Authors (including presenting author) :
Chan VWK(1), Chan PK(1), Fu H(1), Cheung MH(1), Cheung A(1) , Chiu KY(1)
Affiliation :
(1) Department of Orthopaedic and Traumatology, Queen Mary Hospital
Introduction :
Preoperative planning is essential to the success of total joint arthroplasties. The ability to anticipate total knee arthroplasty (TKA) sizes before the surgery can have many advantages. The surgical team can be well-prepared for the potential femur and tibia size mismatch and the need for extreme implant sizes for selected patients. Moreover, the operation procedures can be more streamlined, reducing time in trialing and sizing the components. Correct implant size is vital for ligament balancing and preventing soft tissue impingements, which are fundamentals in ensuring good clinical outcomes after TKA.
Unfortunately, the current methods of preoperative templating have lots of drawbacks.
Preoperative templating with radiographic films, either conventional or digital, is the most commonly adopted method for TKA size templating. However, the reported accuracy varies greatly. It is reported to be accurate in 35-70% of the time for the femur, and the tibia implants. The use of digital radiographic templating did not increase the accuracy compared with conventional films. Besides the variation in accuracy and consistency, radiographic templating requires the surgeon to manually overlay the specific implant onto the radiographic image, which is very time and labor-intensive. With the ever-increasing numbers of total joint arthroplasties performed worldwide, we need to develop a more efficient and sustainable way to predict arthroplasty sizes.
It is logical to believe that patient-specific demographic data, such as gender, age, body height (BH), body weight (BW), and body mass index (BMI), provides information about the patient’s bony morphology and size. Some have published TKA sizing equations using the above demographic variables to predict TKA implant sizes. Demographic-based prediction models achieved 90-100% accuracy within one implant size. In addition to the patient-specific characteristics used in the above studies, other reports show a significant correlation between shoe size with TKA component sizes. Van Egmond et al. published that shoe size alone predicted femur and tibia implant within one size in 94% and 86%, respectively. Hence, patient-specific characteristics to predict TKA sizes appear promising and is an efficient alternative to radiographic templating.
Objectives :
We aim to determine the correlation between TKA size and patient-specific demographics, such as gender, age, BH, BW, BMI, and even the patient’s hand and foot sizes. Then, develop a multivariate regression model using the above variables to predict TKA femur and tibia component sizes without using radiographs.
Methodology :
We retrospectively reviewed our institutes' medical and operation records for primary TKA performed from 2015 to 2020. Revision arthroplasties or conversion TKA were excluded. The same team of fellowship trained arthroplasty surgeons performed all the operations (all authors and co authors). The operating surgeon determined the component sizes intraoperatively according to standardized techniques with conventional instruments. Patient specific demographics, such as age, gender, body weight (BW), body height (BH), and body mass index (BMI) at the time of surgery, were retrieved from the medical records. Glove size was obtained using a standard sizing chart (Ansell glove sizing chart). We measure the hand circumference around the widest part of the palm with a measuring tape. A Brannock device was used to determine the patient's foot size. We used the United States men's scale to document all the foot sizes. We reviewed the operation records for the TKA component sizes. To compare component sizes among different TKA designs, we converted the femur and tibia implants into their anterior posterior (AP) and medial lateral (ML) dimensions in millimeters, respectively. The conversions were based on the published implant dimensions from the manufacturer. A total of 808 primary TKAs were included for analysis. The mean age was 70.3 years old (range 49 – 94 years old, standard deviation (SD) 7.4), and 73.3% of the patients were women. The mean BW, BH, and BMI were 67.8kg (range 32.2 – 108 kg, SD 12.3), 1.55m (range 1.2 – 1.87m, SD 0.1), and 28.3 kg/m2 (range 14.9 – 54.7, SD 5.0), respectively. The median foot and glove size was 5.7 and 7.5, while the average hand circumference was 192.9mm (range 160 – 232, SD 12.7).
Pearson's correlation was used to analyze the correlation between collected demographic factors with TKA implant sizes. We used step-wise multivariate linear regression to formulate a statistical model to predict femur and tibia component sizes with all the patient-specific demographics, including hand and foot measurements. Continuous variables are presented in means and standard deviations (SD), while quantitative variables are presented in percentages. P-value
Result & Outcome :
Concerning femur and tibia components, the patient’s gender, BW, BH, foot size, glove size, and hand circumference significantly correlated with their AP and ML dimensions, respectively (p-value < 0.05). Patients’ foot size had the highest correlation coefficient for both components, and the correlation coefficient for femur AP and tibia ML dimensions was 0.689 and 0.712, respectively.
After step-wise multivariate regression analysis, foot size, gender, BH, hand circumference, and BW remain the significant contributors to the prediction model for femur size, with foot size having the highest standardized coefficient of 0.371. The linear regression model for the femur component had an R and R-square of 0.752 and 0.565, respectively (p-value < 0.05).
Femur size (AP in mm) = 37.956812 + 0.855077(Foot size) + 2.07467(Gender*) + 7.537676 (Body height) + 0.016124 (Hand circumference) + 0.018044 (Body weight)
* Gender: 1 for men, 0 for women
The step-wise multivariate regression analysis revealed that foot size, gender, BH, age, BMI, hand circumference, and BW significantly contributed to the prediction model for tibia size. Again, the foot size had the highest standardized coefficient of 0.354. The linear regression model for the tibia component had an R and R-square of 0.795 and 0.633, respectively (p-value < 0.05).
Tibia size (ML in mm) = 28.791833 + 0.852847 (Foot size) + 2.868646 (Gender) + 17.409496 (Body height) + 0.059051 (Age) + 0.221498 (Body-mass-index) + 0.00974 (Hand circumference) - 0.073321 (Body weight)
* Gender: 1 for men, 0 for women
We applied our regression model to predict the femur and tibia size for the various TKA designs. We predicted the femur component size exactly, within one and two sizes in 49.8%, 94%, and 100%, respectively. While, for the tibia component, the prediction was exact, within one and two sizes in 52.6%, 95.3%, and 100%, respectively.
