Andrew Judge

Objectives: To evaluate the early comparative effectiveness of robotic versus conventional total hip replacement (THR) in the UK. Design: Target trial emulation study. Setting: Public and private hospitals using data from the National Joint Registry of England, Wales, Northern Ireland, the Isle of Man, Guernsey, and Jersey. Participants: 666 283 THRs performed between 2018 and 2024; 656 080 were conventional THRs and 10 203 were robotic THRs. Main outcome measures: Five year THR implant survival (mean 2.5 year follow-up), all cause revision risk, cause specific revision risk, malpositioning revision risk, intraoperative complications, and patient survival up to five years. Kaplan-Meier survival analysis, Cox proportional hazards regression, and Fine and Gray models were used. Propensity score matching created comparable treatment groups to estimate the average treatment effect in the treated population. Results: The five year implant survival of the matched conventional and robotic THR groups was 98.8% (95% confidence interval (CI) 98.6% to 99.0%) and 98.8% (CI 98.3% to 99.1%), respectively, with no difference in revision risk between groups (hazard ratio 0.96, 95% CI 0.75 to 1.22; P=0.73). Cause specific revision risk and intraoperative complication risks did not differ between groups. However, a significantly lower risk of revisions related to implant malposition (dislocation, leg length discrepancy, and malalignment) was observed in the robotic group (hazard ratio 0.53, 95% CI 0.30 to 0.93; P=0.03). The five year patient survival of the matched conventional and robotic THR groups was 96.3% (95% CI 95.8% to 96.8%) and 96.2% (95.2% to 97.0%), respectively, with no difference in all cause mortality risk (hazard ratio 1.12, 95% CI 0.95 to 1.34; P=0.17). Conclusions: No significant differences were found in overall THR or patient survival, cause specific revision, or intraoperative complication risks between robotic and conventional groups. Robotic THR was associated with a lower revision risk from causes related to implant malpositioning. However, the possibility of unmeasured and residual confounding cannot be ruled out owing to the observational design. These results highlight the importance of careful evaluation of robotic technology in publicly funded healthcare systems because of the substantially higher capital and procedural costs.

Objective: To evaluate the early comparative effectiveness of robotic versus conventional total knee replacement (TKR) and unicompartmental knee replacement (UKR) in the UK. Design: Target trial emulation study. Setting: Public and private hospitals using data from the National Joint Registry of England, Wales, Northern Ireland, the Isle of Man, Guernsey, and Jersey. Participants: 697 145 knee replacement surgeries between 2018 and 2024: 675 034 were conventional (TKRs and UKRs) and 22 111 were robotic. Main outcome measures: The main outcome was five year knee replacement survival (mean 2.5 years follow-up); all cause revision risk; cause specific revision risk; intraoperative complications; and revision complexity of robotic versus conventional (TKR and UKR) knee replacement. Kaplan-Meier and Cox regression analyses were used to compare implant survival and indications for revision surgery. Propensity score matching using target trial emulation framework was used to estimate the average treatment effect in the treated population. Results: The five year implant survival of the matched conventional and robotic TKR groups was 98.5% (95% confidence interval (CI) 98.3% to 98.7%) and 98.6% (98.2% to 99.0%), respectively, with no between group difference in revision risk (hazard ratio 1.03, 95% CI 0.86 to 1.26; P=0.71). The five year implant survival of the matched conventional and robotic UKR groups was 97.8% (95% CI 97.2% to 98.3%) and 96.4% (92.4% to 98.3%), respectively, with no between group difference in revision risk (hazard ratio 1.03, 95% CI 0.75 to 1.42; P=0.86). Cause specific revision risk and intraoperative complication risks did not differ between groups for either TKR or UKR. Conclusions: No statistically significant differences were detected in revision risk, cause specific risk, or intraoperative complication risk between the robotic and conventional groups. This was consistent for both TKR and UKR. The possibility of unmeasured and residual confounding cannot be ruled out, however, owing to the observational design of the study. These results highlight the importance of careful evaluation of robotic technology in publicly funded healthcare systems, given the substantially higher capital and procedural costs involved.

