Abstract
Significance
In ideal shared decision making (SDM), evidence-based treatment options, their likelihood of success, and the probability of adverse events is discussed with the patient. However, current SDM is fundamentally flawed because evidence for patient-specific treatment effectiveness and patient-specific adverse event risks is lacking. Observational outcome registries are better than randomized clinical trials for determining patient prognostic factors for outcomes and adverse events. No orthopaedic SDM clinical tools exist to predict patient-specific outcomes. Hypothesis: A patient-specific shared decision making tool can predict clinically significant outcomes and adverse events for total knee replacement (TKR) surgery.
Methods
A web–based prospective observational outcome registry collects patient reported outcomes (PROs) for TKR surgery. The data set for TKR surgery includes: (1) European quality of life (EQ-5D); (2) Oxford Knee Score (OKS); (3) Lower Extremity Activity Scale (LEAS); and (4) Pain Likert Scale (PLS). A TKR outcome calculator predicts patient-specific functional outcome with a regression model using patient-specific pre-operative Oxford Knee Scores, diagnosis, co-morbidities, and demographics. Patient-specific joint infection relative risk is calculated using diagnosis, co-morbidities, and demographics. Functional outcomes are presented as minimum clinically important differences (MCIDs). MCID=σΔ/2.
Results
The MCID for the EQ-5D Health State Score (HSS) is 0.094 (0.000–1.000). The MCID for the EQ-5D Visual Analog Scale (VAS) is 9.1 (0–100). The MCID for the OKS is 4.45 (0–48). The MCID for the LEAS is 1.6 (1–18). The MCID for the PLS is 1.4 (0–10).
Examples
(1) A 55-year-old white male with post-traumatic arthritis (ICD-9 716.16, BMI = 28.7, non-diabetic, recently quit smoking) has a pre-operative Oxford Knee Score of 10. His predicted outcome is 6.3 MCIDs and his relative risk of infection is 6.1 (4.4%) (Figure 1). He is expected to have an excellent outcome. His risk of infection can be reduced by using antibiotic-laden cement. Depending on the patient's preferences, he is an excellent candidate for a total knee replacement. (2) A 60-year-old white male with osteoarthritis (ICD-9 715.16, BMI = 25.0, non-diabetic, non-smoker) has a pre-operative Oxford Knee Score of 45. He has full thickness cartilage loss on his medial femoral condyle by MRI only. His predicted outcome is 0.67 MCIDs and his relative risk of infection is 1.9 (1.4%) (Figure 2). He is expected to have a poor outcome even though his risk of infection is low. Although he has full thickness cartilage loss on MRI, his pre-operative Oxford Knee Score of 45 demonstrates that he is very functional and has minimal opportunity for improving his knee function with a total knee replacement. He is a poor candidate for TKR surgery.
Conclusions
The patient-specific SDM tool for TKR surgery can distinguish between excellent and poor surgical candidates when both patients meet radiographic criteria for surgery. The pre-operative Oxford Knee Score assesses knee function and/or disability. Patients with relatively high OKSs are less likely to achieve clinically significant improvements after total knee replacement surgery.
