Skip to main content
. 2022 Feb 12;25(4):755–761. doi: 10.1038/s41391-022-00497-7

Table 2.

Multivariable Cox models with both PHS and family history of prostate cancer (defined as ≥1 first-degree relative affected) for association with any prostate cancer, with clinically significant prostate cancer, and with fatal prostate cancer.

Dataset Variable   Any prostate cancer
beta z-score p-value HR
ProtecT (iCOGs), n = 5703 PHS 2.11 72.1 <10−16 8.13
Family history 0.06 1.7 0.1 1.07
African (OncoArray), n = 5557 PHS 1.64 53.8 <1016 5.03
Family history 1.13 46.6 <1016 3.09
Asian (OncoArray); n = 1028 PHS 1.99 68.5 <1016 7.64
Family history 0.45 13.1 <10−16 1.56
COSM (OncoArray), n = 2453 PHS 2.13 71.2 <1016 8.71
Family history 0.53 19.0 <1016 1.70
Dataset Variable   Clinically significant prostate cancer
beta z-score p-value HR
ProtecT (iCOGs), n = 5703 PHS 2.32 49.8 <1016 10.01
Family history −0.01 −0.2 0.82 0.99
African (OncoArray), n = 5557 PHS 1.67 37.4 <1016 5.19
Family history 1.17 33.2 <1016 3.22
Asian (OncoArray), n = 1028 PHS 1.89 50.4 <1016 6.90
Family history 0.13 2.5 0.012 1.14
COSM (OncoArray), n = 2453 PHS 2.13 56.9 <1016 6.90
Family history 0.45 12.6 <1016 1.57
Dataset Variable   Fatal prostate cancer
beta z-score p-value HR
COSM (OncoArray), n = 2453 PHS 1.68 19.8 <1016 5.51
Family history 0.43 5.0 4.7 × 107 1.53

Analyses were limited to participants with known family history. Beta and z-scores refer to the overall association (within the multivariable Cox regression) with the endpoint of interest, within the corresponding testing dataset. The p-values reported are two-tailed from the multivariable Cox models.