Table 4.

Hazard ratios, 95% CIs, posterior probabilities and Bayes Factors for the fatty acids selected using the BVS method

Fatty acid	HR (95% CI)a	Posterior probabilities	Bayes Factor
Saturated:
c160	1.25 (1.04, 1.50)	1.00	7672.3
c170	0.77 (0.66, 0.89)	1.00	∞
c180	1.17 (1.00, 1.38)	0.75	60.6
c220	1.25 (0.91, 1.71)	1.00	6646.7
c240	0.67 (0.50, 0.90)	1.00	∞
n-3 polyunsaturated:
c225n3	0.88 (0.75, 1.03)	0.74	55.6
n-6 polyunsaturated:
c182n6	0.85 (0.70, 1.02)	0.60	30.2
c202n6	0.79 (0.68, 0.91)	1.00	∞
c203n6	1.44 (1.22, 1.70)	1.00	∞

Hazard ratios are per 1 standard deviation in the fatty acid.

^aConfidence intervals for 3/9 fatty acids in the modal model included 1. The fatty acids were selected for inclusion using the BVS algorithm based on logistic regression, and hazard ratios were estimated from Prentice-weighted Cox regression. As shown in the simulations, the BVS algorithm has higher sensitivity than methods using weighted Cox regression, so it is possible that the CIs around the hazard ratios for some variables selected using BVS will include 1.