Despite the vast quantities of published artificial intelligence (AI) algorithms that target trauma and orthopaedic applications, very few progress to inform clinical practice. One key reason for this is the lack of a clear pathway from development to deployment. In order to assist with this process, we have developed the Clinical Practice Integration of Artificial Intelligence (CPI-AI) framework – a five-stage approach to the clinical practice adoption of AI in the setting of trauma and orthopaedics, based on the IDEAL principles ( Cite this article:
To examine whether natural language processing (NLP) using a clinically based large language model (LLM) could be used to predict patient selection for total hip or total knee arthroplasty (THA/TKA) from routinely available free-text radiology reports. Data pre-processing and analyses were conducted according to the Artificial intelligence to Revolutionize the patient Care pathway in Hip and knEe aRthroplastY (ARCHERY) project protocol. This included use of de-identified Scottish regional clinical data of patients referred for consideration of THA/TKA, held in a secure data environment designed for artificial intelligence (AI) inference. Only preoperative radiology reports were included. NLP algorithms were based on the freely available GatorTron model, a LLM trained on over 82 billion words of de-identified clinical text. Two inference tasks were performed: assessment after model-fine tuning (50 Epochs and three cycles of k-fold cross validation), and external validation.Aims
Methods
To examine whether Natural Language Processing (NLP) using a state-of-the-art clinically based Large Language Model (LLM) could predict patient selection for Total Hip Arthroplasty (THA), across a range of routinely available clinical text sources. Data pre-processing and analyses were conducted according to the Ai to Revolutionise the patient Care pathway in Hip and Knee arthroplasty (ARCHERY) project protocol ( There were 3911, 1621 and 1503 patient text documents included from the sources of referral letters, radiology reports and clinic letters respectively. All letter sources displayed significant class imbalance, with only 15.8%, 24.9%, and 5.9% of patients linked to the respective text source documentation having undergone surgery. Untrained model performance was poor, with F1 scores (harmonic mean of precision and recall) of 0.02, 0.38 and 0.09 respectively. This did however improve with model training, with mean scores (range) of 0.39 (0.31–0.47), 0.57 (0.48–0.63) and 0.32 (0.28–0.39) across the 5 folds of cross-validation. Performance deteriorated on external validation across all three groups but remained highest for the radiology report cohort. Even with further training on a large cohort of routinely collected free-text data a clinical LLM fails to adequately perform clinical inference in NLP tasks regarding identification of those selected to undergo THA. This likely relates to the complexity and heterogeneity of free-text information and the way that patients are determined to be surgical candidates.
Given the prolonged waits for hip arthroplasty seen across the U.K. it is important that we optimise priority systems to account for potential disparities in patient circumstances and impact. We set out to achieve this through a two-stage approach. This included a Delphi-study of patient and surgeon preferences to determine what should be considered when determining patient priority, followed by a Discrete Choice Experiment (DCE) to decide relative weighting of included attributes. The study was conducted according to the published protocol ([ For the Delphi study there were 43 responses in the first round, with a subsequent 91% participation rate. Final consensus inclusion was achieved for Pain; Mobility/Function; Activities of Daily Living; Inability to Work/Care; Length of Time Waited; Radiological Severity and Mental Wellbeing. 70 individuals subsequently contributed to the DCE, with radiological severity being the most significant factor (Coefficient 2.27 \[SD 0.31\], p<0.001), followed by pain (Coefficient 1.08 \[SD 0.13\], p<0.001) and time waited (Coefficient for 1-month additional wait 0.12 \[SD 0.02\], p<0.001). The calculated trade-off in waiting time for a 1-level change in pain (e.g., moderate to severe pain) was 9.14 months. These results present a new method of determining comparative priority for those on primary hip arthroplasty waiting lists. Evaluation of potential implementation in clinical practice is now required.