Aims: The aim of this study was to quantify the contribution of patient-, surgeon-, and institution-level factors to variation in dual-mobility component use in primary total hip arthroplasty (DM-THA) in England. We sought to determine whether current patterns reflect appropriate patient selection or unwarranted practice variation. We hypothesized that surgeon- and institution-level factors account for a greater proportion of variation in DM-THA use than patient-level characteristics. Methods: We analyzed 238,455 primary THAs recorded in the National Joint Registry which were linked with Hospital Episode Statistics for England between 1 January 2018 and 31 December 2022. Using cross-classified multilevel logistic regression modelling, we partitioned variation in DM-THA use attributable to patient, surgeon, and institutional factors. Patient risk factors were included as fixed effects, with random effects specified for surgeons and hospitals to account for clustering. Results: Of the total cohort, 7,032 underwent DM-THA. These patients were older, more frail, and more frequently treated for neck of femur fractures. Surgeon-level factors explained 32.4% and institutional factors 22.9% of variation in DM-THA use. Patient-level factors accounted for approximately 10% of the variance (marginal R² = 0.10), while the full model explained 59.8% of variation (conditional R² = 0.598). Findings were consistent in analyses restricted to high-volume surgeons. Conclusion: Variation in DM-THA use is predominantly driven by surgeon and institutional practice rather than patient characteristics. Prognostic tools which can identify patients at high dislocation risk, to guide inclusion criteria for targeted randomized trials, are needed to establish evidence-based guidance for DM implant selection and reduce unwarranted practice variation. Cite this article: Bone Joint J 2026;108-B(9):1097–1103.

Purpose: This study assessed the association between (1) annual and (2) cumulative surgeon volume and patient‐reported outcome measures (PROMS) after hip arthroscopic surgery for femoroacetabular impingement syndrome (FAIS) using data from the UK Nonarthroplasty hip registry (NAHR). Methods: All hip arthroscopies (HAs) recorded in the UK NAHR between January 2012 and May 2023 were identified. Annual and cumulative consultant HA volume were derived for each case. International Hip Outcome Tool (iHOT)‐12 and EuroQol‐5 Dimensions (EQ‐5D) index scores were captured preoperatively and at 6 and 12 months postoperatively. Multivariable linear regression models with restricted cubic splines were used to define associations between consultant annual and cumulative volume and outcome. The null hypothesis was that neither annual nor cumulative volume would be significantly associated with postoperative iHOT‐12 or EQ‐5D score change. Results: A total of 3531 HAs were identified (median age 36 years [IQR: 28–44], 62.4% female) from 76 contributing consultant surgeons. Median annual consultant HA volume was 51 (IQR: 22–102), and median cumulative volume was 163 (IQR: 57–404). Cumulative consultant volume of 170 HA cases demonstrated the lowest 12‐month postoperative iHOT‐12 score gain (29.6, confidence interval [CI: 27.0–32.1]) compared to a cumulative consultant HA volume of 972 cases (35.2 [95%CI 32.1–38.3]), where the score difference became significant but remained below the minimal clinically important difference (MCID). No statistically significant relationship between consultant volume and EQ‐5D Index were observed. Cases where the consultant had the highest cumulative volumes were younger (35.7 vs. 37.0 years, p = 0.03) and a greater proportion underwent excision of isolated cam impingement lesions (62.2% vs. 49.3% p = 0.001), compared to cases performed by the lowest volume consultants. Conclusion: This study found a small but statistically significant association (below the MCID) between higher cumulative case volume and improved early functional outcome following HA. Highest‐volume surgeons consistently treated younger patients with less‐severe chondral damage. Level of Evidence: Level III.

