Mean Values (and Standard Deviations) of Indicators Considered for Assessing the Performance of Models Used to Estimate Incidence of 2 Types of Cancer With Small Case Counts, RARECARENet Database, 2000–2007a
| Cancer Type and Model Used . | CI/CrI Coverage, % . | CI/CrI Width, no. of cases . | RMSE . |
|---|---|---|---|
| Adenocarcinoma with variants of middle ear | |||
| WinBUGS modelsb | |||
| Model 1 | 96.04 (16.96) | 2.54 (2.54) | 0.43 (0.69) |
| Model 2 | 96.56 (15.45) | 2.63 (2.57) | 0.43 (0.69) |
| Model 3 | 96.22 (13.22) | 2.74 (2.61) | 0.43 (0.69) |
| Model 4 | 96.59 (11.78) | 2.86 (2.67) | 0.44 (0.70) |
| Model 5 | 96.59 (11.78) | 2.86 (2.67) | 0.44 (0.70) |
| Model 6 | 96.37 (12.67) | 3.61 (2.94) | 0.62 (0.84) |
| Model 7 | 96.37 (12.67) | 2.84 (2.65) | 0.44 (0.70) |
| INLA modelsc | |||
| Model 8 | 79.56 (40.34) | 1.78 (2.23) | 0.46 (0.76) |
| Model 9 | 95.59 (18.15) | 2.57 (2.56) | 0.44 (0.70) |
| Empirical Bayes approach | 51.33 (50.00) | 1.23 (2.12) | 0.45 (0.70) |
| Generalized linear mixed model | 76.44 (42.45) | 1.84 (2.33) | 0.45 (0.71) |
| Poisson approximation | |||
| Exact | 96.74 (17.76) | 6.11 (2.87) | 0.91 (0.99) |
| Byar’s | 98.44 (12.38) | 5.05 (2.94) | 0.91 (0.99) |
| Adenocarcinoma with variants of trachea | |||
| WinBUGS models | |||
| Model 1 | 92.44 (26.44) | 6.28 (5.40) | 1.39 (1.85) |
| Model 2 | 92.81 (25.83) | 6.34 (5.40) | 1.38 (1.83) |
| Model 3 | 94.00 (23.76) | 6.42 (5.39) | 1.37 (1.82) |
| Model 4 | 95.48 (20.78) | 6.57 (5.43) | 1.36 (1.82) |
| Model 5 | 95.48 (20.78) | 6.57 (5.43) | 1.36 (1.82) |
| Model 6 | 96.22 (19.07) | 7.60 (5.61) | 1.55 (1.90) |
| Model 7 | 95.30 (23.20) | 6.49 (5.42) | 1.38 (1.83) |
| INLA models | |||
| Model 8 | 87.04 (33.60) | 5.92 (5.37) | 1.44 (1.91) |
| Model 9 | 92.44 (26.44) | 6.36 (5.41) | 1.39 (1.83) |
| Empirical Bayes approach | 84.15 (36.54) | 5.58 (5.18) | 1.39 (1.87) |
| Generalized linear mixed model | 85.70 (35.02) | 6.01 (5.38) | 1.39 (1.84) |
| Poisson approximation | |||
| Exact | 95.19 (21.42) | 9.68 (5.48) | 1.78 (1.94) |
| Byar’s | 97.70 (14.98) | 8.67 (5.51) | 1.78 (1.94) |
| Cancer Type and Model Used . | CI/CrI Coverage, % . | CI/CrI Width, no. of cases . | RMSE . |
|---|---|---|---|
| Adenocarcinoma with variants of middle ear | |||
| WinBUGS modelsb | |||
| Model 1 | 96.04 (16.96) | 2.54 (2.54) | 0.43 (0.69) |
| Model 2 | 96.56 (15.45) | 2.63 (2.57) | 0.43 (0.69) |
| Model 3 | 96.22 (13.22) | 2.74 (2.61) | 0.43 (0.69) |
