Table 2 Finite sample properties using Monte Carlo simulation study
Estimator | Effect | \(\psi\) | Estimate | Bias | R. Bias(%) | MCEa | rMSEb | \(\alpha\)-level |
|---|---|---|---|---|---|---|---|---|
Scenario 1: No irregular visits, no unmeasured confounder | ||||||||
Naïve | \(A_{j-1}\) | -0.3 | -0.274 | 0.0260 | 8.67% | 0.2966 | 0.2978 | 0.939 |
sIPTW | \(A_{j-1}\) | -0.3 | -0.2796 | 0.0204 | 6.8% | 0.3183 | 0.3189 | 0.944 |
cIPTW | \(A_{j-1}\) | -0.3 | -0.2842 | 0.0158 | 5.27% | 0.3071 | 0.3075 | 0.943 |
Scenario 2: No irregular visits, unmeasured confounder | ||||||||
Naïve | \(A_{j-1}\) | -0.3 | -0.2642 | 0.0358 | 11.93% | 0.3484 | 0.3502 | 0.936 |
sIPTW | \(A_{j-1}\) | -0.3 | -0.2839 | 0.0161 | 5.37% | 0.3181 | 0.3185 | 0.928 |
cIPTW | \(A_{j-1}\) | -0.3 | -0.2909 | 0.0091 | 3.03% | 0.3138 | 0.3139 | 0.940 |
Scenario 3: Irregular visits, no unmeasured confounder | ||||||||
Naïve | \(A_{j-1}\) | -0.3 | -0.2678 | 0.0322 | 10.73% | 0.3421 | 0.3436 | 0.927 |
sIPTW | \(A_{j-1}\) | -0.3 | -0.2719 | 0.0281 | 9.37% | 0.3332 | 0.3343 | 0.934 |
sIPVW | \(A_{j-1}\) | -0.3 | -0.2776 | 0.0224 | 7.47% | 0.3142 | 0.3150 | 0.940 |
sIPTW\(\times\)sIPVW | \(A_{j-1}\) | -0.3 | -0.2859 | 0.0141 | 4.7% | 0.3292 | 0.3295 | 0.935 |
cIPTW\(\times\)cIPVW | \(A_{j-1}\) | -0.3 | -0.3029 | -0.0029 | -0.97% | 0.3341 | 0.3341 | 0.944 |
Scenario 4: Irregular visits, unmeasured confounder | ||||||||
Naïve | \(A_{j-1}\) | -0.3 | -0.2612 | 0.0388 | 12.93% | 0.3397 | 0.3419 | 0.921 |
sIPTW | \(A_{j-1}\) | -0.3 | -0.2672 | 0.0328 | 10.93% | 0.3317 | 0.3333 | 0.938 |
sIPVW | \(A_{j-1}\) | -0.3 | -0.2692 | 0.0308 | 10.27% | 0.3162 | 0.3177 | 0.932 |
sIPTW\(\times\)sIPVW | \(A_{j-1}\) | -0.3 | -0.2777 | 0.0223 | 7.43% | 0.3294 | 0.3301 | 0.940 |
cIPTW\(\times\)cIPVW | \(A_{j-1}\) | -0.3 | -0.2932 | 0.0068 | 2.27% | 0.2934 | 0.2934 | 0.947 |