Skip to main content

Table 4 Scenarios 5 & 6: 30% 'good' prognosis, ×1.2 'good' prognosis survival, 50% and 75% switching probabilities, 100% treatment effect for switchers, uniform distribution for switching times

From: Assessing methods for dealing with treatment switching in randomised controlled trials: a simulation study

True HR and eψ Method Mean estimate Mean SE SE of mean 95% Confidence interval Bias MSE Coverage (%) Successful estimation (%)
      Lower Upper     
  Hazard ratio methods          
  ITT 0.9364 0.0909 0.0909 0.7741 1.1328 0.0364 0.0096 93.6 100.0
  PP - Exclude switchers 0.9204 0.1275 0.1278 0.7016 1.2074 0.0204 0.0167 96.0 100.0
  PP - Censor at switch 2.1663 0.3018 0.2975 1.6488 2.8466 1.2663 1.6921 0.0 100.0
  Time-dependent covariate 3.0639 0.4065 0.4054 2.3625 3.9740 2.1639 4.8466 0.0 100.0
  Law and Kaldor 0.9396 0.1168 0.1144 0.7364 1.1987 0.0396 0.0147 94.2 100.0
  Loeys and Goethebeur 0.8435 - 0.2725 0.4578 1.8486 -0.0565 0.0774 92.8 99.9
0.9 & 1.23 AFT methods          
  ITT 1.1732 0.2280 0.2311 0.8018 1.7173 -0.0613 0.0572 93.2 100.0
  PP - Exclude switchers 1.2492 0.1275 0.3479 0.7276 2.1475 0.0146 0.1212 95.9 100.0
  PP - Censor at switch 0.2373 0.0665 0.0663 0.1371 0.4115 -0.9973 0.9989 0.0 100.0
  Robins and Tsiatis - Logrank 1.2882 - 0.3862 0.6906 2.2552 0.0536 0.1521 93.8 100.0
  Robins and Tsiatis - Cox 1.2855 - 0.3874 0.6868 2.2531 0.0509 0.1527 93.9 96.3
  Robins and Tsiatis - Exponential 1.3036 - 0.3693 0.9172 1.9617 0.0691 0.1412 83.6 100.0
  Robins and Tsiatis - Weibull 1.2879 - 0.3857 0.6988 2.2550 0.0534 0.1516 94.5 100.0
  Branson and Whitehead 1.2841 0.2497 0.3619 0.8772 1.8802 0.0495 0.1334 83.7 100.0
  Walker et al 2.1037 0.7802 0.7791 1.0276 4.4146 0.8692 1.3624 70.0 99.8
  Hazard ratio methods          
  ITT 0.8073 0.0812 0.0814 0.6629 0.9833 0.1073 0.0181 71.5 100.0
  PP - Exclude switchers 0.7179 0.1009 0.1019 0.5451 0.9457 0.0179 0.0107 94.8 100.0
  PP - Censor at switch 1.6825 0.2388 0.2454 1.2740 2.2223 0.9825 1.0256 0.0 100.0
  Time-dependent covariate 2.4211 0.3257 0.3396 1.8602 3.1516 1.7211 3.0776 0.0 100.0
  Law and Kaldor 0.8110 0.1064 0.1046 0.6271 1.0488 0.1110 0.0233 81.2 100.0
  Loeys and Goethebeur 0.5248 - 0.1728 0.2544 1.0233 -0.1752 0.0606 83.8 99.0
0.7 & 2.04 AFT methods          
  ITT 1.5845 0.3208 0.3294 1.0657 2.3569 -0.4563 0.3167 72.2 100.0
  PP - Exclude switchers 2.0610 0.1009 0.6062 1.1886 3.5793 0.0202 0.3679 94.4 100.0
  PP - Censor at switch 0.3894 0.1088 0.1126 0.2254 0.6742 -1.6514 2.7398 0.0 100.0
  Robins and Tsiatis - Logrank 2.1120 - 0.6624 1.1204 3.8159 0.0712 0.4439 94.8 100.0
  Robins and Tsiatis - Cox 2.1062 - 0.6612 1.1095 3.8198 0.0654 0.4415 95.1 92.6
  Robins and Tsiatis - Exponential 2.1121 - 0.6636 1.3323 3.3923 0.0713 0.4455 87.4 100.0
  Robins and Tsiatis - Weibull 2.1128 - 0.6631 1.1249 3.8301 0.0719 0.4449 94.9 100.0
  Branson and Whitehead 2.0536 0.4177 0.6096 1.3787 3.0601 0.0127 0.3718 83.0 100.0
  Walker et al 3.6581 1.3293 1.2498 1.8098 7.5537 1.6172 4.1774 63.9 98.0
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us