Performance Drift in Machine Learning Models for Cardiac Surgery Risk Prediction: Retrospective Analysis.

Background: The Society of Thoracic Surgeons and European System for Cardiac Operative Risk Evaluation (EuroSCORE) II risk scores are the most commonly used risk prediction models for in-hospital mortality after adult cardiac surgery. However, they are prone to miscalibration over time and poor generalization across data sets; thus, their use remains controversial. Despite increased interest, a gap in understanding the effect of data set drift on the performance of machine learning (ML) over time remains a barrier to its wider use in clinical practice. Data set drift occurs when an ML system underperforms because of a mismatch between the data it was developed from and the data on which it is deployed. Objective: In this study, we analyzed the extent of performance drift using models built on a large UK cardiac surgery database. The objectives were to (1) rank and assess the extent of performance drift in cardiac surgery risk ML models over time and (2) investigate any potential influence of data set drift and variable importance drift on performance drift. Methods: We conducted a retrospective analysis of prospectively, routinely gathered data on adult patients undergoing cardiac surgery in the United Kingdom between 2012 and 2019. We temporally split the data 70:30 into a training and validation set and a holdout set. Five novel ML mortality prediction models were developed and assessed, along with EuroSCORE II, for relationships between and within variable importance drift, performance drift, and actual data set drift. Performance was assessed using a consensus metric. Results: A total of 227,087 adults underwent cardiac surgery during the study period, with a mortality rate of 2.76% (n=6258). There was strong evidence of a decrease in overall performance across all models (P<.0001). Extreme gradient boosting (clinical effectiveness metric [CEM] 0.728, 95% CI 0.728-0.729) and random forest (CEM 0.727, 95% CI 0.727-0.728) were the overall best-performing models, both temporally and nontemporally. EuroSCORE II performed the worst across all comparisons. Sharp changes in variable importance and data set drift from October to December 2017, from June to July 2018, and from December 2018 to February 2019 mirrored the effects of performance decrease across models. Conclusions: All models show a decrease in at least 3 of the 5 individual metrics. CEM and variable importance drift detection demonstrate the limitation of logistic regression methods used for cardiac surgery risk prediction and the effects of data set drift. Future work will be required to determine the interplay between ML models and whether ensemble models could improve on their respective performance advantages.

Access to publicly funded weight management services in England using routine data from primary and secondary care (2007-2020): An observational cohort study.

BACKGROUND: Adults living with overweight/obesity are eligible for publicly funded weight management (WM) programmes according to national guidance. People with the most severe and complex obesity are eligible for bariatric surgery. Primary care plays a key role in identifying overweight/obesity and referring to WM interventions. This study aimed to (1) describe the primary care population in England who (a) are referred for WM interventions and (b) undergo bariatric surgery and (2) determine the patient and GP practice characteristics associated with both. METHODS AND FINDINGS: An observational cohort study was undertaken using routinely collected primary care data in England from the Clinical Practice Research Datalink linked with Hospital Episode Statistics. During the study period (January 2007 to June 2020), 1,811,587 adults met the inclusion criteria of a recording of overweight/obesity in primary care, of which 54.62% were female and 20.10% aged 45 to 54. Only 56,783 (3.13%) were referred to WM, and 3,701 (1.09% of those with severe and complex obesity) underwent bariatric surgery. Multivariable Poisson regression examined the associations of demographic, clinical, and regional characteristics on the likelihood of WM referral and bariatric surgery. Higher body mass index (BMI) and practice region had the strongest associations with both outcomes. People with BMI ≥40 kg/m2 were more than 6 times as likely to be referred for WM (10.05% of individuals) than BMI 25.0 to 29.9 kg/m2 (1.34%) (rate ratio (RR) 6.19, 95% confidence interval (CI) [5.99,6.40], p < 0.001). They were more than 5 times as likely to undergo bariatric surgery (3.98%) than BMI 35.0 to 40.0 kg/m2 with a comorbidity (0.53%) (RR 5.52, 95% CI [5.07,6.02], p < 0.001). Patients from practices in the West Midlands were the most likely to have a WM referral (5.40%) (RR 2.17, 95% CI [2.10,2.24], p < 0.001, compared with the North West, 2.89%), and practices from the East of England least likely (1.04%) (RR 0.43, 95% CI [0.41,0.46], p < 0.001, compared with North West). Patients from practices in London were the most likely to undergo bariatric surgery (2.15%), and practices in the North West the least likely (0.68%) (RR 3.29, 95% CI [2.88,3.76], p < 0.001, London compared with North West). Longer duration since diagnosis with severe and complex obesity (e.g., 1.67% of individuals diagnosed in 2007 versus 0.34% in 2015, RR 0.20, 95% CI [0.12,0.32], p < 0.001), and increasing comorbidities (e.g., 2.26% of individuals with 6+ comorbidities versus 1.39% with none (RR 8.79, 95% CI [7.16,10.79], p < 0.001) were also strongly associated with bariatric surgery. The main limitation is the reliance on overweight/obesity being recorded within primary care records to identify the study population. CONCLUSIONS: Between 2007 and 2020, a very small percentage of the primary care population eligible for WM referral or bariatric surgery according to national guidance received either. Higher BMI and GP practice region had the strongest associations with both. Regional inequalities may reflect differences in commissioning and provision of WM services across the country. Multi-stakeholder qualitative research is ongoing to understand the barriers to accessing WM services and potential solutions. Together with population-wide prevention strategies, improved access to WM interventions is needed to reduce obesity levels.

