Background: Preoperative templating of total knee arthroplasty (TKA) components can help in choosing appropriate implant size prior to surgery. While long limb radiographs have been shown to be beneficial in assessing alignment, disease state, and previous pathology or trauma, their accuracy for size prediction has not been proven. In an attempt to improve templating precision, surgeons have looked to develop other predictive models for component size determination utilizing patient characteristics. The purpose of this study was to: 1) Identify which patient characteristics influence the tibial and femoral component sizes; 2) Construct models for size prediction; 3) Test the generated models at five different centers; and 4) Compare implant survivorship and patient characteristics between those who did or did not receive an implant within one size of the prediction.
Materials and methods: Demographic data was collected on 741 patients (845 knees) as part of a multicenter clinical trial. Correlation between component size and patient demographic data were examined using Pearson coefficients, and significant variables were included into a multivariate-linear-regression model to determine "predicted size." Operative surgeon notes and postoperative radiographs were used to determine "actual size." Predictive equations were constructed for both femoral and tibial components and were tested at five different centers. Implant survivorship and patient characteristics were compared between those who did and did not receive an implant within one size of the prediction.
Results: The strongest predictors of component size were height, weight, and gender (p<0.01), followed by ethnicity (p=0.03) and age (p=0.03). Predictive equations were constructed for both tibial and femoral components. The model predicted the component fit within one size in 94% (r2=0.68) and 96% (r2=0.73) of femoral and tibial components. Cases beyond ±1 sizes did not have notable device-specific adverse events with Kaplan-Meier survivorship of 100% at five years.
Conclusion: Demographic models are an effective tool in component size prediction prior to TKA. This model has implications in reducing the need for preoperative radiographic templating, potentially resulting in increasing surgeon efficiency and possibly reducing hospital implant inventory. This may be particularly important for ambulatory or outpatient surgery centers.