Background: Hip fractures (HF) are a major cause of morbidity and mortality and account for substantial health and social care costs. We report the incidence and characteristics of older patients (≥60 years) presenting with hip fractures (HFs) in England, Wales and Northern Ireland. Methods: We conducted a national cohort study of HF events recorded in the National Hip Fracture Database (NHFD) between May 2011 and September 2023. HF incidence was derived using sex and age-banded mid-year population estimates from the Office for National Statistics to calculate at-risk population incidence for the whole cohort and by sex and age strata. Results: 735 436 HF events were included (71% female, mean age 82.9 years). Annual NHFD HF numbers increased across the study period from 55 832 to 63 341 (13.5%). However, overall incidence declined from 422.8 to 412.8 per 100 000 population. Divergence was observed between sexes, with falling incidence in females but rising in males. HF patients are presenting with greater systemic disease burden, with the proportion graded American Society of Anaesthesiologists ≥3 increasing in both females (66.4% to 80.4%) and males (73.4% to 84.3%). Conclusion: Despite modest declines in overall incidence in women, absolute HF numbers continue to rise. The divergent sex-specific trends and substantial increase in patient comorbidity represent important shifts in HF epidemiology with implications for service capacity, risk stratification and secondary prevention strategies. This mandates urgent realignment of care resources to minimise the impact on patient mortality and economic health burden.

Aims: Obesity and diabetes are risk factors for developing osteoarthritis and associated with worse outcomes following total hip (THA) and total knee arthroplasty (TKA). Glucagon-like peptide-1 receptor agonists (GLP-1as) improve glycaemic control and promote weight loss. The aim of this study was to assess the impact of GLP-1a treatment on 90-day surgical complications following THA and TKA. Methods: A systematic review was undertaken according to PRISMA guidelines. Studies were included if they reported at least one outcome of interest following primary THA or TKA. In all the studies which were included, a GLP-1a was prescribed for the management of type 2 diabetes mellitus or for weight loss in obese patients. The primary outcome was the 90-day surgical complication rate in the combined THA and TKA cohort. The 90-day surgical complications were defined as: periprosthetic joint infection (PJI), wound dehiscence, periprosthetic fracture (PPF), haematoma, nerve injury, or surgical site infection (SSI). Secondary outcomes included 90-day medical complications and readmission rates, all-cause revision, healthcare costs, and length of stay. Results: A total of 78 studies were reviewed, and ten matched cohort studies with a total of 96,356 patients (30,350 THAs and 66,606 TKAs) were included. The mean age of the patients was 61.9 years; 60.1% female (n = 57,939). Two studies had a serious risk of overall bias and eight had a moderate risk. The confidence of evidence according to the GRADE assessment was very low for all but one outcome measure (healthcare cost: moderate). The rate of 90-day surgical complications was lower in the GLP-1a cohort (pooled risk ratio (RR) 0.73). The use of a GLP-1a lowered the 90-day medical complications and readmission rates (RR 0.78 and RR 0.79 (I ² = 61.6%), respectively). At two years, the revision rate ranged from 1.7% to 3.3% for the GLP-1a cohort and 1.7% to 4.5% for the controls. Sub-group analysis showed a stronger association in the THA group, in which the use of a GLP-1a was associated with significantly lower rates of 90-day surgical complications (RR 0.63), 90-day medical complications (RR 0.55), and 90-day readmissions (RR 0.82). In contrast, the use of a GLP-1a in the TKA group was only sigificantly associated with a lower 90-day readmission rate (RR 0.77). Conclusion: GLP-1a treatment before THA and TKA potentially reduces the 90-day surgical and medical complications and 90-day readmission rates in high-risk diabetic and obese patients. The use of a GLP-1a may also be associated with substantial cost savings. Cite this article: Bone Joint J 2026;108-B(8):986–994.