Cardiac surgery risk prediction using ensemble machine learning to incorporate legacy risk scores: A benchmarking study.

Objective: The introduction of new clinical risk scores (e.g. European System for Cardiac Operative Risk Evaluation (EuroSCORE) II) superseding original scores (e.g. EuroSCORE I) with different variable sets typically result in disparate datasets due to high levels of missingness for new score variables prior to time of adoption. Little is known about the use of ensemble learning to incorporate disparate data from legacy scores. We tested the hypothesised that Homogenenous and Heterogeneous Machine Learning (ML) ensembles will have better performance than ensembles of Dynamic Model Averaging (DMA) for combining knowledge from EuroSCORE I legacy data with EuroSCORE II data to predict cardiac surgery risk. Methods: Using the National Adult Cardiac Surgery Audit dataset, we trained 12 different base learner models, based on two different variable sets from either EuroSCORE I (LogES) or EuroScore II (ES II), partitioned by the time of score adoption (1996-2016 or 2012-2016) and evaluated on holdout set (2017-2019). These base learner models were ensembled using nine different combinations of six ML algorithms to produce homogeneous or heterogeneous ensembles. Performance was assessed using a consensus metric. Results: Xgboost homogenous ensemble (HE) was the highest performing model (clinical effectiveness metric (CEM) 0.725) with area under the curve (AUC) (0.8327; 95% confidence interval (CI) 0.8323-0.8329) followed by Random Forest HE (CEM 0.723; AUC 0.8325; 95%CI 0.8320-0.8326). Across different heterogenous ensembles, significantly better performance was obtained by combining siloed datasets across time (CEM 0.720) than building ensembles of either 1996-2011 (t-test adjusted, p = 1.67×10-6) or 2012-2019 (t-test adjusted, p = 1.35×10-193) datasets alone. Conclusions: Both homogenous and heterogenous ML ensembles performed significantly better than DMA ensemble of Bayesian Update models. Time-dependent ensemble combination of variables, having differing qualities according to time of score adoption, enabled previously siloed data to be combined, leading to increased power, clinical interpretability of variables and usage of data.

The Sensitivity and Specificity of Parental Report of Concern for Identifying Language Disorder in Children With Craniosynostosis.

ABSTRACT: Many factors that may co-occur with craniosynostosis, such as oral structural anomalies, hearing impairment, visual impairment, cognitive difficulties and psychosocial factors, may predispose this population to communication difficulties. At the Oxford Craniofacial Unit, children's speech, language and communication are regularly monitored in accordance with a systematic developmental screening protocol developed by the Speech and Language Therapists in the 4 United Kingdom (UK) Highly Specialized Craniofacial Centers. In addition to routine assessments, when parents attend routine multidisciplinary clinic appointments, they are asked about their child's communication development, and whether they have any concerns.A retrospective review was undertaken of parental concerns about hearing, speech development, behavior, physical development, concentration, school and friendships as indicated by parents on the Oxford Craniofacial Unit Pre-Clinic Questionnaire. The areas of concern were then correlated with the results of a standardized, guided parent questionnaire about children's language development, (Children's Communication Checklist - 2 (CCC-2)), to determine whether parental concern alone is a reliable way of identifying whether patients require further assessment for Language Disorder associated with Craniosynostosis.Participants were parents of 89 monolingual English-speaking children with craniosynostosis (62 male; 27 female), age range four to 13 years (mean age = 8 years 7 months), receiving active care at the Oxford Craniofacial Unit (June 2017-July 2018). Results of the pre-clinic questionnaire indicated that 6% of parents had concerns about their child's communication development. Results of the CCC-2 indicated that 29/89 (32.6%) of children required further assessment for Language Disorder associated with Craniosynostosis. When language difficulties were identified on the CCC-2, only 14% (n = 4/29) parents indicated concern on the pre clinic questionnaire. Results indicated that parental concern about behavior was the most important factor in identifying language disorder (P = 0.023).Results reinforce that the pre-clinic questionnaire is useful for identifying areas of parental concern. Results also indicate that parental concern alone is not sufficient to identify language disorder, and that further, detailed assessment is warranted. The results are consistent with previously reported links between behavior and language in the general population.

