Preoperative Risk Prediction Models for Short-Term Revision and Death After Total Hip Arthroplasty: Data from the Finnish Arthroplasty Register

Clinical prediction models for patients undergoing total hip arthroplasty: an external validation based on a systematic review and the Dutch Arthroplasty Register

BACKGROUND AND PURPOSE

 External validation is a crucial step after prediction model development. Despite increasing interest in prediction models, external validation is frequently overlooked. We aimed to evaluate whether joint registries can be utilized for external validation of prediction models, and whether published prediction models are valid for the Dutch population with a total hip arthroplasty.

METHODS

 We identified prediction models developed in patients undergoing arthroplasty through a systematic literature search. Model variables were evaluated for availability in the Dutch Arthroplasty Registry (LROI). We assessed the model performance in terms of calibration and discrimination (area under the curve [AUC]). Furthermore, the models were updated and evaluated through intercept recalibration and logistic recalibration.

RESULTS

 After assessing 54 papers, 19 were excluded for not describing a prediction model (n = 16) or focusing on non-TJA populations (n = 3), leaving 35 papers describing 44 prediction models. 90% (40/44) of the prediction models used outcomes or predictors missing in the LROI, such as diabetes, opioid use, and depression. 4 models could be externally validated on LROI data. The models' discrimination ranged between poor and acceptable and was similar to that in the development cohort. The calibration of the models was insufficient. The model performance improved slightly after updating.

CONCLUSION

 External validation of the 4 models resulted in suboptimal predictive performance in the Dutch population, highlighting the importance of external validation studies.

The artificial intelligence advantage: Supercharging exploratory data analysis

Explorative data analysis (EDA) is a critical step in scientific projects, aiming to uncover valuable insights and patterns within data. Traditionally, EDA involves manual inspection, visualization, and various statistical methods. The advent of artificial intelligence (AI) and machine learning (ML) has the potential to improve EDA, offering more sophisticated approaches that enhance its efficacy. This review explores how AI and ML algorithms can improve feature engineering and selection during EDA, leading to more robust predictive models and data-driven decisions. Tree-based models, regularized regression, and clustering algorithms were identified as key techniques. These methods automate feature importance ranking, handle complex interactions, perform feature selection, reveal hidden groupings, and detect anomalies. Real-world applications include risk prediction in total hip arthroplasty and subgroup identification in scoliosis patients. Recent advances in explainable AI and EDA automation show potential for further improvement. The integration of AI and ML into EDA accelerates tasks and uncovers sophisticated insights. However, effective utilization requires a deep understanding of the algorithms, their assumptions, and limitations, along with domain knowledge for proper interpretation. As data continues to grow, AI will play an increasingly pivotal role in EDA when combined with human expertise, driving more informed, data-driven decision-making across various scientific domains. Level of Evidence: Level V - Expert opinion.

Development of a patient-centred tool for use in total hip arthroplasty

BACKGROUND

The aim of this project was to develop a tool using the experience of previous patients to inform patient-centred clinical decision-making in the context of total hip arthroplasty (THA). We sought out the patients' views on what is important for them, leveraging registry data, and providing outcome information that is perceived as relevant, understandable, adapted to a specific patient's profile, and readily available.

METHODS

We created the information tool "Patients like me" in four steps. (1) The knowledge basis was the systematically collected detailed exposure and outcome information from the Geneva Arthroplasty Registry established 1996. (2) From the registry we randomly selected 275 patients about to undergo or having already undergone THA and asked them via interviews and a survey which benefits and harms associated with the operation and daily life with the prosthesis they perceived as most important. (3) The identified relevant data (39 predictor candidates, 15 outcomes) were evaluated using Conditional Inference Trees analysis to construct a classification algorithm for each of the 15 outcomes at three different time points/periods. Internal validity of the results was tested using bootstrapping. (4) The tool was designed by and pre-tested with patients over several iterations.

