategies. Median duration of antimicrobial treatment was 13.5 weeks (range, 0.5 to 218 weeks) and similar in both groups. After six years of follow-up, the probability of failure-free survival was significantly lower in haematogenous compared to non-haematogenous PJI (55% vs 74%; p = 0.021). Infection-related mortality was significantly higher in haematogenous than non-haematogenous PJI (7% vs 0% episodes; p = 0.001). Pathogenesis of failure was similar in both groups. Retention of the prosthesis was the only independent risk factor for failure in multivariate analysis in both groups. Conclusion. Treatment failure was significantly higher in haematogenous compared to non-haematogenous PJI. Retention of the prosthesis was the only independent risk factor for failure in both groups. Cite this article: Bone Joint J 2023;105-B(12):1294–1302

Implants in total hip replacement (THR) are associated with different clinical and cost-effectiveness profiles,. We estimate the costs and outcomes for NHS patients in the year after THR associated with implant bearing materials using linked routinely collected data. We linked NJR primary elective THR patients for osteoarthritis to HES and National PROMs. We estimated health care costs, health-related quality of life indices, and revision risks, in the year after primary and revision THRs overall. We used generalised linear models adjusting for patient and hospital characteristics and estimated 10-year cumulative probability of revision. We imputed utilities using chained equations for half the sample with missing PROMS. We linked 577,973 elective primary THRs and 11,812 subsequent revisions. One year after primary THR, patients with the cemented THRs using cobalt chrome or stainless steel head with HCLPE liner/cup cost the NHS, on average, £13,101 (95%CI £13,080,£13,122), had an average quality-of-life score of 0.788 (95%CI 0.787,0.788), and a 10-year revision probability of 1.9% (95%CI 1.6,2.3). Compared to the reference, patients receiving a cemented THR with delta ceramic head and HCLPE liner/cup, hybrid THR with delta ceramic head and HCLPE liner/cup, and hybrid THR with alumina head and HCLPE liner/cup had lower 1-year costs (-£572 \[95% CI -£775,-£385\], -£346 \[-£501,-£192\], -£371 \[-£574,-£168\] respectively), better quality of life (0.007 \[95% CI 0.003,0.011\], 0.013 \[0.010,0.016\], 0.009 \[0.005,0.013\] respectively), and lower 10-year revision probabilities (1.4% \[1.03,2.0\], 1.5 \[1.3,1.7\], 1.6%\[1.2,2.1\] respectively). Implant bearing materials are associated with varying mean costs and health outcomes after primary THR. Ours is the first study to derive costs and health outcomes from large, linked databases using multiple imputation methods to deal with bias. Our findings are useful for commissioning and procurement decisions and to inform a subsequent cost-effectiveness model with more granular detail on THR implant types

Aims. The aim of this study was to estimate time to arthroplasty among patients with hip and knee osteoarthritis (OA), and to identify factors at enrolment to first-line intervention that are prognostic for progression to surgery. Methods. In this longitudinal register-based observational study, we identified 72,069 patients with hip and knee OA in the Better Management of Patients with Osteoarthritis Register (BOA), who were referred for first-line OA intervention, between May 2008 and December 2016. Patients were followed until the first primary arthroplasty surgery before 31 December 2016, stratified into a hip and a knee OA cohort. Data were analyzed with Kaplan-Meier and multivariable-adjusted Cox regression. Results. At five years, Kaplan-Meier estimates showed that 46% (95% confidence interval (CI) 44.6 to 46.9) of those with hip OA, and 20% (95% CI 19.7 to 21.0) of those with knee OA, had progressed to arthroplasty. The strongest prognostic factors were desire for surgery (hazard ratio (HR) hip 3.12 (95% CI 2.95 to 3.31), HR knee 2.72 (95% CI 2.55 to 2.90)), walking difficulties (HR hip 2.20 (95% CI 1.97 to 2.46), HR knee 1.95 (95% CI 1.73 to 2.20)), and frequent pain (HR hip 1.56 (95% CI 1.40 to 1.73), HR knee 1.77 (95% CI 1.58 to 2.00)). In hip OA, the probability of progression to surgery was lower among those with comorbidities (e.g. ≥ four conditions; HR 0.64 (95% CI 0.59 to 0.69)), with no detectable effects in the knee OA cohort. Instead, being overweight or obese increased the probability of OA progress in the knee cohort (HR 1.25 (95% CI 1.15 to 1.37)), but not among those with hip OA. Conclusion. Patients with hip OA progressed faster and to a greater extent to arthroplasty than patients with knee OA. Progression was strongly influenced by patients’ desire for surgery and by factors related to severity of OA symptoms, but factors not directly related to OA symptoms are also of importance. However, a large proportion of patients with OA do not seem to require surgery within five years, especially among those with knee OA. Cite this article: Bone Joint J 2022;104-B(7):792–800