| Model 4 | 96.59 (11.78) | 2.86 (2.67) | 0.44 (0.70) |
| Model 5 | 96.59 (11.78) | 2.86 (2.67) | 0.44 (0.70) |
| Model 6 | 96.37 (12.67) | 3.61 (2.94) | 0.62 (0.84) |
| Model 7 | 96.37 (12.67) | 2.84 (2.65) | 0.44 (0.70) |
| INLA modelsc | |||
| Model 8 | 79.56 (40.34) | 1.78 (2.23) | 0.46 (0.76) |
| Model 9 | 95.59 (18.15) | 2.57 (2.56) | 0.44 (0.70) |
| Empirical Bayes approach | 51.33 (50.00) | 1.23 (2.12) | 0.45 (0.70) |
| Generalized linear mixed model | 76.44 (42.45) | 1.84 (2.33) | 0.45 (0.71) |
| Poisson approximation | |||
| Exact | 96.74 (17.76) | 6.11 (2.87) | 0.91 (0.99) |
| Byar’s | 98.44 (12.38) | 5.05 (2.94) | 0.91 (0.99) |
| Adenocarcinoma with variants of trachea | |||
| WinBUGS models | |||
| Model 1 | 92.44 (26.44) | 6.28 (5.40) | 1.39 (1.85) |
| Model 2 | 92.81 (25.83) | 6.34 (5.40) | 1.38 (1.83) |
| Model 3 | 94.00 (23.76) | 6.42 (5.39) | 1.37 (1.82) |
| Model 4 | 95.48 (20.78) | 6.57 (5.43) | 1.36 (1.82) |
| Model 5 | 95.48 (20.78) | 6.57 (5.43) | 1.36 (1.82) |
| Model 6 | 96.22 (19.07) | 7.60 (5.61) | 1.55 (1.90) |
| Model 7 | 95.30 (23.20) | 6.49 (5.42) | 1.38 (1.83) |
| INLA models | |||
| Model 8 | 87.04 (33.60) | 5.92 (5.37) | 1.44 (1.91) |
| Model 9 | 92.44 (26.44) | 6.36 (5.41) | 1.39 (1.83) |
| Empirical Bayes approach | 84.15 (36.54) | 5.58 (5.18) | 1.39 (1.87) |
| Generalized linear mixed model | 85.70 (35.02) | 6.01 (5.38) | 1.39 (1.84) |
| Poisson approximation | |||
| Exact | 95.19 (21.42) | 9.68 (5.48) | 1.78 (1.94) |
| Byar’s | 97.70 (14.98) | 8.67 (5.51) | 1.78 (1.94) |
Abbreviations: CI, confidence interval; CrI, credible interval; INLA, integrated nested Laplace approximation; RARECARENet, Information Network on Rare Cancers; RMSE, root mean squared error.
a Coverage and width of the 95% CI or CrI and RMSE between the observed and predicted numbers of cases across the 1,000 simulated data sets for 2 cancer sites with small case counts in all areas.
b WinBUGS models: model 1, |$\sigma\!\!\sim\!\! \mathrm{Uniform}(0,500)$|; model 2, |$\sigma\!\!\sim\!\! \mathrm{Uniform}(0.1,500)$|; model 3, |$\sigma\!\!\sim\!\!\mathrm{Uniform}(0.2,500)$|; model 4, σ |$\sim \mathrm{Uniform}(0.3,500)$|; model 5, σ |$\sim \mathrm{Uniform}(0.5,500)$|; model 6, σ |$\sim \mathrm{Uniform}(1,500)$|; model 7, σ |$\sim \mathrm{Uniform}(a,b)$|.
c INLA models: model 8, |$\tau \sim \gamma (0.00001,0.00001)$|; model 9, σ |$\sim \mathrm{Uniform}(0,\infty)$|.