Treatment effects of reverse total shoulder arthroplasty - a simple method to measure outcomes at 6, 12, 24 and 60 months for each patient.

BACKGROUND: Although shoulder arthroplasty is less common than knee or hip arthroplasty, the number of procedures being performed is increasing rapidly. The treatment effect is a simple method to measure outcome of joint replacement. The method was applied to measure results of total hip/knee arthroplasty but not yet for shoulder arthroplasty. METHODS: Included were patients with unilateral cuff arthropathy (Hamada grades > = 2) treated with reversed total shoulder arthroplasty (RSA) in this prospective multicenter study. The patients were assessed with the ASES questionnaire. The treatment effects (TE) was calculated for each patient. TE = score reduction/baseline score. A positive TE means amelioration, TE = 0 unchanged, and a negative TE means worse. The primary aim was to calculate the TE's for RSA at 6, 12, 24, and 60 months postoperatively. The secondary aim was to analyze the influence of confounders (preoperative Hamada grade, age, gender, dominance, side of the affected shoulder, general co-morbidities measured using ASA grade). RESULTS: Two hundred three patients were included for this analysis of whom 183 patients had a complete 2 year follow up. Two years postoperatively the mean ASES score augmented significant from 20.5 to 78.7 (p < 0.001). The 2 year TE's ranged from 1 to 0.09. We had no patient with a negative TE. A higher Hamada grade was associated with better TE's (Hamada grade 4+ vs. 2, p-value 0.042). For age and dominant side there were weak associations where those aged 80+ and dominant side had better TE's. The patients with higher ASA grade had lower TE's (ASA grade 4+ vs. 1, p-value 0.013). The mean TE's were 0.77 at 6-months, 0.81 at 1 year, 0.76 at 2 years and 0.73 at 5 years. CONCLUSIONS: The outcome for reverse shoulder arthroplasty can be measured with the treatment effect method; the 2 years TE's vary from 1 to 0.09. The mean treatment effects change little in the first five postoperative years (from 0.73 to 0.81). The confounders for better TE's were: higher severity of cuff arthropathy (Hamada grade 3, 4 and 5), less co-morbidities (ASA Grade 1), higher age (80+) and dominant side. Gender did not influence the 2-year TE's. TRIAL REGISTRATION: Comité intercantonal d'éthique (Jura, Fribourg, Neuchâtel), number 01/2008, 24.09.2008.

Geographical Variation in Outcomes of Primary Hip and Knee Replacement.