Aims. We report the natural course of Baker’s cysts following total knee arthroplasty (TKA) at short- and mid-term follow-up. Methods. In this prospective case series, 105 TKA patients were included. All patients who received surgery had a diagnosis of primary osteoarthritis and had preoperatively presented with a Baker’s cyst. Sonography and MRI were performed to evaluate the existence and the gross size of the cyst before TKA, and sonography was repeated at a mean follow-up time of 1.0 years (0.8 to 1.3; short-term) and 4.9 years (4.0 to 5.6; mid-term) after TKA. Symptoms potentially attributable to the Baker’s cyst were recorded at each assessment. Results. At the one-year follow-up analysis, 102 patients were available. Of those, 91 patients were available for the 4.9-year assessment (with an 86.7% follow-up rate (91/105)). At the short- and mid-term follow-up, a Baker’s cyst was still present in 87 (85.3%) and 30 (33.0%) patients, respectively. Of those patients who retained a Baker’s cyst at the short-term follow-up, 31 patients (35.6%) had popliteal symptoms. Of those patients who continued to have a Baker’s cyst at the mid-term follow-up, 17 patients (56.7%) were still symptomatic. The mean preoperative cyst size was 14.5 cm. 2. (13.1 to 15.8). At the short- and mid-term follow-up, the mean cyst size was 9.7 cm. 2. (8.3 to 11.0) and 10.4 cm. 2. (9.8 to 11.4), respectively. A significant association was found between the size of the cyst at peroperatively and the probability of resolution, with lesions smaller than the median having an 83.7% (36/43) probability of resolution, and larger lesions having a 52.1% (25/48) probability of resolution (p < 0.001). At the mid-term follow-up, no association between cyst size and popliteal symptoms was found. Conclusion. At a mean follow-up of 4.9 years (4.0 to 5.6) after TKA, the majority (67.0%, 61/91) of the Baker’s cysts that were present preoperatively had disappeared. The probability of cyst resolution was dependent on the size of the Baker’s cyst at baseline, with an 83.7% (36/43) probability of resolution for smaller cysts and 52.1% (25/48) probability for larger cysts. Cite this article: Bone Joint J. 2020;102-B(1):132–136

Revision Total Knee Arthroplasty (rTKA) is predicted to increase by more than 600% between 2005 and 2030. The survivorship of primary TKA has been extensively investigated, however more granular information on the risks of rTKA is needed. The aim of the study was to investigate the incidence of re-revision TKA, with explanatory variables of time from primary to revision, and indication (aseptic vs septic). Secondary aim was to investigate mortality. This is an analysis of the Scottish Arthroplasty Project data set, a national audit prospectively recording data on all joint replacements performed in Scotland. The period from 2000 to 2019 was studied. 4723 patients underwent revision TKA. The relationship between time from primary to revision TKA and 2nd revision was significant (p<0.001), with increasing time lowering probability of re-revision (OR 0.99 95% CI 0.987 to 0.993). There was no significant association in time to first revision on time from 1st revision to re-revision (p>0.05). Overall mortality for all patients was 32% at 10 years (95% CI 31-34), Time from primary TKA to revision TKA had a significant effect on mortality: p=0.004 OR 1.03 (1.01-1.05). Septic revisions had a reduced mortality compared to aseptic, OR 0.95 (0.71-1.25) however this was not significant (p=0.69). This is the first study to demonstrate time from primary TKA to revision TKA having a significant effect on probability of re-revision TKA. Furthermore the study suggests mortality is increased with increasing time from primary procedure to revision, however decreased if the indication is septic rather than aseptic