BACKGROUND Knee replacement surgery is performed for people with end-stage knee osteoarthritis to reduce pain and improve activity participation. However, patients have limited evidence on how quickly they can expect to return to normal physical activity after surgery, and how much better their activity will get. Smartphones and consumer-grade wearables capture step count data on individuals' devices. Linked with detailed surgical information, these data could provide a unique opportunity to address this question. However, linking consumer data with national health data, which are currently held separately, is uncharted territory and comes with operational challenges. OBJECTIVE This study has two objectives. The first is to describe physical activity patterns after knee replacement surgery in people with knee osteoarthritis, and the second is to understand how researchers can successfully conduct retrospective cohort studies that link consumer activity data collected through smartphones and consumer wearables with national health datasets through this worked example. METHODS We aim to recruit up to 1,000 people aged 18 or older who had knee osteoarthritis and a knee replacement in England, Wales, Northern Ireland, Isle of Man, or Guernsey between 1st January 2017 to 31st December 2023. Participants are eligible if they used Fitbit, Apple (iPhone or Apple Watch), or Oura Ring devices to track physical activity in the six months before and one year after surgery. Study participants will have to complete three tasks: answer questions about their surgery, provide identifiers for linkage, and provide access to physical activity information from their devices. Physical activity data will be linked to an extract of data from their surgical health record in the National Joint Registry. The primary outcome for the first objective will be post-operative relative daily step count, calculated by dividing post-operative daily step count by a pre-operative baseline level. We will assess the extent to which step count increases post-surgery using linear mixed-effects models and explore trajectory groupings using latent class mixed models. For the second objective, we will summarise sources of referral to the study and their corresponding conversion rate, cumulative recruitment numbers, website engagement, representativeness, and completeness of physical activity data. RESULTS The study started recruitment in November 2025, with 132 participants onboarded as of 24 June 2026. We will complete recruitment on 31 July 2026 and aim to publish results by December 2026. CONCLUSIONS Understanding how physical activity changes after knee replacement surgery could help patients and clinicians make more informed decisions based on anticipated future physical activity levels and recovery patterns. Through this worked example, we will generate learning about how to conduct retrospective cohort studies linking consumer health measures to national healthcare datasets, particularly regarding recruitment and study onboarding. CLINICALTRIAL ISRCTN41610298

Objectives To evaluate the early implementation of tirzepatide for the management of obesity in primary care in England, including variations in Integrated Care Board (ICB) service delivery models, access criteria, and initial prescribing activity following national policy introduction (June 2025). This paper reports the initial phase of a wider mixed methods national evaluation of tirzepatide delivery models for initial priority cohorts defined in NHS England interim commissioning guidance. Design National study combining 1) mapping of ICB implementation plans using a questionnaire, publicly available data and documents, and stakeholder discussions, and 2) analysis of English Prescribing Dataset (EPD) to investigate tirzepatide prescribing trends using interrupted time-series methods to compare monthly prescriptions before and after June 2025 (implementation start) against a synthetic control series based on semaglutide prescribing. Setting ICBs across England. Participants 23/42 ICBs (55%) responded, of which 13 also took part in stakeholder meetings. For 17 ICBs (40%), we additionally extracted information from publicly available documents. Main outcome measures Descriptions of service delivery models, access criteria, and ICB characteristics; stakeholder‑reported implementation challenges; and prescribing trends over time. Results We were able to categorise models of care for 40/42 ICBs. A general practice‑led delivery model was most commonly planned (18/40; 45%), followed by community/local‑based delivery (8; 20%), custom models (8; 20%), and specialist weight management service (SWMS) community outreach (6; 15%). Four ICBs (10%) planned to use more than one model of care. Most ICBs reported following national priority cohort eligibility criteria 31/39 (79%), while eight (21%) applied additional prioritisation criteria due to funding constraints. Stakeholder discussions highlighted variation in the interpretation of model definitions, variation in operationalisation, and challenges related to affordability, insufficient workforce capacity, and tight timelines, with some ICBs not yet prescribing at the time of stakeholder meetings (November 2025). Analyses of EPD showed a steady increase in tirzepatide prescribing over time, with no evidence that the June 2025 policy implementation date produced any additional increase beyond background diabetes‑related trends. Conclusions Early implementation of tirzepatide prescribing in primary care has been characterised by heterogeneous service models, local adaptations of access criteria, and slow implementation of obesity‑related prescribing. Findings highlight the challenges of implementing new pharmacological treatments including patient management within routine care and the need for clear guidance, realistic timelines, and system capacity. Ongoing evaluation is needed to understand how service models evolve and implications for access, equity, and outcomes.