This study aimed to determine whether lateral femoral wall thickness (LWT) < 20.5 mm was associated with increased revision risk of intertrochanteric fracture (ITF) of the hip following sliding hip screw (SHS) fixation when the medial calcar was intact. Additionally, the study assessed the association between LWT and patient mortality. This retrospective study included ITF patients aged 50 years and over treated with SHS fixation between 2019 and 2021 at a major trauma centre. Demographic information, fracture type, delirium status, American Society of Anesthesiologists grade, and length of stay were collected. LWT and tip apex distance were measured. Revision surgery and mortality were recorded at a mean follow-up of 19.5 months (1.6 to 48). Cox regression was performed to evaluate independent risk factors associated with revision surgery and mortality.Aims
Methods
Periprosthetic femur fracture (PPF) are heterogeneous, complex, and thought to be increasingly prevalent. The aims were to evaluate PPF prevalence, casemix, management, and outcomes. This nationwide study included all PPF patients aged >50 years from 16 Scottish hospitals in 2019. Variables included: demographics; implant and fracture factors; management factors, and outcomes. There were 332 patients, mean age 79.5 years, and 220/332 (66.3%) were female. One-third (37.3%) were ASA1-2 and two-thirds (62.3%) were ASA3+, 91.0% were from home/sheltered housing, and median Clinical Frailty Score was 4.0 (IQR 3.0). Acute medical issues featured in 87/332 (26.2%) and 19/332 (5.7%) had associated injuries. There were 251/332 (75.6%) associated with a proximal femoral implant, of which 232/251 (92.4%) were arthroplasty devices (194/251 [77.3%] total hip, 35/251 [13.9%] hemiarthroplasty, 3/251 [1.2%] resurfacing). There were 81/332 (24.4%) associated with a distal femoral implant (76/81 [93.8%] were total knee arthroplasties). In 38/332 (11.4%) there were implants proximally and distally. Most patients (268/332; 80.7%) were treated surgically, with 174/268 (64.9%) requiring fixation only and 104/268 (38.8%) requiring an arthroplasty or combined solution. Median time to theatre was longer for arthroplasty versus fixation procedures (120 vs 46 hours), and those requiring inter-hospital transfer waited longer (94 vs 48 hours). Barriers to investigating PPF include varied classification, coding challenges, and limitations of existing registries. This is the first study to examine a national PPF cohort and presents important data to guide service design and research. Additional findings relating to fracture patterns, implant types, surgeon skill-mix, and outcomes are reported herein.
Current levels of hip fracture morbidity contribute greatly to the overall burden on health and social care services. Given the anticipated ageing of the population over the coming decade, there is potential for this burden to increase further, although the exact scale of impact has not been identified in contemporary literature. We therefore set out to predict the future incidence of hip fracture and help inform appropriate service provision to maintain an adequate standard of care. Historical data from the Scottish Hip Fracture Audit (2017 to 2021) were used to identify monthly incidence rates. Established time series forecasting techniques (Exponential Smoothing and Autoregressive Integrated Moving Average) were then used to predict the annual number of hip fractures from 2022 to 2029, including adjustment for predicted changes in national population demographics. Predicted differences in service-level outcomes (length of stay and discharge destination) were analyzed, including the associated financial cost of any changes.Aims
Methods
Surgery is often delayed in patients who sustain a hip fracture and are treated with a total hip arthroplasty (THA), in order to await appropriate surgical expertise. There are established links between delay and poorer outcomes in all patients with a hip fracture, but there is little information about the impact of delay in the less frail patients who undergo THA. The aim of this study was to investigate the influence of delayed surgery on outcomes in these patients. A retrospective cohort study was undertaken using data from the Scottish Hip Fracture Audit between May 2016 and December 2020. Only patients undergoing THA were included, with categorization according to surgical treatment within 36 hours of admission (≤ 36 hours = ‘acute group’ vs > 36 hours = ‘delayed’ group). Those with delays due to being “medically unfit” were excluded. The primary outcome measure was 30-day survival. Costs were estimated in relation to the differences in the lengths of stay.Aims
Methods
