Table 3B. Variable Impact on the Apparent C-Statistic for the Colorectal Cancer Model in Women

Variable	C-Statistic	Change in C-Statistic	Variables (n)
No Model	0.5	—	0
Age	0.657655	0.157655	1
Race/ethnicity	0.667666	0.010010	2
Years of education	0.670909	0.003243	3
Use of estrogen	0.674048	0.003139	4
History of diabetes	0.676648	0.002600	5
Pack-years of smoking	0.679123	0.002476	6
Family history of colon cancer	0.680963	0.001839	7
Regular use of multivitamins	0.682497	0.001534	8
Body mass index	0.683796	0.001300	9
Regular use of NSAIDs	0.684893	0.001097	10
Alcoholic drinks per day	0.685869	0.000976	11
Preference for well-done meat^*	0.686268	0.000399	12
Moderate physical activity per day (hours)^*	0.686525	0.000257	13
Regular use of aspirin^*	0.686790	0.000266	14
Red meat intake per day (oz)^*	0.686869	0.000078	15
History of cancer^*	0.686865	-0.000003	16 (Full)

The outcome was colorectal cancer.
↵* The variable was excluded from the final model because of the relatively small amount of additional prediction accuracy that it could have contributed to the model. The final model, which included 11 variables, had an apparent c-statistic of 0.6859, which was within 0.001 of the accuracy associated with the c-statistic of the full model, which was 0.6869.
NSAIDS, nonsteroidal anti-inflammatory drugs.