Prospective Validation of a Demographically Based Primary Total Knee Arthroplasty Size Calculator

Demographic Data Reliably Predicts Total Hip Arthroplasty Component Size

BACKGROUND

Preoperative radiographic templating for total hip arthroplasty (THA) has been shown to be inaccurate, although essential for streamlining operating room efficiency. Although demographic data have shown to predict total knee arthroplasty component sizes, the unique contour and design among femoral stem implants have limited a similar application for hip arthroplasty. The purpose of this study was to determine whether demographic data may predict cementless THA size independent of the stem design.

METHODS

A consecutive series of 1,653 index cementless metaphyseal-fitting THAs were reviewed between 2007 and 2019. This included 12 unique femoral component designs, 6 acetabular component designs, 60 femur size-design combinations, and 23 acetabular size-design combinations. Implanted component sizes and patient demographic data were collected, including gender, height, weight, laterality, age, race, and ethnicity. Multivariate linear regressions were formulated to predict implanted femur and acetabular component sizes from the demographic data.

RESULTS

There was a significant linear correlation between gender, implant model, age, height, and weight for femur (R² = 0.778; P < .001) and acetabular (R² = 0.491; P < .001) sizes. Calculated femur and acetabular component sizes averaged within 0.97 and 0.95 sizes of those implants, respectively. Femur and acetabular sizes were predicted within 1 size 79.1% and 78.2% and within 2 sizes 94.3% and 94.6% of the time, respectively.

CONCLUSIONS

Multivariate regression models were created based on specific demographics data to predict femur and acetabular component sizes. The model allows for simplified preoperative planning and potential cost savings implementation. A free phone application named EasyTJA was constructed for ease of implementation.

Machine learning algorithms predict within one size of the final implant ultimately used in total knee arthroplasty with good-to-excellent accuracy

PURPOSE

To develop a novel machine learning algorithm capable of predicting TKA implant sizes using a large, multicenter database.

METHODS

A consecutive series of primary TKA patients from two independent large academic and three community medical centers between 2012 and 2020 was identified. The primary outcomes were final tibial and femoral implant sizes obtained from an automated inventory system. Five machine learning algorithms were trained using six routinely collected preoperative features (age, sex, height, weight, and body mass index). Algorithms were validated on an independent set of patients and evaluated through accuracy, mean absolute error (MAE), and root mean-squared error (RMSE).

RESULTS

A total of 11,777 patients were included. The support vector machine (SVM) algorithm had the best performance for femoral component size(MAE = 0.73, RMSE = 1.06) with accuracies of 42.2%, 88.3%, and 97.6% for predicting exact size, ± one size, and ± two sizes, respectively. The elastic-net penalized linear regression (ENPLR) algorithm had the best performance for tibial component size (MAE 0.70, RMSE = 1.03) with accuracies of 43.8%, 90.0%, and 97.7% for predicting exact size, ± one size, and ± two sizes, respectively.

CONCLUSION

Machine learning algorithms demonstrated good-to-excellent accuracy for predicting within one size of the final tibial and femoral components used for TKA. Patient height and sex were the most important factors for predicting femoral and tibial component size, respectively. External validation of these algorithms is imperative prior to use in clinical settings.

LEVEL OF EVIDENCE

Case-control, III.

Can Component Size in Total Knee Arthroplasty Be Predicted Preoperatively?-An Analysis of Patient Characteristics

INTRODUCTION

 Accurately predicting component sizing in total knee arthroplasty (TKA) can ensure appropriate implants are readily available, avoiding complications from malsizing while also reducing cost by improving workflow efficiency through a reduction in instrumentation. This study investigated the utility of demographic variables to reliably predict TKA component sizes.

METHODS AND MATERIALS

 A retrospective chart review of 337 patients undergoing primary TKA was performed. Patient characteristics (age, sex, race, height, weight) were recorded along with implant and shoe size. Correlation between shoe size and TKA component size was assessed using Pearson's correlation coefficient and linear regression analysis using three models: (A) standard demographic variables, (B) shoe size, and (C) combination of both models.