Economic evaluation provides a framework for assessing the costs and consequences of alternative programmes or interventions. One common vehicle for economic evaluations in the healthcare context is the decision-analytic model, which synthesizes information on parameter inputs (for example, probabilities or costs of clinical events or health states) from multiple sources and requires application of mathematical techniques, usually within a software program. A plethora of decision-analytic modelling-based economic evaluations of orthopaedic interventions have been published in recent years. This annotation outlines a number of issues that can help readers, reviewers, and decision-makers interpret evidence from decision-analytic modelling-based economic evaluations of orthopaedic interventions. Cite this article: Bone Joint J 2023;105-B(1):17–20

Aims. This study was designed to develop a model for predicting bone mineral density (BMD) loss of the femur after total hip arthroplasty (THA) using artificial intelligence (AI), and to identify factors that influence the prediction. Additionally, we virtually examined the efficacy of administration of bisphosphonate for cases with severe BMD loss based on the predictive model. Methods. The study included 538 joints that underwent primary THA. The patients were divided into groups using unsupervised time series clustering for five-year BMD loss of Gruen zone 7 postoperatively, and a machine-learning model to predict the BMD loss was developed. Additionally, the predictor for BMD loss was extracted using SHapley Additive exPlanations (SHAP). The patient-specific efficacy of bisphosphonate, which is the most important categorical predictor for BMD loss, was examined by calculating the change in predictive probability when hypothetically switching between the inclusion and exclusion of bisphosphonate. Results. Time series clustering allowed us to divide the patients into two groups, and the predictive factors were identified including patient- and operation-related factors. The area under the receiver operating characteristic (ROC) curve (AUC) for the BMD loss prediction averaged 0.734. Virtual administration of bisphosphonate showed on average 14% efficacy in preventing BMD loss of zone 7. Additionally, stem types and preoperative triglyceride (TG), creatinine (Cr), estimated glomerular filtration rate (eGFR), and creatine kinase (CK) showed significant association with the estimated patient-specific efficacy of bisphosphonate. Conclusion. Periprosthetic BMD loss after THA is predictable based on patient- and operation-related factors, and optimal prescription of bisphosphonate based on the prediction may prevent BMD loss. Cite this article: Bone Joint Res 2024;13(4):184–192

Successful estimation of postoperative PROMs prior to a joint replacement surgery is important in deciding the best treatment option for a patient. However, estimation of the outcome is associated with substantial noise around individual prediction. Here, we test whether a classifier neural network can be used to simultaneously estimate postoperative PROMs and uncertainty better than current methods. We perform Oxford hip score (OHS) estimation using data collected by the NJR from 249,634 hip replacement surgeries performed from 2009 to 2018. The root mean square error (RMSE) of the various methods are compared to the standard deviation of outcome change distribution to measure the proportion of the total outcome variability that the model can capture. The area under the curve (AUC) for the probability of the change score being above a certain threshold was also plotted. The proposed classifier NN had a better or equivalent RMSE than all other currently used models. The standard deviation for the change score for the entire population was 9.93, which can be interpreted as the RMSE that would be achieved for a model that gives the same estimation for all patients regardless of the covariates. However, most of the variation in the postoperative OHS/OKS change score is not captured by the models, confirming the importance of accurate uncertainty estimation. The threshold AUC shows similar results for all methods close to a change score of 20 but demonstrates better accuracy of the classifier neural network close to 0 change and greater than 30 change, showing that the full probability distribution performed by the classifier neural network resulted in a significant improvement in estimating the upper and lower quantiles of the change score probability distribution. Consequently, probabilistic estimation as performed by the classifier NN is the most adequate approach to this problem, since the final score has an important component of uncertainty. This study shows the importance of uncertainty estimation to accompany postoperative PROMs prediction and presents a clinically-meaningful method for personalised outcome that includes such uncertainty estimation