Importance: Little is known about variation in outcomes of surgery or about the factors associated with such variation. Objectives: To evaluate variation in patient outcomes and costs for primary hip and knee replacement across health areas in England and to identify whether patient, surgical, or hospital factors are associated with such variation. Design, Setting, and Participants: This cohort study used data from the National Joint Registry, linked to English Hospital Episode Statistics and Patient Reported Outcome Measures data sets, for 383 382 adult patients who underwent primary total hip replacement (THR) or primary total and unicompartmental knee replacement (TKR) surgical procedures from January 2014 to December 2016. Geographical Information Systems were used to display maps describing adjusted estimates of variation in outcomes across health areas. Data analysis took place from January 2018 to August 2019. Exposures: Patient characteristics (eg, age, sex, body mass index [BMI], and socioeconomic deprivation), surgical factors (eg, surgeon volume and grade), and hospital organizational factors (eg, number of operating theaters, number of specialist consultants, and hospital volume). Main Outcomes and Measures: Length of stay (LOS), bed-day costs, change in Oxford hip or knee scores 6 months after surgery, and complications 6 months after surgery. Results: A total of 173 107 patients (mean [SD] age, 69.3 [10.7] years; mean [SD] BMI, 28.9 [5.2]) underwent primary THR and 210 275 patients (mean [SD] age 69.7 [9.4] years; mean [SD] BMI, 31.1 [5.5]) underwent primary TKR, nested in 207 health areas. A number of factors were associated with longer LOS, higher bed-day costs, smaller changes in Oxford hip or knee scores, and a higher percentage of complications, including a workforce with a higher number of less experienced physicians (eg, LOS for less experienced surgeons, THR: regression coefficient, 0.02; 95% CI, 0.01 to 0.03; P < .001; TKR: regression coefficient, 0.01; 95% CI, 0.01 to 0.02; P < .001), public hospitals (eg, bed-day costs for private hospitals, THR: regression coefficient, -0.15; 95% CI, -0.15 to -0.14; P < .001; TKR: regression coefficient, -0.19; 95% CI, -0.19 to -0.19; P < .001), low volume of surgical procedures per surgeon (eg, change in Oxford hip or knee scores for lead surgeon with ≤10 vs >150 surgical procedures per year, THR: regression coefficient, -1.03; 95% CI, -1.47 to -0.58; P < .001; TKR: regression coefficient, -0.54; 95% CI, -1.01 to -0.06), and low volume of surgical procedures per hospital (eg, percentage of complications for hospitals with ≤200 vs ≥500 surgical procedures per year, THR: regression coefficient, 0.12; 95% CI, 0.04 to 0.21; P < .001; TKR: regression coefficient, 0.09; 95% CI, 0.01 to 0.18; P = .03). Although these factors did not attenuate the magnitude of variation across health areas, they had ecological correlations with the observed geographical variations in outcomes of surgery by health area. For example, the percentage of public and private hospitals was ecologically correlated at the health area level with longer and shorter stays, respectively (public hospital, THR: ρ, 0.41; public hospital, TKR: ρ, 0.44; private hospital, THR: ρ, -0.37; private hospital, THR: ρ, -0.38). Across health areas, estimated mean length of stay ranged from 3 to 7 days, and associated bed-day costs ranged from £4727 ($5827) to £8800 ($10 848) for both total hip and knee replacement. The absolute estimated mean change in Oxford hip score varied from 18.7 to 24.6 points and, for Oxford knee score, from 13.1 to 18.8. Estimated 6-month complications ranged from 2.9% to 5.8% for both THR and TKR. Conclusions and Relevance: In this study, models indicated that higher surgical volume by surgeon and by hospital as well as private hospitals were associated with better patient outcomes, which could be explained by the changing case mix of public hospitals treating an increasing number of more complex patients. A higher proportion of less experienced physicians was associated with poorer outcomes. This variation was observed geographically.

Protective effect of antirheumatic drugs on dementia in rheumatoid arthritis patients.

INTRODUCTION: Rheumatoid arthritis is a systemic inflammatory disease, and classical disease-modifying antirheumatic drugs (cDMARDs) have proven efficacy. It is unknown what impact cDMARDs might have on dementia as an outcome. METHODS: Incident diagnoses of rheumatoid arthritis in persons over 18 years from 1995 to 2011 were identified from the UK Clinical Practice Research Datalink. There were 3876 cDMARD users and were propensity score matched to 1938 nonusers, on a wide range of confounders. Impact on dementia was assessed using survival models. RESULTS: cDMARD users were at reduced risk of dementia (hazard ratio: 0.60; 95% confidence interval: 0.42-0.85). The effect was strongest in methotrexate users (hazard ratio: 0.52; 95% confidence interval; 0.34-0.82). DISCUSSION: The strong effect of cDMARD use on halving of dementia risk requires replication in a trial and may provide an important therapeutic pharmacological treatment.

A population-based survival analysis describing the association of body mass index on time to revision for total hip and knee replacements: results from the UK general practice research database.

OBJECTIVES: Against a backdrop of rising levels of obesity, we describe and estimate associations of body mass index (BMI), age and gender with time to revision for participants undergoing primary total hip (THR) or knee (TKR) replacement in the UK. DESIGN: Population-based cohort study. SETTING: Routinely collected primary care data from a representative sample of general practices, including linked data on all secondary care events. PARTICIPANTS: Population-based cohort study of 63 162 patients with THR and 54 276 with TKR in the UK General Practice Research Database between 1988 and 2011. PRIMARY AND SECONDARY OUTCOMES: Risk of THR and TKR revision associated with BMI, age and gender, after adjusting for the competing risk of death. RESULTS: The 5-year cumulative incidence rate for THR was 2.2% for men and 1.8% for women (TKR 2.3% for men, 1.6% for women). The adjusted overall subhazard ratio (SHR) for patients with THR undergoing subsequent hip revision surgery, with a competing risk of death, were estimated at 1.020 (95% CI 1.009 to 1.032) per additional unit (kg/m(2)) of BMI, 1.23 (95% CI 1.10 to 1.38) for men compared with women and 0.970 (95% CI 0.967 to 0.973) per additional year of age. For patients with TKR, the equivalent estimates were 1.015 (95% CI 1.002 to 1.028) for BMI; 1.51 (95% CI 1.32 to 1.73) for gender and 0.957 (95% CI 0.951 to 0.962) for age. Morbidly obese patients with THR had a 65.5% increase (95% CI 15.4% to 137.3%, p=0.006) in the subhazard of revision versus the normal BMI group (18.5-25). The effect for TKR was smaller (a 43.9% increase) and weaker (95% CI 2.6% to 103.9%, p=0.040). CONCLUSIONS: BMI is estimated to have a small but statistically significant association with the risk of hip and knee revision, but absolute numbers are small. Further studies are needed in order to distinguish between effects for specific revision surgery indications.