RESULTS

 Shoe size demonstrated the strongest correlation with femoral anteroposterior (FAP) (p < 0.001) followed by height (p < 0.001). Conversely, height exhibited the strongest correlation with tibial mediolateral (TML) (p < 0.001) followed by shoe size (p < 0.001). Model C was able to correctly predict both the femur and tibia within one and two sizes in 83.09 and 98.14% of cases, respectively. Individually, model C predicted the FAP within one and two sizes in 83.09 and 96.14% of cases, and the TML in 98.81 and 100% of cases, respectively.

CONCLUSION

 A patient's shoe size demonstrates a strong correlation to the TKA implant size, and when combined with standard demographic variables the predictive reliability is further increased. Here, we present a predictive model for implant sizing based solely on easily attainable demographic variables, that will be useful for preoperative planning to improve surgical efficiency.

LEVEL OF EVIDENCE

 II, Diagnostic.

Magnification assessment of radiographs for knee replacement (MARKeR) - A pilot study in a low-resource setting

Background

Selecting the correct size of implants to be used in total knee arthroplasty is critical for a successful outcome. Marker-less templating systems use an institutionally derived magnification factor for all radiographs.

Purpose

To determine the institutional magnification of knee radiographs for patients awaiting total knee arthroplasty.

Material and Methods

Eighty patients awaiting total knee arthroplasty underwent preoperative knee radiographs using a standardized protocol. A marker attached to the patients’ knees at the level of the knee joint was used to calculate the magnification factor on both anteroposterior (AP) and lateral (LAT) views. Two independent observers estimated the magnification to determine the intra and inter-observer reliability.

Results

The mean magnification of the AP (15.3%) radiograph was significantly greater than the LAT (12.1%) radiograph (p< 0.0001). Patients with absent markers on their radiographs were heavier than patients in whom the marker was visible (84.7 kgs vs. 76.6 kgs, p=0.01). No marker was visible on the radiographs in 56.3% (45/80) of patients. There was excellent inter and intra-observer reliability of both the AP and LAT measurements.

Conclusion

After standardizing the protocol for preoperative knee radiographs, our results show significantly greater institutional magnification of the anteroposterior compared with the lateral images. Accurate templating in knee arthroplasty requires both radiographic images. To reduce errors in implant sizing, we recommend surgeons use different institutional magnification factors for the anteroposterior and lateral radiographs.

Validation and performance of a machine-learning derived prediction guide for total knee arthroplasty component sizing

INTRODUCTION

Anticipation of patient-specific component sizes prior to total knee arthroplasty (TKA) is essential to avoid excessive cost associated with additional surgical trays and morbidity associated with imperfect sizing. Current methods of size prediction, including templating, are inconsistent and time-consuming. Machine learning (ML) algorithms may allow for accurate TKA component size prediction with the ability to make predictions in real-time.

METHODS

Consecutive patients receiving primary TKA between 2012 and 2020 from two large tertiary academic and six community hospitals were identified. The primary outcomes were the final femoral and tibial component sizes extracted from automated inventory systems. Five ML algorithms were trained with routinely corrected demographic variables (age, height, weight, body mass index, and sex) using 80% of the study population and internally validated on an independent set of the remaining 20% of patients. Algorithm performance was evaluated through accuracy, mean absolute error (MAE), and root mean-squared error (RMSE).

RESULTS

A total of 17,283 patients that received one of 9 TKA implants from independent manufacturers were included. The SGB model accuracy for predicting ± 4-mm of the true femoral anteroposterior diameter was 83.6% and for ± 1 size of the true femoral component size was 95.0%. The SGB model accuracy for predicting ± 4-mm of the true tibial medial/lateral diameter was 83.0% and for ± 1 size of the true tibial component size was 97.8%. Patient sex was the most influential feature in terms of informing the SGB model predictions for both femoral and tibial component sizing. A TKA implant sizing application was subsequently created.

