Table 3.

Summary accuracy measures in the validation sample for our proposed model and two of the models proposed in [2]^a

	Our proposed model Estimate (95 % confidence interval)	Previously validated core model Estimate (95 % confidence interval)	Previously validated extended model Estimate (95 % confidence interval)
Sensitivity	72.3 % (66.4–77.6 %)	83.8 % (78.7–88.0 %)	92.7 % (88.6–95.4 %)
Specificity	62.5 % (54.9–69.6 %)	47.7 % (40.2–55.4 %)	44.3 % (36.9–52.0 %)
Positive predictive value	74.0 % (68.1–79.2 %)	70.3 % (64.8–75.3 %)	71.1 % (65.9–75.8 %)
Negative predictive value	60.4 % (52.9–67.5 %)	66.7 % (57.6–74.7 %)	80.4 % (70.9–87.5 %)
Correct classification	68.3 % (63.7–72.6 %)	69.3 % (64.7–73.5 %)	73.2 % (68.7–77.2 %)

^aIn calculating these measures, we used ‘positive’ to denote an acceptable outcome (eGOS > 4)