RESULTS

Data from 6836 primary elective THAs operated between 1996 and 2019 were included. The trajectories for the 15 outcomes from the domains pain relief, activity improvement, complication (infection, dislocation, peri-prosthetic fracture) and what to expect in the future (revision surgery, need for contralateral hip replacement) over up to 20 years after surgery were presented for all patients and for specific patient profiles. The tool was adapted to various purposes including individual use, group sessions, patient-clinician interaction and surgeon information to complement the preoperative planning. The pre-test patients' feedback to the tool was unanimously positive. They considered it interesting, clear, complete, and complementary to other information received.

CONCLUSION

The tool based on a survey of patients' perceived concerns and interests and the corresponding long-term data from a large institutional registry makes past patients' experience accessible, understandable, and visible for today's patients and their clinicians. It is a comprehensive illustration of trajectories of relevant outcomes from previous "Patients like me". This principle and methodology can be applied in other medical fields.

Prediction of Early Adverse Events After THA: A Comparison of Different Machine-Learning Strategies Based on 262,356 Observations From the Nordic Arthroplasty Register Association (NARA) Dataset

OBJECTIVE

Preoperative risk prediction models can support shared decision-making before total hip arthroplasties (THAs). Here, we compare different machine-learning (ML) approaches to predict the six-month risk of adverse events following primary THA to obtain accurate yet simple-to-use risk prediction models.

METHODS

We extracted data on primary THAs (N = 262,356) between 2010 and 2018 from the Nordic Arthroplasty Register Association dataset. We benchmarked a variety of ML algorithms in terms of the area under the receiver operating characteristic curve (AUROC) for predicting the risk of revision caused by periprosthetic joint infection (PJI), dislocation or periprosthetic fracture (PPF), and death. All models were internally validated against a randomly selected test cohort (one-third of the data) that was not used for training the models.

RESULTS

The incidences of revisions because of PJI, dislocation, and PPF were 0.8%, 0.4%, and 0.3%, respectively, and the incidence of death was 1.2%. Overall, Lasso regression with stable iterative variable selection (SIVS) produced models using only four to five input variables but with AUROC comparable to more complex models using all 32 variables available. The SIVS-based Lasso models based on age, sex, preoperative diagnosis, bearing couple, fixation, and surgical approach predicted the risk of revisions caused by PJI, dislocations, and PPF, as well as death, with AUROCs of 0.61, 0.67, 0.76, and 0.86, respectively.

CONCLUSION

Our study demonstrates that satisfactory predictive potential for adverse events following THA can be reached with parsimonious modeling strategies. The SIVS-based Lasso models may serve as simple-to-use tools for clinical risk assessment in the future.

Frailty Among Total Hip and Knee Arthroplasty Recipients: Epidemiology and Propensity Score-weighted Analysis of Effect on In-hospital Postoperative Outcomes

INTRODUCTION

Total joint arthroplasty (TJA) is one of the most successful and frequently performed procedures in the United States. The number of these procedures is projected to continue growing rapidly in the coming years, and with it comes the demand for more sophisticated perioperative risk and complication assessment. This study examines the effect of frailty on postoperative inpatient complications and hospital resource utilization after TJA.

METHODS

Discharge data from the National Inpatient Sample were used to identify all patients aged 50 years or older who underwent TJA between 2006 and 2015. Nonelective admissions and hip fractures were excluded. Patients were stratified into two groups with and without concomitant ICD-9 diagnostic criteria that qualified them has having frailty. An analysis comparing the 2 groups' epidemiology, medical comorbidities, and propensity score-weighted postoperative clinical and economic outcomes was done.

RESULTS

A total of 7,854,890 TJAs were included in this analysis, with 136,516 meeting the criteria for frailty and 7,718,374 being nonfrail. Among these patients, the average age was 67.3 years and the female distribution was 61.1%. Frail patients were found to have markedly higher rates of all but two individual comorbidities constituting the Modified Elixhauser Profile compared with nonfrail patients. Compared with the control group, frail patients were found to have increased risk of any postoperative complication, central nervous system complications, hematoma/seroma, wound dehiscence, infection, and postoperative anemia. Frail patients also had longer length of stay, higher discharge to rehabilitation facilities, and higher hospital charges.