Aims: Within one year after hip fracture in older adults living in England, this study aimed to estimate the inpatient hospital costs associated with subsequent complications, including further fractures, reoperations, and death. Methods: The study linked national datasets to identify all patients aged ≥ 60 years who underwent an index hip fracture operation in all hospitals in England between 1 April 2016 and 31 March 2019, with one-year follow-up. Patients were categorized into eight mutually exclusive groups based on combinations of further fracture (after 30 days from index hip fracture, at any skeletal site, excluding facial, digit, and skull fractures), reoperation, and death in the year after hip fracture. Inpatient costs were valued using 2019/20 NHS reference costs. Marginal cost differences over patients with no complications were estimated using generalized linear models. Results: Among 164,691 patients (70.8% female (n = 116,534); mean age 82.7 years (SD 8.7)), across 159 hospitals, the mean one-year inpatient cost was £14,581 (SD 8,583). Further fracture (£24,306 (SD 11,635); n = 11,226) and reoperation (£23,923 (SD 13,537); n = 7,522) were each associated with approximately twice the cost of patients without complications (£12,859 (SD 7,097); n = 107,638). Additional costs, adjusted for prefracture costs, fracture type, and patient case-mix variables, were highest for patients with further fracture and reoperation who died within one year (+£16,971 (95% CI 15,914 to 18,028); n = 541), closely followed by patients with both further fracture and reoperation who survived (+£13,562 (95% CI 13,065 to 14,059); n = 2,348), compared with adjusted cost of patients who survived with no complications (mean £13,159 (95% CI 13,118 to 13,201)). Nationally, annual inpatient expenditure for hip fracture care was estimated to £647 million in 2018/19. Of this, £265 million was attributable to patients with a complication (35% of all patients). Conclusion: Nearly half of total national inpatient expenditure on hip fracture care is concentrated among patients who experienced complications or died. Further fracture and reoperation markedly increase inpatient expenditure, highlighting the importance of preventing complications and refractures. Cite this article: Bone Joint J 2026;108-B(7):952–960.

Objectives: To characterise the use of intra-articular corticosteroid injections (IACIs) in a large patient cohort with OA in UK primary care from 2005 to 2020. We sought to evaluate IACI practices, identify patterns of administration across different joints, assess regional variation and examine potential demographic factors influencing IACI utilisation for managing OA. Methods: This retrospective cohort study used data from the UK Clinical Practice Research Datalink GOLD, a database of primary care health records. Data from other healthcare settings were not included. Patients ≥20 years of age with a first diagnosis of OA between 2005 and 2019 were included. OA and IACIs were identified through Read and Gemscript codes, respectively. Results: Of 220 125 patients with OA, 11% received at least one IACI. The anatomical location was only specified in 56% of IACI records. The highest proportion of injections occurred in patients with multiple joints diagnosed with OA (19.0%), followed by the shoulder (14.9%) and knee (12.5%). The median time from OA diagnosis to first injection was 0.9 years and was shortest for the shoulder (0.3 years) and longest for the elbow (2.1 years). We observed substantial regional variation, with incidence rates ranging from 1.2 injections per 100 patient-years in the South East Coast to 3.8 in Yorkshire and the Humber. Conclusion: This study observed notable variation in the median time from OA diagnosis to injection, across different joints, with substantial heterogeneity in the incidence of IACIs across UK regions.