Assessing the accuracy of multiparametric MRI to predict clinically significant prostate cancer in biopsy naïve men across racial/ethnic groups

Table 4 Comparison of all models’ area under the curve by race/ethnic group

Model	Black AUC [95% CI]	White AUC [95% CI]	p value^a	Hispanic AUC [95% CI]	p value^a
Clinical variables	0.60 [0.50, 0.71]	0.55 [0.43, 0.66]	0.49	0.59 [0.45, 0.73]	0.64
PIRADS	0.70 [0.60, 0.80]	0.72 [0.63, 0.82]	0.76	0.58 [0.44, 0.73]	0.12
Clinical variables + PIRADS	0.75 [0.66, 0.84]	0.73 [0.64, 0.83]	0.79	0.59 [0.44, 0.74]	0.11

AUC area under the receiver operating characteristic curve, PIRADS prostate imaging reporting and data system
^ap values correspond to comparison of AUCs across race/ethnic group using White patients as the referent group; p values were compared using Kolmogorov Smirnov Test; PIRADS was modeled as an ordinal variable with 4 values (i.e., 1–2, 3, 4, 5). The p values comparing the AUCs of the models within each race/ethnic group are all > 0.05

ISSN: 1471-2490