Skip to main content
. 2017 May 3;25(7):854–862. doi: 10.1038/ejhg.2017.78

Table 2. Simulation results for case-cohort studies with sampling fraction of 10% from 40,000 individuals for a SNP with RAF=0.10.

  Cox Regression (HR) Logistic Regression (OR)  
True HR Mean(Bias) Mean(SE) Power Mean(Bias) Mean(SE) Power Diff in Power
Complete Follow-up, 5% Cumulative Disease Incidence
 1.1 –0.0007 0.0635 0.3284 0.0021 0.0647 0.3318 –0.0034
 1.2 –0.0002 0.0622 0.8308 0.0053 0.0634 0.8392 –0.0084
 1.5 0.0010 0.0596 1.0000 0.0143 0.0608 1.0000 0.0000
 2 0.0007 0.0582 1.0000 0.0259 0.0588 1.0000 0.0000
Complete Follow-up, 10% Cumulative Disease Incidence
 1.1 0.0001 0.0524 0.4480 0.0054 0.0542 0.4580 –0.0100
 1.2 0.0006 0.0517 0.9430 0.0119 0.0536 0.9522 –0.0092
 1.5 –0.0001 0.0505 1.0000 0.0270 0.0524 1.0000 0.0000
 2 0.0004 0.0503 1.0000 0.0531 0.0518 1.0000 0.0000
Complete Follow-up, 15% Cumulative Disease Incidence
 1.1 0.0011 0.0484 0.5126 0.0098 0.0510 0.5402 –0.0276
 1.2 –0.0008 0.0481 0.9674 0.0161 0.0507 0.9782 –0.0108
 1.5 0.0003 0.0475 1.0000 0.0418 0.0504 1.0000 0.0000
 2 0.0022 0.0479 1.0000 0.0816 0.0507 1.0000 0.0000
               
Survey Follow-up, 5% Cumulative Disease Incidence
 1.1 0.0002 0.0659 0.3120 0.0027 0.0643 0.3378 –0.0258
 1.2 0.0008 0.0647 0.8050 0.0049 0.0631 0.8402 –0.0352
 1.5 –0.0002 0.0624 1.0000 0.0094 0.0607 1.0000 0.0000
 2 0.0015 0.0612 1.0000 0.0195 0.0590 1.0000 0.0000
Survey Follow-up, 10% Cumulative Disease Incidence
 1.1 –0.0004 0.0569 0.3786 0.0026 0.0541 0.4350 –0.0564
 1.2 0.0009 0.0563 0.9038 0.0076 0.0536 0.9422 –0.0384
 1.5 0.0007 0.0553 1.0000 0.0171 0.0525 1.0000 0.0000
 2 0.0024 0.0551 1.0000 0.0334 0.0519 1.0000 0.0000
Survey Follow-up, 15% Cumulative Disease Incidence
 1.1 0.0019 0.0542 0.4332 0.0064 0.0510 0.5130 –0.0798
 1.2 –0.0005 0.0538 0.9294 0.0089 0.0507 0.9682 –0.0388
 1.5 0.0016 0.0533 1.0000 0.0230 0.0504 1.0000 0.0000
 2 0.0008 0.0535 1.0000 0.0424 0.0506 1.0000 0.0000
               
Random Follow-up, 5% Cumulative Disease Incidence
 1.1 –0.0007 0.0681 0.2864 –0.0001 0.0645 0.3212 –0.0348
 1.2 0.0005 0.0670 0.7768 0.0030 0.0633 0.8334 –0.0566
 1.5 –0.0005 0.0648 1.0000 0.0067 0.0609 1.0000 0.0000
 2 0.0007 0.0639 1.0000 0.0143 0.0593 1.0000 0.0000
Random Follow-up, 10% Cumulative Disease Incidence
 1.1 0.0006 0.0599 0.3536 0.0021 0.0543 0.4290 –0.0754
 1.2 –0.0005 0.0593 0.8756 0.0037 0.0537 0.9398 –0.0642
 1.5 0.0004 0.0583 1.0000 0.0097 0.0526 1.0000 0.0000
 2 0.0004 0.0583 1.0000 0.0192 0.0520 1.0000 0.0000
Random Follow-up, 15% Cumulative Disease Incidence
 1.1 –0.0002 0.0574 0.3740 0.0016 0.0509 0.4726 –0.0986
 1.2 0.0021 0.0571 0.9062 0.0055 0.0507 0.9578 –0.0516
 1.5 0.0004 0.0566 1.0000 0.0112 0.0503 1.0000 0.0000
 2 0.0031 0.0571 1.0000 0.0199 0.0505 1.0000 0.0000

Abbreviations: Diff in Power, difference in power (Cox − logistic); HR, hazard ratio; OR, odds ratio; RAF, risk allele frequency; SE, standard error of the logarithm of hazard or odds ratio. Bias refers to either the estimated log(HR) or log(OR) minus the underlying log(HR). It is important to note that HRs and ORs are different measures of comparison and as such ORs are not 'biased' if different from the underlying HR. The Cox model was Prentice weighted and robust SEs were applied to account for the sampling process. This table is a subset of Supplementary Table S5.