Skip to main content
. 2021 May 29;23:62. doi: 10.1186/s13058-021-01439-2

Table 4.

Discrimination of the CKB 10-year prediction model in the test subcohort of China Kadoorie Biobank (CKB) and Shanghai Women’s Health Study (SWHS)

Test subcohort of CKB SWHS
AUCs (95% CIs) AUCs (95% CIs)
Basic modela
 Overall 0.609 (0.581 to 0.637) 0.505 (0.487 to 0.524)
 Residence-adjusted 0.533 (0.506 to 0.561) --
Full model
 Overallb 0.658 (0.631 to 0.684) 0.573 (0.553 to 0.593)
Age-specific, year
 < 50
  30–34 0.542 (0.346 to 0.738) --
  35–39 0.602 (0.520 to 0.684) --
  40–44 0.661 (0.600 to 0.723) 0.562 (0.522 to 0.601)
  45–49 0.579 (0.502 to 0.657) 0.545 (0.500 to 0.589)
 ≥ 50
  50–54 0.638 (0.578 to 0.698) 0.566 (0.511 to 0.621)
  55–59 0.594 (0.515 to 0.673) 0.611 (0.558 to 0.665)
  60–64 0.682 (0.595 to 0.770) 0.604 (0.545 to 0.662)
  65–69 0.740 (0.636 to 0.844) 0.673 (0.622 to 0.723)
  70–74 0.777 (0.693 to 0.862) --
  75–79 0.960 (0.927 to 0.994) --
Residence-specific
 Urban 0.646 (0.614 to 0.679) --
 Rural 0.615 (0.568 to 0.661) --
Age- and residence-adjustedc 0.634 (0.608 to 0.661) 0.585 (0.564 to 0.605)

Abbreviations: AUC area under the receiver characteristic operating curve, CI confidence interval, -- not applicable

aBasic model included age and residence in the CKB and included age only in the SWHS

bOverall AUC indicated the discriminating ability of the absolute risk predicted by our full model

cAge- and residence-adjusted AUC was estimated by testing the full model while adjusting for residence (urban/rural) and age at entry in a 5-year interval, i.e., the prediction effect of age and residence was removed