Approximately 20% of patients feel unsatisfied 12 months after primary total knee arthroplasty (TKA). Current predictive tools for TKA focus on the clinician as the intended user rather than the patient. The aim of this study is to develop a tool that can be used by patients without clinician assistance, to predict health-related quality of life (HRQoL) outcomes 12 months after total knee arthroplasty (TKA). All patients with primary TKAs for osteoarthritis between 2012 and 2019 at a tertiary institutional registry were analysed. The predictive outcome was improvement in Veterans-RAND 12 utility score at 12 months after surgery. Potential predictors included patient demographics, co-morbidities, and patient reported outcome scores at baseline. Logistic regression and three machine learning algorithms were used. Models were evaluated using both discrimination and calibration metrics. Predictive outcomes were categorised into deciles from 1 being the least likely to improve to 10 being the most likely to improve. 3703 eligible patients were included in the analysis. The logistic regression model performed the best in out-of-sample evaluation for both discrimination (AUC = 0.712) and calibration (gradient = 1.176, intercept = -0.116, Brier score = 0.201) metrics. Machine learning algorithms were not superior to logistic regression in any performance metric. Patients in the lowest decile (1) had a 29% probability for improvement and patients in the highest decile (10) had an 86% probability for improvement. Logistic regression outperformed machine learning algorithms in this study. The final model performed well enough with calibration metrics to accurately predict improvement after TKA using deciles. An ongoing randomised controlled trial (ACTRN12622000072718) is evaluating the effect of this tool on patient willingness for surgery. Full results of this trial are expected to be available by April 2023. A free-to-use online version of the tool is available at . smartchoice.org.au.

Swelling following an ankle fracture is commonly believed to preclude surgical fixation, delaying operative treatment to allow the swelling to subside. This is in an attempt to achieve better soft tissue outcomes. We aim to identify whether pre-operative ankle swelling influences postoperative wound complications following ankle fracture surgery. This is a prospective cohort study of 80 patients presenting to a tertiary referral centre with operatively managed malleolar ankle fractures. Ankle swelling was measured visually and then quantitatively using the validated ‘Figure-of-eight’ technique. Follow-up was standardised at 2, 6, and 12 weeks post-operatively. Wound complications, patient co-morbidities, operative time, surgeon experience, and hospital stay duration were recorded. The complication rate was 8.75% (n=7), with 1 deep infection requiring operative intervention and all others resolving with oral antibiotics and wound cares. There was no significant difference in wound complication rates associated with quantitative ankle swelling (p=0.755), visual assessment of ankle swelling (p=0.647), or time to operative intervention (p=0.270). Increasing age (p=0.006) and female gender (p=0.049) had a significantly greater probability of wound complications. However, BMI, smoking status, level of the operating surgeon, and tourniquet time were not significantly different. Visual assessment of ankle swelling had a poor to moderate correlation to ‘Figure-of-eight’ ankle swelling measurements ICC=0.507 (0.325- 0.653). Neither ankle swelling nor time to surgery correlates with an increased risk of postoperative wound complication in surgically treated malleolar ankle fractures. Increasing patient age and female gender had a significantly greater probability of wound infection, irrespective of swelling. Visual assessment of ankle swelling is unreliable for quantifying true ankle swelling. Operative intervention at any time after an ankle fracture, irrespective of swelling, is safe and showed no better or worse soft tissue outcomes than those delayed for swelling

Aims. To calculate how the likelihood of obtaining measurable benefit from hip or knee arthroplasty varies with preoperative patient-reported scores. Methods. Existing UK data from 222,933 knee and 209,760 hip arthroplasty patients were used to model an individual’s probability of gaining meaningful improvement after surgery based on their preoperative Oxford Knee or Hip Score (OKS/OHS). A clinically meaningful improvement after arthroplasty was defined as ≥ 8 point improvement in OHS, and ≥ 7 in OKS. Results. The upper preoperative score threshold, above which patients are unlikely to achieve any meaningful improvement from surgery, is 41 for knees and 40 for hips. At lower scores, the probability of improvement increased towards a maximum of 88% (knee) and 95% for (hips). Conclusion. By our definition of meaningful improvement, patients with preoperative scores above 41 (OKS) and 40 (OHS) should not be routinely referred to secondary care for possible arthroplasty. Using lower thresholds would incrementally increase the probability of meaningful benefit for those referred but will exclude some patients with potential to benefit. The findings are useful to support the complex shared decision-making process in primary care for referral to secondary care; and in secondary care for experienced clinicians counselling patients considering knee or hip arthroplasty, but should not be used in isolation. Cite this article: Bone Joint J 2020;102-B(7):941–949