Skip to main content

Table 2 Observed vs predicted goodness of fit statistics

From: The validation of published utility mapping algorithms: an example of EORTC QLQ-C30 and EQ-5D in non-small cell lung cancer

Mapping

Source of utility data

N

Observed mean (95% CI)

Predicted mean (95% CI)

O-P

Mean absolute error (MAE)

Root mean squared error (RMSE)

Young et al. (2015) [12]

Taken from the source paper

771

0.579 (0.555, 0.603)

0.573 (0.552, 0.594)

0.007

0.134

NR

Mapped to AURA data via ‘crosswalk’

4382

0.799 (0.793, 0.805)

0.777 (0.771, 0.783)

0.022

0.087

0.119

Khan and Morris (2014) [13]

Taken from the source paper

2038

0.610 (0.597, 0.623)

0.608 (0.600, 0.616)

0.002

0.10

0.09

Mapped to AURA data via ‘crosswalk’

4382

0.799 (0.793, 0.805)

0.67 (0.668, 0.672)

0.129

0.176

0.211

Khan et al. (2016) [14]

Taken from the source paper

985

0.515 (0.496, 0.534)

0.518 (0.507, 0.529)

−0.003

0.099

0.113

Mapped to AURA data via ‘crosswalk’

4382

0.799 (0.793, 0.805)

0.677 (0.676, 0.678)

0.122

0.178

0.219

  1. Key: N Number of questionnaires completed, NR Not reported, O-P Observed mean utility minus predicted mean utility