Table 2.
Estimation method and treatment | Prostate cancer–specific death |
All-cause death |
||
% died (95% CI) | Survival, mo (95% CI) | % died (95% CI) | Survival, mo (95% CI) | |
All patients (unweighted) | 0.028 | 73.2 | 0.200 | 83.0 |
Observational (unweighted) | ||||
RP | 0.025 (0.022 to 0.028) | 75.0 (74.46 to 75.62) | 0.177 (0.170 to 0.185) | 85.1 (84.49 to 85.70) |
CM | 0.036 (0.030 to 0.041) | 69.2 (68.38 to 70.05) | 0.249 (0.237 to 0.263) | 78.4 (77.57 to 79.33) |
P for difference in means | <.001 | <.001 | <.001 | <.001 |
PS reweighted (IPTW) | ||||
RP | 0.026 (0.023 to 0.030) | 75.2 (74.6 to 75.7) | 0.185 (0.177 to 0.193) | 85.1 (84.5 to 85.7) |
CM | 0.035 (0.029 to 0.040) | 68.4 (67.6 to 69.2) | 0.236 (0.223 to 0.248) | 77.8 (76.9 to 78.7) |
P for difference in means | <.001 | <.001 | <.001 | <.001 |
PS reweighted (SMRW) | ||||
RP | 0.030 (0.026 to 0.033) | 75.4 (74.8 to 76.0) | 0.203 (0.195 to 0.211) | 85.2 (84.6 to 85.8) |
CM | 0.036 (0.030 to 0.041) | 69.2 (68.4 to 70.0) | 0.250 (0.237 to 0.263) | 78.4 (77.6 to 79.3) |
P for difference in means | <.001 | <.001 | <.001 | <.001 |
Instrumental variable† | ||||
Less than median value of instrument | 0.027 (0.023 to 0.031) | 73.7 (73.05 to 74.41) | 0.192 (0.183 to 0.201) | 83.6 (82.94 to 84.35) |
Median value of instrument or higher | 0.030 (0.026 to 0.034) | 72.7 (72.05 to 73.39) | 0.208 (0.199 to 0.218) | 82.4 (81.71 to 83.12) |
P for difference in means | .351 | .038 | .012 | .016 |
RCT (12 y of follow-up)‡ | ||||
RP | 0.131 (0.088 to 0.195) | NR | 0.420 (0.350 to 0.505) | NR |
CM | 0.132 (0.089 to 0.196) | NR | 0.393 (0.325 to 0.477) | NR |
P for difference in means | NR | NR |
A two-sided t test was used. CM = conservative management; IPTW = inverse probability of treatment weights; NR = not reported; PS = propensity score; RCT = randomized controlled trial; RP = radical prostatectomy; SMRW = standardized mortality ratio weights.
Lagged area treatment was the only instrument.
Bill-Axelson et al. (29).