The principles of evidence-based medicine (EBM) are the foundation of modern medical practice. Surgeons are familiar with the commonly used statistical techniques to test hypotheses, summarize findings, and provide answers within a specified range of probability. Based on this knowledge, they are able to critically evaluate research before deciding whether or not to adopt the findings into practice. Recently, there has been an increased use of artificial intelligence (AI) to analyze information and derive findings in orthopaedic research. These techniques use a set of statistical tools that are increasingly complex and may be unfamiliar to the orthopaedic surgeon. It is unclear if this shift towards less familiar techniques is widely accepted in the orthopaedic community. This study aimed to provide an exploration of understanding and acceptance of AI use in research among orthopaedic surgeons. Semi-structured in-depth interviews were carried out on a sample of 12 orthopaedic surgeons. Inductive thematic analysis was used to identify key themes.Aims
Methods
Hip fracture represents a significant challenge, placing increasing pressure on health and social care services in Scotland. This study establishes the ‘historic’ hip fracture burden, namely, the annual number of hip fractures in Scotland, and respective incidence, between 2017 – 2021. Furthermore, the ‘projected’ hip fracture burden and incidence from 2022 – 2029 was estimated, to inform future capacity and funding of health and social care services. The number of individuals with a hip fracture in Scotland between 2017 and 2021 was identified through the Scottish Hip Fracture Audit, enabling the annual number of hip fractures and respective incidence between 2017 – 2021 to be calculated. Projection modelling was performed using Exponential Smoothing and Auto Regressive Integrated Moving Average to estimate the number of hip fractures occurring annually from 2022 – 2029. A combined average projection was employed to provide a more accurate forecast. Accounting for predicted changes within the population demographics of Scotland, the projected hip fracture incidence up to 2029 was calculated. Between 2017 and 2021 the annual number of hip fractures in Scotland increased from 6675 to 7797 (15%), with an increase in incidence from 313 to 350 per 100,000 (11%) of the at-risk population. Hip fracture was observed to increase across all groups, notably males, and the 70–79 and 80–89 age cohorts. By 2029, the combined average projection estimated the annual number of hip fractures at 10311, with an incidence rate of 463 per 100,000, representing a 32% increase from 2021. The largest percentage increase in hip fracture by 2029 occurs in the 70–79 and 80–89 age cohorts (57% and 53% respectively). Based upon these projections, overall length of hospital stay following hip fracture will increase by 60699 days per annum by 2029, incurring an additional cost of at least £25 million. Projection modelling demonstrates the annual number of hip fractures in Scotland will increase substantially by 2029, with significant implications for health and social care services. This increase in hip fracture burden and incidence is influenced strongly by changing population demographics, primarily an ageing population.
Appropriate surgical management of hip fractures has major clinical and economic consequences. Recently IMN use has increased compared to SHS constructs, despite no clear evidence demonstrating superiority of outcome. We therefore set out to provide further evidence about the clinical and economic implications of implant choice when considering hip fracture fixation strategies. A retrospective cohort study using Scottish hip fracture audit (SHFA) data was performed for the period 2016–2022. Patients ≥50 with a hip fracture and treated with IMN or SHS constructs at Scottish Hospitals were included. Comparative analyses, including adjustment for confounders, were performed utilising Multivariable logistic regression for dichotomous outcomes and Mann-Whitney-U tests for non-parametric data. A sub-group analysis was also performed focusing on AO-A1/A2 configurations which utilised additional regional data. Cost differences in Length of Stay (LOS) were calculated using defined costs from the NHS Scotland Costs book. In all analyses p<0.05 denoted significance. 13638 records were included (72% female). 9867 received a SHS (72%). No significant differences were identified in 30 or 60-day survival (Odds Ratio [OR] 1.05, 95%CI 0.90–1.23; p=0.532), (OR 1.10, 95%CI 0.97–1.24; p=0.138) between SHS and IMN's. There was however a significantly lower early mobilisation rate with IMN vs SHS (OR 0.64, 95%CI 0.59–0.70; p<0.001), and lower likelihood of discharge to domicile by day-30 post-admission (OR 0.77, 95%CI 0.71–0.84; p<0.001). Acute and overall, LOS were significantly lower for SHS vs IMN (11 vs 12 days and 20 vs 24 days respectively; p<0.001). Findings were similar across a sub-group analysis of 559 AO A1/A2 fracture configurations. Differences in LOS potentially increases costs by £1230 per-patient, irrespective of the higher costs of IMN's v SHS. Appropriate SHS use is associated with early mobilisation, reduced LOS and likely with reduced cost of treatment. Further research exploring potential reasons for the identified differences in early mobilisation are warranted.
