Table 3 AIC-selected logistic regression model statistics for 2% response threshold. Bolded odds ratio values and p-values indicate statistical significance (α = 0.05)

WEDG

WEDG + V-ARCH

Clinically-Accessible

Laboratory Derived

Clinically-Accessible

Laboratory Derived

Responder: Non-Responder

33:20

33:20

28:25

28:25

Model AIC

58.975

59.445

65.252

63.453

Model Likelihood Ratio (p-value)

p

< 0.001

< 0.001

0.001

< 0.001

H&L Goodness of Fit (p-value)

p

0.815

0.575

0.074

0.109

AUC ROC(c)

c

0.859

0.794

0.829

0.843

95%CI

(0.753, 0.965)

(0.670, 0.918)

(0.709, 0.949)

(0.729, 0.957)

Odds Ratios by Predictor Variable

Gait speed (Clin)

Gait speed (Lab)

Gait speed (Clin)

Gait speed (Lab)

OR

3.118

2.224

2.040

2.240

95%CI

(1.465, 6.633)

(1.301, 3.795)

(1.243, 3.349)

(1.315, 3.815)

p

0.003

0.003

0.005

0.003

KL Grade

KL Grade

Sex

Sex

OR

0.380

0.374

0.180

0.157

95%CI

(0.090, 1.602)

(0.099, 1.416)

(0.034, 0.954)

(0.028, 0.888)

p

0.188

0.148

0.044

0.036

FFI Pain

FFI Pain

FFI Pain

OR

0.948

0.948

0.948

95%CI

(0.900, 0.999)

(0.904, 0.995)

(0.903, 0.994)

p

0.046

0.029

0.029

NRS Pain

NRS Pain

NRS Pain

OR

1.609

1.463

1.512

95%CI

(0.996, 2.598)

(0.974, 2.199)

(0.991, 2.308)

p

0.052

0.067

0.055

FPA (Clin)

OR

0.901

95%CI

(0.779, 1.043)

p

0.164