CONCLUSION

Novel machine learning algorithms demonstrated good to excellent performance for predicting TKA component size. Patient sex appears to contribute an important role in predicting TKA size. A web-based real-time prediction application was created capable of integrating patient specific data to predict TKA size, which will require external validation prior to clinical use.

Accuracy of one-dimensional templating on linear EOS radiography allows template-directed instrumentation in total knee arthroplasty

BACKGROUND

Templating for total knee arthroplasty (TKA) is routinely performed on two-dimensional standard X-ray images and allows template-directed instrumentation. To date, there is no report on one-dimensional (1D) anteroposterior (AP) templating not requiring specific templating software. We aim to describe a novel technique and explore its reliability, accuracy and potential cost-savings.

METHODS

We investigated a consecutive series of TKAs at one institution between January and July 2019. Patients with preoperative low-dose linear AP EOS radiography images were included. Implant component sizes were retrospectively templated on the AP view with the hospitals imaging viewing software by two observers who were blinded to the definitive implant size. Planning accuracy as well as inter- and intra-observer reliability was calculated. Cost-savings were estimated based on the reduction of trays indicated by the 1D templating size estimations.

RESULTS

A total of 141 consecutive TKAs in 113 patients were included. Accuracy of 1D templating was as follows: exact match in 53% femoral and 63% tibial components, within one size in 96% femoral and 98% tibial components. Overall 58% of TKA components were planned correctly and 97% within one size. Inter- and intra-rater reliability was good (κ = 0.66) and very good (κ = 0.82), respectively. This templating process can reduce instrumentation from six to three trays per case and therefore halve sterilisation costs.

CONCLUSIONS

The new 1D templating method using EOS AP imaging predicts component sizes in TKA within one size 97% of the time and can halve the number of instrumentation trays and sterilisation costs.

Machine Learning Predicts Femoral and Tibial Implant Size Mismatch for Total Knee Arthroplasty

Background

Despite reasonable accuracy with preoperative templating, the search for an optimal planning tool remains an unsolved dilemma. The purpose of the present study was to apply machine learning (ML) using preoperative demographic variables to predict mismatch between templating and final component size in primary total knee arthroplasty (TKA) cases.

Methods

This was a retrospective case-control study of primary TKA patients between September 2012 and April 2018. The primary outcome was mismatch between the templated and final implanted component sizes extracted from the operative database. The secondary outcome was mismatch categorized as undersized and oversized. Five supervised ML algorithms were trained using 6 demographic features. Prediction accuracies were obtained as a metric of performance for binary mismatch (yes/no) and multilevel (undersized/correct/oversized) classifications.

Results

A total of 1801 patients were included. For binary classification, the best-performing algorithm for predicting femoral and tibial mismatch was the stochastic gradient boosting model (area under the curve: 0.76/0.72, calibration intercepts: 0.05/0.05, calibration slopes: 0.55/0.7, and Brier scores: 0.20/0.21). For multiclass classification, the best-performing algorithms had accuracies of 83.9% and 82.9% for predicting the concordance/mismatch of the femoral and tibial implant, respectively. Model predictions of greater than 51.0% and 47.9% represented high-risk thresholds for femoral and tibial sizing mismatch, respectively.

Conclusions

ML algorithms predicted templating mismatch with good accuracy. External validation is necessary to confirm the performance and reliability of these algorithms. Predicting sizing mismatch is the first step in using ML to aid in the prediction of final TKA component sizes. Further studies to optimize parameters and predictions for the algorithms are ongoing.

Demographic data is more predictive of component size than digital radiographic templating in total knee arthroplasty

Background

Preoperative radiographic templating for total knee arthroplasty (TKA) has been shown to be inaccurate. Patient demographic data, such as gender, height, weight, age, and race, may be more predictive of implanted component size in TKA.