The association of patient characteristics and surgical variables on symptoms of pain and function over 5 years following primary hip-replacement surgery: a prospective cohort study.

OBJECTIVES: To identify patient characteristics and surgical factors associated with patient-reported outcomes over 5 years following primary total hip replacement (THR). DESIGN: Prospective cohort study. SETTING: Seven hospitals across England and Scotland. PARTICIPANTS: 1431 primary hip replacements for osteoarthritis. MAIN OUTCOME MEASURES: The Oxford Hip Score (OHS) was collected preoperatively and each year up to 5 years postoperatively. Repeated measures such as linear regression modelling are used to identify patient and surgical predictors of outcome and describe trends over time. RESULTS: The majority of patients demonstrated substantial improvement in pain/function in the first year after surgery-between 1 and 5 years follow-up, there was neither further improvement nor decline. The strongest determinant of attained postoperative OHS was the preoperative OHS-those with worse preoperative pain/function had worse postoperative pain/function. Other predictors with small but significant effects included: femoral component offset-women with an offset of 44 or more had better outcomes; age-compared to those aged 50-60, younger (age <50) and older patients (age >60) had worse outcome, increasing body mass index (BMI), more coexisting diseases and worse Short Form 36 mental health (MH) was related to worse postoperative pain/function. Assessment of change in OHS between preoperative  and postoperative assessments revealed that patients achieved substantial and clinically relevant symptomatic improvement (change), regardless of variation in these patient and surgical factors. CONCLUSIONS: Patients received substantial benefit from surgery, regardless of their preoperative assessments and surgical characteristics (baseline pain/function, age, BMI, comorbidities, MH and femoral component offset). Further research is needed to identify other factors that can improve our ability to identify patients at risk of poor outcomes from THR surgery.

Predictors of outcomes of total knee replacement surgery.

OBJECTIVE: To identify pre-operative predictors of patient-reported outcomes of primary total knee replacement (TKR) surgery. METHODS: The Elective Orthopaedic Centre database is a large prospective cohort of 1991 patients receiving primary TKR in south-west London from 2005 to 2008. The primary outcome is the 6-month post-operative Oxford Knee Score (OKS). To classify whether patients had a clinically important outcome, we calculated a patient acceptable symptom state (PASS) for the 6-month OKS related to satisfaction with surgery. Potential predictor variables were pre-operative OKS, age, sex, BMI, deprivation, surgical side, diagnosis, operation type, American Society of Anesthesiologists grade and EQ5D anxiety/depression. Regression modelling was used to identify predictors of outcome. RESULTS: The strongest determinants of outcome include pre-operative pain/function-those with less severe pre-operative disease obtain the best outcomes; diagnosis in relation to pain outcome-patients with RA did better than those with OA; deprivation-those living in poorer areas had worse outcomes; and anxiety/depression-worse pre-operative anxiety/depression led to worse pain. Differences were observed between predictors of pain and functional outcomes. Diagnosis of RA and anxiety/depression were associated with pain, whereas age and gender were specifically associated with function. BMI was not a clinically important predictor of outcome. CONCLUSION: This study identified clinically important predictors of attained pain/function post-TKR. Predictors of pain were not necessarily the same as functional outcomes, which may be important in the context of a patient's expectations of surgery. Other predictive factors need to be identified to improve our ability to recognize patients at risk of poor TKR outcomes.

Clinical tool to identify patients who are most likely to achieve long-term improvement in physical function after total hip arthroplasty.