DISCUSSION

Patients with frailty undergoing TJA procedures are at a markedly higher risk for developing postoperative complications and worse hospital economic outcomes. As this patient population continues to increase, it is imperative for clinicians to use their risk factors in optimizing their perioperative care and support.

Patient-rated satisfaction and improvement following hip and knee replacements: Development of prediction models

RATIONALE

Effective preoperative assessments of determinants of health status and function may improve postoperative outcomes.

AIMS AND OBJECTIVES

We developed risk scores of preoperative patient factors and patient-reported outcome measures (PROMs) as predictors of patient-rated satisfaction and improvement following hip and knee replacements.

PATIENTS AND METHODS

Prospectively collected National Health Service and independent sector patient data (n = 30,457), including patients' self-reported demographics, comorbidities, PROMs (Oxford Hip/Knee score (OHS/OKS) and European Quality of Life (EQ5D index and health-scale), were analysed. Outcomes were defined as patient-reported satisfaction and improvement following surgery at 7-month follow-up. Univariable and multivariable-adjusted logistic regressions were undertaken to build prediction models; model discrimination was evaluated with the concordance index (c-index) and nomograms were developed to allow the estimation of probabilities.

RESULTS

Of the 14,651 subjects with responses for satisfaction following hip replacements 564 (3.8%) reported dissatisfaction, and 1433 (9.2%) of the 15,560 following knee replacement reported dissatisfaction. A total of 14,662 had responses for perceived improvement following hip replacement (lack of improvement in 391; 2.7%) and 15,588 following knee replacement (lack of improvements in 1092; 7.0%). Patients reporting poor outcomes had worse preoperative PROMs. Several factors, including age, gender, patient comorbidities and EQ5D, were included in the final prediction models: C-indices of these models were 0.613 and 0.618 for dissatisfaction and lack of improvement, respectively, for hip replacement and 0.614 and 0.598, respectively, for knee replacement.

CONCLUSIONS

Using easily accessible preoperative patient factors, including PROMs, we developed models which may help predict dissatisfaction and lack of improvement following hip and knee replacements and facilitate risk stratification and decision-making processes.

A retrospective study of risk factors, causative micro-organisms and healthcare resources consumption associated with prosthetic joint infections (PJI) using the Clinical Practice Research Datalink (CPRD) Aurum database

BACKGROUND

Prosthetic joint infection (PJI) is a serious complication after joint replacement surgery and it is associated with risk of mortality and morbidity along with high direct costs.

METHODS

The Clinical Practice Research Datalink (CPRD) data were utilized to quantify PJI incidence after hip or knee replacement up to 5 years after implant and a variety of risk factors related to patient characteristics, medical and treatment history along with characteristics of the original surgery were analyzed through Cox proportional hazard.

RESULTS

221,826 patients (individual joints 283,789) met all the inclusion and exclusion criteria of the study; during the study follow-up period (5 years), 707 and 695 PJIs were diagnosed in hip and knee, respectively. Patients undergoing joint replacement surgery during an unscheduled hospitalization had greater risk of PJI than patients whose surgery was elective; similarly, the risk of developing PJI after a secondary hip or knee replacement was about 4 times greater than after primary arthroplasty when adjusted for all other variables considered. A previous diagnosis of PJI, even in a different joint, increased the risk of a further PJI. Distribution of average LoS per each hospitalization caused by PJI exhibited a right skewed profile with median duration [IQR] duration of 16 days [8-32] and 13 days [7.25-32] for hip and knee, respectively. PJIs causative micro-organisms were dependent on the time between initial surgery and infection offset; early PJI were more likely to be multispecies than later (years after surgery); the identification of Gram- pathogens decreased with increasing post-surgery follow-up.

CONCLUSIONS

This study offers a contemporary assessment of the budgetary and capacity (number and duration of hospitalizations along with the number of Accident and Emergency (A&E) visits) posed by PJIs in UK for the national healthcare system (NHS). The results to provide risk management and planning tools to health providers and policy makers in order to fully assess technologies aimed at controlling and preventing PJI. The findings add to the existing evidence-based knowledge surrounding the epidemiology and burden of PJI by quantifying patterns of PJI in patients with a relatively broad set of prevalent comorbidities.

