curve (ROC) methods to minimize misclassification bias. As the positive predictive value (PPV) increased when II was calculated by tertile (T) of baseline severity, we report those results. Results: Sample mean age was 65.1 years (range 43e90) and 55% were female. 80% reported at least one comorbid condition. For pain, T1 (mean 26.0, range 0e35) the II was 45; T2 (mean 48.7, range 40e55) the II was 30; and, T3 (mean 71.1, range 60e100) the II was 25. The overall PPV was 0.92. For function, T1 (mean 29.1, range 0e41) the II was 37; T2 (mean 50.6, range 43e59) the II was 20; and, T3 (mean 71.3, range 60e99) the II was 17. The overall PPV was 0.91. For higher demand activities, T1 (mean 49.3, range 18.7e56.2) the II was 8.8; T2 (mean 61.7, range 56.8e66.4) the II was 8.8; and, T3 (mean 75.6, range 67.3e100) the II was 5.3. The overall PPV was 0.87. Overall, 62%, 77% and 74% achieved the II for pain, function and high demand activities. The vast majority of participants responded 'yes' to the PASS anchor question (only 28, 29 and 33 (<15%) participants reported 'no' for pain, function and higher demand activities respectively) precluding PASS calculation by baseline severity. The PASS for each of pain, function and higher demand activities were 65.0, 68.0 and 70.2/100 respectively based on ROC analyses. The PPV ranged from 0.97e0.98. Conclusions: This work quantifies the II for the WOMAC pain and function subscales and, for the first time, for higher demand activities as measured by the LLDI limitations subscale. These II values accurately classified between 87 and 92% of participants. Given the known large effect for many with TKR, we determined cut-points based on a 'good deal better or more' resulting in somewhat higher values than those reported in some literature depending on baseline severity. Unlike