OBJECTIVE: To develop a clinical risk prediction tool to identify patients most likely to experience long-term clinically meaningful functional improvement following total hip arthroplasty (THA). METHODS: We studied 282 patients from 2 health districts in England (Portsmouth and North Staffordshire) who were ≥45 years of age and undergoing THA for primary osteoarthritis. Baseline data on age, sex, comorbidity, body mass index (BMI), functional status (Short Form 36 [SF-36]), and preoperative radiographic severity were collected by interview and examination. The outcome was a clinically significant (30-point) improvement in SF-36 physical function score assessed ~8 years after THA. Logistic regression modeling was used to identify predictors of functional improvement. RESULTS: Improvement in physical function was less likely in patients with better preoperative functioning (odds ratio [OR] 0.73 [95% confidence interval (95% CI) 0.60, 0.89]), older people (OR 0.94 [95% CI 0.90, 0.98]), women (OR 0.37 [95% CI 0.19, 0.72]), those with a previous hip injury (OR 0.14 [95% CI 0.03, 0.74]), and those with a greater number of painful joint sites (OR 0.61 [95% CI 0.46, 0.80]). Patients with worse radiographic grades were most likely to improve (OR 2.15 [95% CI 1.17, 3.93]). We found no influence of BMI or patient comorbidity on functional outcome. Predictors of good outcomes were the same as those of bad outcomes, acting in the opposite direction. A clinical risk prediction tool was developed to identify patients who are most likely to receive functional improvement following THA. CONCLUSION: This prediction tool has the potential to inform health care professionals and patients about functional improvement following THA (as distinct from driving rationing or commissioning decisions regarding who should have surgery); it requires introduction into clinical practice under research conditions to investigate its impact on decisions made by patients and clinicians.

Inequalities in rates of renal replacement therapy in England: does it matter who you are or where you live?

UNLABELLED: Introduction. This study explores the geographical variation in renal replacement therapy (RRT) incidence and prevalence after adjusting for general population socio-demographics, renal unit treatment patterns and travel times. METHODS: The UK Renal Registry provided data on all patients in England commencing RRT in 2007 and receiving RRT on 31 December 2007. Multilevel Poisson regression models were constructed separately for incidence and prevalence. Geographical Information Systems software enabled estimation of road travel times and renal unit catchment areas. Small area estimates of RRT prevalence were produced for all 354 local authority districts. RESULTS: Adjusted RRT incidence rates were 1.4 (95% confidence interval 1.2-1.6) times higher in the most deprived areas and 1.7 (1.5-2.0) and 1.5 (1.3-1.7) times higher in areas with most Black and South Asian inhabitants (10+%), respectively. The proportion of a centre's patients on haemodialysis or transplanted were positively associated with RRT incidence (not prevalence); numbers of satellite units were negatively associated with RRT incidence (not prevalence). While only 3% of patients lived >30 min from a dialysis unit, there was an effect of travel time on RRT rates; individuals living 45+ min from a dialysis unit were 20% less likely to commence or receive RRT than those living within 15 min (Ptrend=0.36 and Ptrend<0.001, respectively). A 4-fold variation in adjusted local authority district RRT prevalence rates could not be explained. CONCLUSION: Expansion of renal unit facilities in England has reduced travel times in most areas though the possibility of inequitable geographic access to RRT persists.

Patient preoperative expectations of total hip replacement in European orthopedic centers.

OBJECTIVE: Patient expectations have been identified as a factor that may account for individual differences in recovery after total hip replacement (THR) surgery. However, patient expectations have not been studied within a valid theoretical framework. This study employed the World Health Organization's model of health, the International Classification of Functioning, Disability and Health (ICF), to classify the content of preoperative patient expectations of THR. METHODS: A European cohort of 1,108 patients preoperatively reported 2 types of expectations. Patients reported what they anticipated surgery would enable them to do that they needed to be able to do ("need" expectation), and what they would like to be able to do ("desire" expectation) in a year's time. Free-text responses were independently classified by 2 researchers to 1 or more of the ICF constructs of impairment, activity limitation, and participation restriction. RESULTS: Interrater reliability was high (κ = ≥0.87). All patient expectations were classified to the ICF constructs. Less than 5% of patient expectations were identified as impairment, 58% of "need" expectations were identified as activity limitations, and 45% of "desire" expectations were identified as activity limitations and participation restrictions combined. CONCLUSION: The ICF is a suitable theoretical framework to study patient expectations of THR. THR targets impairment; however, few patient expectations were classified to the ICF definition of impairment. The majority of patient expectations were classified as activity limitation or a combination of activity limitation and participation restriction. Therefore, patient expectations of surgery focus on recovering valued activities rather than reversal of bodily impairments.

Equity in access to total joint replacement of the hip and knee in England: cross sectional study.