Materials and methods

A multivariate linear regression model was designed to predict implanted femoral and tibial component size using demographic data along a consecutive series of 201 patients undergoing index TKA. Traditional, two-dimensional, radiographic templating was compared to demographic-based regression predictions on a prospective 181 consecutive patients undergoing index TKA in their ability to accurately predict intraoperative implanted sizes. Surgeons were blinded of any predictions.

Results

Patient gender, height, weight, age, and ethnicity/race were predictive of implanted TKA component size. The regression model more accurately predicted implanted component size compared to radiographically templated sizes for both the femoral (P = 0.04) and tibial (P < 0.01) components. The regression model exactly predicted femoral and tibial component sizes in 43.7 and 43.7% of cases, was within one size 90.1 and 95.6% of the time, and was within two sizes in every case. Radiographic templating exactly predicted 35.4 and 36.5% of cases, was within one size 86.2 and 85.1% of the time, and varied up to four sizes for both the femoral and tibial components. The regression model averaged within 0.66 and 0.61 sizes, versus 0.81 and 0.81 sizes for radiographic templating for femoral and tibial components.

Conclusions

A demographic-based regression model was created based on patient-specific demographic data to predict femoral and tibial TKA component sizes. In a prospective patient series, the regression model more accurately and precisely predicted implanted component sizes compared to radiographic templating.

Level of evidence

Prospective cohort, level II.

Compilation and Analysis of Web-Based Orthopedic Personalized Predictive Tools: A Scoping Review

Web-based personalized predictive tools in orthopedic surgery are becoming more widely available. Despite rising numbers of these tools, many orthopedic surgeons may not know what tools are available, how these tools were developed, and how they can be utilized. The aim of this scoping review is to compile and synthesize the profile of existing web-based orthopedic tools. We conducted two separate PubMed searches-one a broad search and the second a more targeted one involving high impact journals-with the aim of comprehensively identifying all existing tools. These articles were then screened for functional tool URLs, methods regarding the tool's creation, and general inputs and outputs required for the tool to function. We identified 57 articles, which yielded 31 unique web-based tools. These tools involved various orthopedic conditions (e.g., fractures, osteoarthritis, musculoskeletal neoplasias); interventions (e.g., fracture fixation, total joint arthroplasty); outcomes (e.g., mortality, clinical outcomes). This scoping review highlights the availability and utility of a vast array of web-based personalized predictive tools for orthopedic surgeons. Increased awareness and access to these tools may allow for better decision support, surgical planning, post-operative expectation management, and improved shared decision-making.

Patient Demographics and Anthropometric Measurements Predict Tibial and Femoral Component Sizing in Total Knee Arthroplasty

Background

Accurate sizing is critical for the overall success of a total knee arthroplasty (TKA). This study's primary purpose was to investigate the ability to predict the tibial and femoral component size in a single implant system from patient demographics and anthropometric data. A secondary goal was to compare the predicted tibial and femoral component sizes from our statistical model with a previously validated electronic application used to predict the implant size.

Methods

A consecutive series of 484 patients undergoing a primary TKA at a single institution was reviewed. Data on height, weight, body mass index, sex, age, and component size were collected. A proportional odds model was developed to predict tibial and femoral component sizes. The relationship between the proportional odds model predictions was also compared with the component sizes determined by the Arthroplasty Size Predictor electronic application.

Results

Weight, height, and sex predicted the implanted component size with an accuracy of 54.0% (n = 247/484) for the tibia and 51.1% (n = 231/484) for the femur. The accuracy improved to 94.4% (n = 457/484) for the tibia and 93.4% (n = 452/484) for the femur within ±1 component size. Our data are highly correlated to the Arthroplasty Size Predictor for the predicted tibial component size (ρ = 0.91, P < .001) and femoral component size (ρ = 0.89, P < .001).

