Table 2.

Patient characteristics for the multivariate Fine–Gray model: complete data cohort in training and validation datasets.

Variable	Training data (N = 14,781)		Validation data (N = 29,727)
	N	%	N	%
Racea
White	10,853	73	16,812	57
Black	1391	9	1303	4
Hispanic	1767	12	784	3
Other	770	5	785	3
Unknown	0	0	10,043	34
Age
<40	1582	11	1913	6
40–69	11,351	77	22,151	75
≥70	1848	12	5663	19
Menopausal statusb
Premenopausal	5182	35	17,332	58
Postmenopausal	9599	65	12,395	42
Anatomic stage
I	7688	52	16,607	56
IIA	3694	25	8327	28
IIB	1870	13	3331	11
IIIA–IIIB	1076	7	737	3
IIIC	453	3	725	2
Nuclear gradec
1	1885	13	6723	23
2	6627	45	13,884	47
3	6269	42	9120	31
ER statusd
Positive	11,623	79	24,646	83
Negative	3154	21	5077	17
Unknown	4	0	4	0
PR statusd
Positive	9790	66	21,827	74
Negative	4944	33	7856	27
Unknown	47	0	44	0
Hormone receptor (HR) statuse
HR+c	11,888	80	24,989	84
HR−d	2893	20	4738	16
HER2 statusd,f
Positive	1175	8	1086	4
Negative	13,606	92	28,641	96
Biomarker subgroups
TNBC	2431	16	4401	15
HR+/HER2–	11,175	76	24,240	82
HR+/HER2+e	713	5	749	3
HR−/HER2+	462	3	337	1
Chemotherapyf
TNBC	1914	79	3459	79
HR+/HER2−	5035	45	8866	37
HR+/HER2+	657	92	748	100
HR−/HER2+	426	92	335	99

^aFor NCCN data, race background and ethnicity variables were used to obtain this variable.

^bClinically defined based on history; if not recorded, age is considered (≥60: postmenopausal; <60: premenopausal).

^cHistologic grade or nuclear grade (if histologic grade is not available); Training Data, HG (n = 4724); NG (n = 10,057); Validation Data, HG (n = 29,466); NG (n = 12,434).

^dBiomarker definitions in database are reflective of evolution of national guidelines^{28,38,54–56}.

^eHR+:ER+ or PR+; HR−: ER− and PR−.

^fAll HER2+ patients included in the multivariate analysis received adjuvant trastuzumab.