OBJECTIVE: To explore geographical and sociodemographic factors associated with variation in equity in access to total hip and knee replacement surgery. DESIGN: Combining small area estimates of need and provision to explore equity in access to care. SETTING: English census wards. SUBJECTS: Patients throughout England who needed total hip or knee replacement and numbers who received surgery. MAIN OUTCOME MEASURES: Predicted rates of need (derived from the Somerset and Avon Survey of Health and English Longitudinal Study of Ageing) and provision (derived from the hospital episode statistics database). Equity rate ratios comparing rates of provision relative to need by sociodemographic, hospital, and distance variables. RESULTS: For both operations there was an "n" shaped curve by age. Compared with people aged 50-59, those aged 60-84 got more provision relative to need, while those aged >or=85 received less total hip replacement (adjusted rate ratio 0.68, 95% confidence interval 0.65 to 0.72) and less total knee replacement (0.87, 0.82 to 0.93). Compared with women, men received more provision relative to need for total hip replacement (1.08, 1.05 to 1.10) and total knee replacement (1.31, 1.28 to 1.34). Compared with the least deprived, residents in the most deprived areas got less provision relative to need for total hip replacement (0.31, 0.30 to 0.33) and total knee replacement (0.33, 0.31 to 0.34). For total knee replacement, those in urban areas got higher provision relative to need, but for total hip replacement it was highest in villages/isolated areas. For total knee replacement, patients living in non-white areas received more provision relative to need (1.04, 1.00 to 1.07) than those in predominantly white areas, but for total hip replacement there was no effect. Adjustment for hospital characteristics did not attenuate the effects. CONCLUSIONS: There is evidence of inequity in access to total hip and total knee replacement surgery by age, sex, deprivation, rurality, and ethnicity. Adjustment for hospital and distance did not attenuate these effects. Policy makers should examine factors at the level of patients or primary care to understand the determinants of inequitable provision.

Patient-reported outcomes one year after primary hip replacement in a European Collaborative Cohort.

OBJECTIVE: To identify whether patients have symptomatic improvement 12 months following total hip replacement (THR) surgery. METHODS: The European Collaborative Database of Cost and Practice Patterns of Total Hip Replacement study consists of 1,327 patients receiving primary THR for osteoarthritis (OA) across 20 European orthopedic centers. The primary outcome was the difference in Western Ontario and McMaster Universities OA Index (WOMAC) score between preoperative and 12-month postoperative measurements. To classify whether patients responded to THR at 12 months, we used return to normal, Outcome Measures in Rheumatology Clinical Trials (OMERACT)-OA Research Society International (OARSI) criteria, minimum important difference (MID), and minimum clinically important difference. Exposures were age, sex, obesity, employment, educational attainment, American Society of Anesthesiologists status, and radiographs. RESULTS: On average, there was a large improvement in WOMAC scores 12 months after surgery, but whereas some patients improved, others got worse. The OMERACT-OARSI method classified 85.7% of patients as responders, MID 70.1%, and return to normal 64.1%. In general, each approach classified the same groups of patients as responding to THR. Based on total WOMAC score, patients who were younger, morbidly obese, employed, and better educated were more likely to respond to THR, but the effects were attenuated after adjustment for confounding, with only the effect of education remaining important. CONCLUSION: The overall average response to THR was good, but approximately 14-36% of patients did not improve, or were worse, 12 months postsurgery. Although the OMERACT-OARSI criteria were originally designed for use in clinical drug trials, they performed well in classifying patient response 12 months post-THR. Further research is required to understand the determinants of patient outcomes following THR.

Modeling the need for hip and knee replacement surgery. Part 2. Incorporating census data to provide small-area predictions for need with uncertainty bounds.

OBJECTIVE: To develop methods to produce small-area estimates of need for hip and knee replacement surgery to inform local health service planning. METHODS: Multilevel Poisson regression modeling was used to estimate rates of need for hip/knee replacement by age, sex, deprivation, rurality, and ethnic mix using a nationally representative population-based survey (the English Longitudinal Study of Ageing, n = 11,392 people age > or =50 years). Estimates of need from the regression model were then combined with stratified census population counts to produce small-area predictions of need. Uncertainty in the predictions was obtained by taking a Bayesian simulation-based approach using WinBUGS software. This allows correlations in parameter estimates to be appropriately incorporated in the credible intervals for the small-area predictions. RESULTS: Small-area estimates of need for hip/knee replacement have been produced for wards and districts in England. Rates of need are adjusted for the sociodemographic characteristics of an area and include 95% credible intervals. Need for hip/knee replacement varies geographically, dependant on the sociodemographic characteristics of an area. CONCLUSION: For the first time, small-area estimates of need for hip/knee replacement surgery have been produced together with estimates of uncertainty to inform local health planning. The methodologic approach described here could be reproduced in other countries and for other disease indicators. Further research is required to combine small-area estimates of need with provision to determine whether there is equitable access to care.

Modeling the need for hip and knee replacement surgery. Part 1. A two-stage cross-cohort approach.