A Surgeon's Guide to Understanding Artificial Intelligence and Machine Learning Studies in Orthopaedic Surgery

PURPOSE OF REVIEW

In recent years, machine learning techniques have been increasingly utilized across medicine, impacting the practice and delivery of healthcare. The data-driven nature of orthopaedic surgery presents many targets for improvement through the use of artificial intelligence, which is reflected in the increasing number of publications in the medical literature. However, the unique methodologies utilized in AI studies can present a barrier to its widespread acceptance and use in orthopaedics. The purpose of our review is to provide a tool that can be used by practitioners to better understand and ultimately leverage AI studies.

RECENT FINDINGS

The increasing interest in machine learning across medicine is reflected in a greater utilization of AI in recent medical literature. The process of designing machine learning studies includes study design, model choice, data collection/handling, model development, training, testing, and interpretation. Recent studies leveraging ML in orthopaedics provide useful examples for future research endeavors. This manuscript intends to create a guide discussing the use of machine learning and artificial intelligence in orthopaedic surgery research. Our review outlines the process of creating a machine learning algorithm and discusses the different model types, utilizing examples from recent orthopaedic literature to illustrate the techniques involved.

Improved risk prediction of chemotherapy-induced neutropenia-model development and validation with real-world data

BACKGROUND

The existing risk prediction models for chemotherapy-induced febrile neutropenia (FN) do not necessarily apply to real-life patients in different healthcare systems and the external validation of these models are often lacking. Our study evaluates whether a machine learning-based risk prediction model could outperform the previously introduced models, especially when validated against real-world patient data from another institution not used for model training.

METHODS

Using Turku University Hospital electronic medical records, we identified all patients who received chemotherapy for non-hematological cancer between the years 2010 and 2017 (N = 5879). An experimental surrogate endpoint was first-cycle neutropenic infection (NI), defined as grade IV neutropenia with serum C-reactive protein >10 mg/l. For predicting the risk of NI, a penalized regression model (Lasso) was developed. The model was externally validated in an independent dataset (N = 4594) from Tampere University Hospital.

RESULTS

Lasso model accurately predicted NI risk with good accuracy (AUROC 0.84). In the validation cohort, the Lasso model outperformed two previously introduced, widely approved models, with AUROC 0.75. The variables selected by Lasso included granulocyte colony-stimulating factor (G-CSF) use, cancer type, pre-treatment neutrophil and thrombocyte count, intravenous treatment regimen, and the planned dose intensity. The same model predicted also FN, with AUROC 0.77, supporting the validity of NI as an endpoint.

CONCLUSIONS

Our study demonstrates that real-world NI risk prediction can be improved with machine learning and that every difference in patient or treatment characteristics can have a significant impact on model performance. Here we outline a novel, externally validated approach which may hold potential to facilitate more targeted use of G-CSFs in the future.

Stable Iterative Variable Selection

Motivation

The emergence of datasets with tens of thousands of features, such as high-throughput omics biomedical data, highlights the importance of reducing the feature space into a distilled subset that can truly capture the signal for research and industry by aiding in finding more effective biomarkers for the question in hand. A good feature set also facilitates building robust predictive models with improved interpretability and convergence of the applied method due to the smaller feature space.

Results

Here, we present a robust feature selection method named Stable Iterative Variable Selection (SIVS) and assess its performance over both omics and clinical data types. As a performance assessment metric, we compared the number and goodness of the selected feature using SIVS to those selected by Least Absolute Shrinkage and Selection Operator regression. The results suggested that the feature space selected by SIVS was, on average, 41% smaller, without having a negative effect on the model performance. A similar result was observed for comparison with Boruta and caret RFE.

Availability and implementation

The method is implemented as an R package under GNU General Public License v3.0 and is accessible via Comprehensive R Archive Network (CRAN) via https://cran.r-project.org/package=sivs.

Supplementary information

Supplementary data are available at Bioinformatics online.