Conclusions

Our novel templating model may improve operative efficiency for a single TKA system. Our findings have a high concordance with a widely available electronic application used to predict implant sizes for a variety of TKA systems.

Predicting implant size in total knee arthroplasty using demographic variables

AIMS

The purpose of this investigation was to determine the relationship between height, weight, and sex with implant size in total knee arthroplasty (TKA) using a multivariate linear regression model and a Bayesian model.

METHODS

A retrospective review of an institutional registry was performed of primary TKAs performed between January 2005 and December 2016. Patient demographics including patient age, sex, height, weight, and body mass index (BMI) were obtained from registry and medical record review. In total, 8,100 primary TKAs were included. The mean age was 67.3 years (SD 9.5) with a mean BMI of 30.4 kg/m² (SD 6.3). The TKAs were randomly split into a training cohort (n = 4,022) and a testing cohort (n = 4,078). A multivariate linear regression model was created on the training cohort and then applied to the testing cohort . A Bayesian model was created based on the frequencies of implant sizes in the training cohort. The model was then applied to the testing cohort to determine the accuracy of the model at 1%, 5%, and 10% tolerance of inaccuracy.

RESULTS

Height had a relatively strong correlation with implant size (femoral component anteroposterior (AP) Pearson correlation coefficient (ρ) = 0.73, p < 0.001; tibial component mediolateral (ML) ρ = 0.77, p < 0.001). Weight had a moderately strong correlation with implant size, (femoral component AP ρ = 0.46, p < 0.001; tibial ML ρ = 0.48, p < 0.001). There was a significant linear correlation with height, weight, and sex with implant size (femoral component R² = 0.607, p < 0.001; tibial R² = 0.695, p < 0.001). The Bayesian model showed high accuracy in predicting the range of required implant sizes (94.4% for the femur and 96.6% for the tibia) accepting a 5% risk of inaccuracy.

CONCLUSION

Implant size was correlated with basic demographic variables including height, weight, and sex. The linear regression and Bayesian models accurately predicted required implant sizes across multiple manufacturers based on height, weight, and sex alone. These types of predictive models may help improve operating room and implant supply chain efficiency. Level of Evidence: Level IV Cite this article: Bone Joint J 2020;102-B(6 Supple A):85-90.

Prospective Comparison of Available Primary Total Knee Arthroplasty Sizing Equations

BACKGROUND

Several studies have proposed regression equations that can increase the accuracy of predicting femur and tibia component sizes for total knee arthroplasty (TKA). This study compared available regression equations in their ability to prospectively predict component size in a unique patient series.

METHODS

Demographic data and implanted femur and tibia TKA component sizes were collected on a consecutive 382 patients undergoing index TKA. Equations by Bhowmik-Stoker et al, Ren et al, Sershon et al, and Miller et al were identified that used age, race, ethnicity, gender, height, weight, or body mass index. Equation outputs were converted to implant-corrected sizes and compared to the implanted component.

RESULTS

Femur and tibia sizes were accurately predicted within 1 size 88% and 92%, 84% and 86%, and 79% and 92% for Bhowmik-Stoker et al, Sershon et al, and Miller et al, respectively. Ren et al was within 1 tibia size 88% of the time. Adding one more common implant size improved this accuracy by an average of 9.1% and 6.6% for the femur and tibia, respectively. For femur components, Bhowmik-Stoker et al outperformed Sershon et al by 0.14 sizes (P < .001) and Miller et al by 0.21 sizes (P < .001) on average. For tibia components, Bhowmik-Stoker et al outperformed Sershon et al by 0.09 sizes (P = .028) and Ren et al by 0.11 sizes (P = .005) on average.

CONCLUSION

Equations by Bhowmik-Stoker et al more accurately predicted implanted TKA size. In cases of greater uncertainty, the practicing surgeon may err on having more common TKA sizes available.