Table 1.
Prior distributions, posterior estimates, and posterior predictive distributions of overdiagnosis for model parameters in base case analysis model
| Prior distribution* | Mean (95% CI) estimated values† | |||
|---|---|---|---|---|
| Type | Mean | |||
| Parameter | ||||
| Lifetime probability of breast cancer for women born in 1900 | Uniform (6.0%, 12.0%) | 9.0% | 7.0% (6.0% to 8.0%) | |
| Increase in lifetime probability for women born 1900-50 | Uniform (0.5, 2.0) | 1.25 | 1.06 (0.50 to 1.69) | |
| Proportion of increase in lifetime probability from carcinomas in situ | Uniform (0.0%, 100.0%) | 50.0% | 41.0% (5.0% to 85.0%) | |
| Age at onset of preclinical cancer: parameter 1 for scaled β distribution | Uniform (7.0, 8.0) | 7.50 | 7.48 (6.93 to 7.99) | |
| Age at onset of preclinical cancer: parameter 2 for scaled β distribution | Uniform (5.5, 7.0) | 6.25 | 5.94 (5.79 to 6.12) | |
| Proportion of invasive cancer at onset of preclinical phase | β (3, 7) | 30.0% | 52.0% (23.0% to 74.0%) | |
| Proportion of in situ cancer evolving to invasive during preclinical phase | β (8, 2) | 80.0% | 91.0% (84.0% to 97.0%) | |
| Proportion of non-progressive carcinoma in situ | Uniform (0.0%, 50.0%) | 25.0% | 6.0% (0.0% to 17.0%) | |
| Proportion of non-progressive invasive cancer | Uniform (0.0%, 50.0%) | 25.0% | 3.0% (3.0% to 4.0%) | |
| Probability of undergoing screening mammography in 1987-8 and 1989-90 | Uniform (15.0%, 25.0%) | 20.0% | 20.0% (19.0% to 21.0%) | |
| Probability of undergoing screening mammography in 1991-2 | Uniform (19.0%, 30.0%) | 25.0% | 21.0% (18.0% to 24.0%) | |
| Probability of undergoing screening mammography in 2005-6 | Uniform (46.0%, 74.0%) | 60.0% | 57.0% (46.0% to 71.0%) | |
| Overdiagnosis for in situ cancers | ||||
| Among cancer diagnosed | — | — | 28.0% (2.2% to 59.8%) | |
| Among cancer screened | — | — | 31.9% (2.9% to 62.3%) | |
| Overdiagnosis for invasive cancers | ||||
| Among cancer diagnosed | — | — | 1.5% (0.3% to 2.9%) | |
| Among cancer screened | — | — | 3.3% (0.7% to 6.5%) | |
CI=credibility interval.
*Proportion of overdiagnosis for carcinoma in situ and invasive cancer was not parameter of model and consequently did not have prior distributions. Values derived from proportion of non-progressive disease for each type of cancer.
†Obtained from posterior distributions for 12 parameters and from posterior predictive distributions for proportion of overdiagnosis.