Table 4 Error measurements for predicting utility values based on OHS and OKS questionnaires using the different models. Estimation sample

From: Mapping analysis to predict EQ-5D-5 L utility values based on the Oxford Hip Score (OHS) and Oxford Knee Score (OKS) questionnaires in the Spanish population suffering from lower limb osteoarthritis

		MAE	MSE	SE^a	ICC
Model	Dependent Variable
Hip osteoarthritis, OHS,n = 352
OLS	Utility	0.1263 (0.1157–0.1368)	0.0259 (0.0217–0.0301)	0.0200 (0.0194–0.0207)	0.825 (0.789–0.856)
Tobit	Utility	0.1228 (0.1125–0.1331)	0.0247 (0.0206–0.0288)	0.0223 (0.0216–0.0231)	0.832 (0.797–0.862)
GLM	Utility	0.1156 (0.1058–0.1256)	0.0222 (0.0184–0.0260)	0.0216 (0.0204–0.0227)	0.856 (0.826–0.882)
Beta reg	Utility	0.1199 (0.1093–0.1304)	0.0244 (0.0200–0.0288)	0.0067 (0.0066–0.0069)	0.861 (0.832–0.886)
Knee Osteoarthritis, OKS,n = 390
OLS	Utility	0.1340 (0.1232–0.1448)	0.0297 (0.0249–0.0343)	0.0210 (0.0204–0.0216)	0.750 (0.703–0.790)
Tobit	Utility	0.1313 (0.1206–0.1421)	0.0296 (0.0242–0.0334)	0.0221 (0.0216–0.0228)	0.751 (0.705–0.792)
GLM	Utility	0.1248 (0.1140–0.1356)	0.0272 (0.0224–0.0319)	0.0228 (0.0218–0.0239)	0.774 (0.731–0.811)
Beta reg	Utility	0.1287 (0.1176–0.1397)	0.0287 (0.0240–0.0335)	0.0075 (0.0074–0.0076)	0.798 (0.769–0.823)

MAE Mean absolute error, MSE Mean squared error
SE^a: Standard error reported are the mean values for the original predictions (disutility for GLM and transformed utility (0–1) for beta reg)
CI 95% in parentheses
ICC Intraclass correlation coefficient. Observed-predicted values (absolute agreement)

Back to article page