Table 4 Scenario (II): performance of the estimated propensity score (PS) model by simulation setting assuming n = 200
From: Treatment effect estimation using the propensity score in clinical trials with historical control
Performance measurement | PS model | Outcome event rate | Odds ratio | |||
|---|---|---|---|---|---|---|
1.0 | 2.0 | 5.0 | 10.0 | |||
Bias | \(\pi\) (without \(X_\textit{r}\)) | 50% | 0.168 | 0.164 | 0.168 | 0.224 |
10% | 0.032 | 0.132 | 0.177 | 0.268 | ||
5% | -0.908 | -0.106 | 0.516 | 1.758 | ||
\({\pi }^{*}\) (with \({X}_{{\text{r}}}\)) | 50% | 0.050 | 0.050 | 0.060 | 0.126 | |
10% | -0.096 | 0.011 | 0.077 | 0.187 | ||
5% | -0.989 | -0.203 | 0.447 | 1.750 | ||
MSE | \(\pi\) | 50% | 0.144 | 0.151 | 0.202 | 0.862 |
10% | 1.686 | 0.459 | 0.326 | 1.431 | ||
5% | 19.574 | 6.274 | 7.605 | 27.879 | ||
\({\pi }^{*}\) | 50% | 0.174 | 0.193 | 0.261 | 0.880 | |
10% | 1.640 | 0.532 | 0.421 | 1.584 | ||
5% | 18.505 | 6.120 | 8.210 | 29.796 | ||
Coverage (%) | \(\pi\) | 50% | 92.4 | 94.0 | 95.2 | 95.8 |
10% | 93.9 | 94.2 | 95.1 | 96.6 | ||
5% | 89.8 | 94.6 | 95.5 | 88.4 | ||
\({\pi }^{*}\) | 50% | 94.5 | 94.0 | 93.6 | 94.0 | |
10% | 94.0 | 94.0 | 93.9 | 93.7 | ||
5% | 90.0 | 93.4 | 93.4 | 86.8 | ||
Type I error and power (%) | \(\pi\) | 50% | 7.5 | 68.6 | 99.9 | 100.0 |
10% | 6.0 | 39.1 | 93.1 | 99.5 | ||
5% | 10.2 | 25.2 | 72.3 | 92.3 | ||
\({\pi }^{*}\) | 50% | 5.4 | 44.4 | 94.1 | 98.7 | |
10% | 5.9 | 26.6 | 81.1 | 96.2 | ||
5% | 9.9 | 20.7 | 98.8 | 99.9 | ||