Table 4.

Model performance statistics when linking multiple datasets.

Cancer	Performance statistics	Model A1	Model A2	Model A3
Breast	AIC	34,409.25	34,400.09	34,381.607
	c-Statistic	0.800	0.801	0.801
	IDI	–	0.002 (0.000 to 0.003)	0.005 (0.001 to 0.013)
	Relative IDI	–	0.006 (0.002 to 0.010)	0.019 (0.003 to 0.048)
	Continuous NRI	–	0.10 (0.05 to 0.14)	0.12 (0.06 to 0.20)
	Event NRI	–	0.02 (−0.03 to 0.05)	0.01 (−0.03 to 0.05)
	Non-event NRI	–	0.09 (0.07 to 0.10)	0.11 (0.07 to 0.18)
Lung	AIC	277,780.96	277,708.46	277,701.67
	c-Statistic	0.734	0.734	0.734
	IDI	–	0.001 (0.000 to 0.002)	0.002 (0.001 to 0.004)
	Relative IDI	–	0.004 (0.002 to 0.006)	0.006 (0.002 to 0.013)
	Continuous NRI	–	0.09 (0.05 to 0.14)	0.16 (0.06 to 0.23)
	Event NRI	–	0.03 (0.02 to 0.05)	0.11 (0.02 to 0.13)
	Non-event NRI	–	0.06 (0.02 to 0.11)	0.05 (0.02 to 0.11)
Colorectal	AIC	51,526.92	51,475.53	51,448.46
	c-Statistic	0.782	0.784	0.785
	IDI	–	0.003 (0.002 to 0.005)	0.009 (0.002 to 0.013)
	Relative IDI	–	0.009 (0.005 to 0.013)	0.024 (0.007 to 0.035)
	Continuous NRI	–	0.09 (0.04 to 0.16)	0.25 (0.05 to 0.31)
	Event NRI	–	−0.02 (−0.05 to 0.03)	0.10 (−0.04 to 0.14)
	Non-event NRI	–	0.11 (0.08 to 0.14)	0.15 (0.09 to 0.18)

AIC: Akaike information criterion; IDI: integrated discrimination improvement; NRI: net reclassification improvement. Sex was not included as a predictor in the breast models. Model A1: Basic dataset with individual-level predictors only. Model A2: Integrated dataset with individual- and contextual-level predictors. Model A3: Integrated dataset with individual- and contextual-level predictors, and cross-level interactions.

Note: Values in parentheses are lower and upper bounds of 95% confidence intervals.