Table 3.
Twelve-year prostate cancer mortality (95% confidence interval) for observed and modeled Prostate Cancer Intervention Versus Observation Trial (PIVOT)*
| Subgroup (version) |
Watchful waiting | Radical prostatectomy | Absolute mortality difference | |||
|---|---|---|---|---|---|---|
| PIVOT | Model | PIVOT | Model | PIVOT | Model | |
| Not screen detected (baseline) |
― | 16.7% (16.6% to 16.8%) |
― | 12.1% (12.0% to 12.1%) |
― | 4.6% (4.4% to 4.8%) |
| Not overdiagnosed (baseline) |
― | 6.0% (3.9% to 8.1%) |
― | 3.1% (1.1% to 5.1%) |
― | 2.9% (0.7% to 5.1%) |
| Overall (baseline) |
7.4% | 7.9% (5.2% to 10.6%) |
4.4% | 5.9% (3.5% to 8.3%) |
3.0% | 2.0% (−1.6% to 5.6%) |
| Overall (sensitivity analysis 1) |
7.4% | 7.8% (5.1% to 10.6%) |
4.4% | 6.0% (3.5% to 8.5%) |
3.0% | 1.8% (−1.7% to 5.5%) |
| Overall (sensitivity analysis 2) |
7.4% | 7.3% (5.2% to 9.4%) |
4.4% | 5.6% (3.4% to 7.8%) |
3.0% | 1.7% (−1.7% to 5.1%) |
* Model results are averages over 500 000 simulations. Ninety-five percent confidence intervals are based on the uncertainty ranges in overdiagnosis and mean lead times shown in Table 2. Baseline results assume exponential lead time distributions and cause-specific survival. Sensitivity analysis 1 replaces exponential with Weibull lead time distributions (see Supplementary Material, available online). Sensitivity analysis 2 replaces exponential with shifted exponential cause-specific survival (see Supplementary Material, available online). — = not available from published data.