OBJECTIVE: To explore inequalities in the need for hip/knee replacement surgery using a 2-stage cross-cohort approach. METHODS: In the first stage, a small-area population-based survey, the Somerset and Avon Survey of Health, was used to provide a high-quality measure of need for hip/knee replacement using the New Zealand (NZ) score. Receiver operating characteristic curve analyses were used to validate a simplified NZ score, excluding information from clinical examination. In the second stage, a nationally representative population-based survey, the English Longitudinal Study of Ageing, was used to explore inequalities in need for hip/knee replacement using the simplified NZ score. Multilevel Poisson regression modeling was used to estimate rates of need for surgery. Exposures considered were age, sex, social class, ethnicity, obesity, Index of Multiple Deprivation 2004 deprivation quintiles, rurality, and ethnic mix of area. RESULTS: Rates of need for hip/knee replacement increase with age and are lower in men than in women (rate ratio [RR] 0.7, 95% confidence interval [95% CI] 0.6-0.9 for hips; RR 0.8, 95% CI 0.7-1.0 for knees). Those of lowest social class have greater need. Need was greatest for people living in more deprived areas. Individual ethnic group did not predict the need for surgery. For hip replacement, there was no rurality effect; for knee replacement, those in town and fringe areas had greater need. Obesity was a strong predictor of need for surgery (RR 2.3, 95% CI 1.9-2.8 for hips; RR 2.4, 95% CI 2.0-2.8 for knees). CONCLUSION: This study provides evidence of greater variations of inequalities in need for hip/knee replacement than previous studies. Further research should explore geographic variation and produce small-area estimates of need to inform local health planning. It is important to complement data on need with willingness to undergo surgery.

Geographical variation in the provision of elective primary hip and knee replacement: the role of socio-demographic, hospital and distance variables.

BACKGROUND: To explore inequalities in the provision of hip/knee replacement surgery and produce small-area estimates of provision to inform local health planning. METHODS: Hospital Episode Statistics were used to explore inequalities in the provision of primary hip/knee operations in English NHS hospitals in 2002. Multilevel Poisson regression modelling was used to estimate rates of surgical provision by socio-demographic, hospital and distance variables. GIS software was used to estimate road travel times and create hospital catchment areas. RESULTS: Rates of joint replacement increased with age before falling in those aged 80+. Women received more operations than men. People living in the most deprived areas obtained fewer hip, but more knee operations. Those in urban areas received less hip surgery, but there was no association for knee replacement. Controlling for hospital and distance measures did not attenuate the effects. Geographical variation across districts was observed with some districts showing inequality in socio-demographic factors, whereas others showed none at all. CONCLUSIONS: This study found evidence of inequalities in the provision of joint replacement surgery. However, before we can conclude that there is inequity in receipts of healthcare, future research must consider whether these patterns are explained by variations in need across socio-demographic groups.

Knee arthroscopy in England and Ontario: patterns of use, changes over time, and relationship to total knee replacement.

BACKGROUND: The role of knee arthroscopy in the management of osteoarthritis is unclear. The purpose of this study was to examine patterns of use of knee arthroscopy, overall and by diagnostic and sociodemographic subgroups, in countries with comparable health-care systems. METHODS: Administrative databases were used to construct cohorts of adults, twenty years of age or older, who had undergone their first knee arthroscopy in 1993, 1997, 2002, or 2004 either in Ontario, Canada, or in England. For each year, age and sex-standardized rates of knee arthroscopy per 100,000 population were determined overall and by diagnosis, sex, age, and income quintile. Regression analysis, with control for confounders, was used to examine predictors of readmission for primary total knee replacement up to five years after an index knee arthroscopy performed in 1993 or 1997. We also analyzed the records of patients who had undergone primary knee replacement in 2002 to determine the rates of knee arthroscopy in the two years prior to that replacement. RESULTS: In both countries, the proportion of arthroscopic procedures performed to treat internal derangement or dislocation of the knee increased over time; the rates were highest in the highest income quintiles. The study revealed that 4.8% of the patients in England and 8.5% of those in Ontario who had an arthroscopy to treat osteoarthritis in 1997 received a knee replacement within one year after that procedure. The risk of readmission for knee replacement was greater in association with a diagnosis of osteoarthritis, female sex, and an older age at the time of the arthroscopy. Of the patients who had a primary knee replacement in 2002, 2.7% in England and 5.7% in Ontario had undergone a knee arthroscopy in the previous year; the likelihood of the patient having had a prior arthroscopy increased with higher income and increasing age. CONCLUSIONS: Variations in knee arthroscopy rates according to age, sex, income, and diagnosis were identified in both countries. Research to determine if these differences are consistent with need is warranted.