Cite this article:
The hip fracture burden on health and social care services in Scotland is anticipated to increase significantly, primarily driven by an ageing population. This study forecasts future hip fracture incidence and the annual number of hip fractures in Scotland until 2029. The monthly number of patients with hip fracture aged ≥ 50 admitted to a Scottish hospital between 01/01/2017 and 31/12/2021 was identified through data collected by the Scottish Hip Fracture Audit. This data was analysed using Exponential Smoothing and Auto Regressive Integrated Moving Average forecast modelling to project future hip fracture incidence and the annual number of hip fractures until 2029. Adjustments for population change were accounted for by integrating population projections published by National Records of Scotland. Between 2017 and 2021 the annual number of hip fractures in Scotland increased from 6675 to 7797, with a respective increase in hip fracture incidence from 313 to 350 per 100,000. By 2029, the averaged projected annual number of hip fractures is 10311, with an incidence rate of 463 per 100,000. The largest percentage increase in hip fracture occurs in the 70-79 age group (57%), with comparable increases in both sexes (30%). Based upon these projections, overall length of stay following hip fracture will increase from 142713 bed days per annum in 2021, to 203412 by 2029, incurring an additional cost of over £25 million. Forecast modelling demonstrates that the annual number of hip fractures in Scotland will rise substantially by 2029, with considerable implications for health and social care services.
Cite this article:
The aim of this study was to examine perioperative blood transfusion practice, and associations with clinical outcomes, in a national cohort of hip fracture patients. A retrospective cohort study was undertaken using linked data from the Scottish Hip Fracture Audit and the Scottish National Blood Transfusion Service between May 2016 and December 2020. All patients aged ≥ 50 years admitted to a Scottish hospital with a hip fracture were included. Assessment of the factors independently associated with red blood cell transfusion (RBCT) during admission was performed, alongside determination of the association between RBCT and hip fracture outcomes.Aims
Methods
The extended wait that most patients are now experiencing for hip and knee arthroplasty has raised questions about whether reliance on waiting time as the primary driver for prioritization is ethical, and if other additional factors should be included in determining surgical priority. Our Prioritization of THose aWaiting hip and knee ArthroplastY (PATHWAY) project will explore which perioperative factors are important to consider when prioritizing those on the waiting list for hip and knee arthroplasty, and how these factors should be weighted. The final product will include a weighted benefit score that can be used to aid in surgical prioritization for those awaiting elective primary hip and knee arthroplasty. There will be two linked work packages focusing on opinion from key stakeholders (patients and surgeons). First, an online modified Delphi process to determine a consensus set of factors that should be involved in patient prioritization. This will be performed using standard Delphi methodology consisting of multiple rounds where following initial individual rating there is feedback, discussion, and further recommendations undertaken towards eventual consensus. The second stage will then consist of a Discrete Choice Experiment (DCE) to allow for priority setting of the factors derived from the Delphi through elicitation of weighted benefit scores. The DCE consists of several choice tasks designed to elicit stakeholder preference regarding included attributes (factors).Aims
Methods
There is increasing popularity in the use of artificial intelligence and machine-learning techniques to provide diagnostic and prognostic models for various aspects of Trauma & Orthopaedic surgery. However, correct interpretation of these models is difficult for those without specific knowledge of computing or health data science methodology. Lack of current reporting standards leads to the potential for significant heterogeneity in the design and quality of published studies. We provide an overview of machine-learning techniques for the lay individual, including key terminology and best practice reporting guidelines. Cite this article:
The aim of this study was to determine the impact of hospital-level service characteristics on hip fracture outcomes and quality of care processes measures. This was a retrospective analysis of publicly available audit data obtained from the National Hip Fracture Database (NHFD) 2018 benchmark summary and Facilities Survey. Data extraction was performed using a dedicated proforma to identify relevant hospital-level care process and outcome variables for inclusion. The primary outcome measure was adjusted 30-day mortality rate. A random forest-based multivariate imputation by chained equation (MICE) algorithm was used for missing value imputation. Univariable analysis for each hospital level factor was performed using a combination of Tobit regression, Siegal non-parametric linear regression, and Mann-Whitney U test analyses, dependent on the data type. In all analyses, a p-value < 0.05 denoted statistical significance.Aims
Methods