Mean Values (and Standard Deviations) of Indicators Considered for Assessing the Performance of Models Used to Estimate Incidence of 2 Types of Cancer With Small Case Counts, RARECARENet Database, 2000–2007a
| Cancer Type and Model Used . | CI/CrI Coverage, % . | CI/CrI Width, no. of cases . | RMSE . |
|---|---|---|---|
| Adenocarcinoma with variants of middle ear | |||
| WinBUGS modelsb | |||
| Model 1 | 96.04 (16.96) | 2.54 (2.54) | 0.43 (0.69) |
| Model 2 | 96.56 (15.45) | 2.63 (2.57) | 0.43 (0.69) |
| Model 3 | 96.22 (13.22) | 2.74 (2.61) | 0.43 (0.69) |
| Model 4 | 96.59 (11.78) | 2.86 (2.67) | 0.44 (0.70) |
| Model 5 | 96.59 (11.78) | 2.86 (2.67) | 0.44 (0.70) |
| Model 6 | 96.37 (12.67) | 3.61 (2.94) | 0.62 (0.84) |
| Model 7 | 96.37 (12.67) | 2.84 (2.65) | 0.44 (0.70) |
| INLA modelsc | |||
| Model 8 | 79.56 (40.34) | 1.78 (2.23) | 0.46 (0.76) |
| Model 9 | 95.59 (18.15) | 2.57 (2.56) | 0.44 (0.70) |
| Empirical Bayes approach | 51.33 (50.00) | 1.23 (2.12) | 0.45 (0.70) |
| Generalized linear mixed model | 76.44 (42.45) | 1.84 (2.33) | 0.45 (0.71) |
| Poisson approximation | |||
| Exact | 96.74 (17.76) | 6.11 (2.87) | 0.91 (0.99) |
| Byar’s | 98.44 (12.38) | 5.05 (2.94) | 0.91 (0.99) |
| Adenocarcinoma with variants of trachea | |||
| WinBUGS models | |||
| Model 1 | 92.44 (26.44) | 6.28 (5.40) | 1.39 (1.85) |
| Model 2 | 92.81 (25.83) | 6.34 (5.40) | 1.38 (1.83) |
| Model 3 | 94.00 (23.76) | 6.42 (5.39) | 1.37 (1.82) |
| Model 4 | 95.48 (20.78) | 6.57 (5.43) | 1.36 (1.82) |
| Model 5 | 95.48 (20.78) | 6.57 (5.43) | 1.36 (1.82) |
| Model 6 | 96.22 (19.07) | 7.60 (5.61) | 1.55 (1.90) |
| Model 7 | 95.30 (23.20) | 6.49 (5.42) | 1.38 (1.83) |
| INLA models | |||
| Model 8 | 87.04 (33.60) | 5.92 (5.37) | 1.44 (1.91) |
| Model 9 | 92.44 (26.44) | 6.36 (5.41) | 1.39 (1.83) |
| Empirical Bayes approach | 84.15 (36.54) | 5.58 (5.18) | 1.39 (1.87) |
| Generalized linear mixed model | 85.70 (35.02) | 6.01 (5.38) | 1.39 (1.84) |
| Poisson approximation | |||
| Exact | 95.19 (21.42) | 9.68 (5.48) | 1.78 (1.94) |
| Byar’s | 97.70 (14.98) | 8.67 (5.51) | 1.78 (1.94) |
| Cancer Type and Model Used . | CI/CrI Coverage, % . | CI/CrI Width, no. of cases . | RMSE . |
|---|---|---|---|
| Adenocarcinoma with variants of middle ear | |||
| WinBUGS modelsb | |||
| Model 1 | 96.04 (16.96) | 2.54 (2.54) | 0.43 (0.69) |
| Model 2 | 96.56 (15.45) | 2.63 (2.57) | 0.43 (0.69) |
| Model 3 | 96.22 (13.22) | 2.74 (2.61) | 0.43 (0.69) |
| Model 4 | 96.59 (11.78) | 2.86 (2.67) | 0.44 (0.70) |
| Model 5 | 96.59 (11.78) | 2.86 (2.67) | 0.44 (0.70) |
| Model 6 | 96.37 (12.67) | 3.61 (2.94) | 0.62 (0.84) |
| Model 7 | 96.37 (12.67) | 2.84 (2.65) | 0.44 (0.70) |
| INLA modelsc | |||
| Model 8 | 79.56 (40.34) | 1.78 (2.23) | 0.46 (0.76) |
| Model 9 | 95.59 (18.15) | 2.57 (2.56) | 0.44 (0.70) |
| Empirical Bayes approach | 51.33 (50.00) | 1.23 (2.12) | 0.45 (0.70) |
| Generalized linear mixed model | 76.44 (42.45) | 1.84 (2.33) | 0.45 (0.71) |
| Poisson approximation | |||
| Exact | 96.74 (17.76) | 6.11 (2.87) | 0.91 (0.99) |
| Byar’s | 98.44 (12.38) | 5.05 (2.94) | 0.91 (0.99) |
| Adenocarcinoma with variants of trachea | |||
| WinBUGS models | |||
| Model 1 | 92.44 (26.44) | 6.28 (5.40) | 1.39 (1.85) |
| Model 2 | 92.81 (25.83) | 6.34 (5.40) | 1.38 (1.83) |
| Model 3 | 94.00 (23.76) | 6.42 (5.39) | 1.37 (1.82) |
| Model 4 | 95.48 (20.78) | 6.57 (5.43) | 1.36 (1.82) |
| Model 5 | 95.48 (20.78) | 6.57 (5.43) | 1.36 (1.82) |
| Model 6 | 96.22 (19.07) | 7.60 (5.61) | 1.55 (1.90) |
| Model 7 | 95.30 (23.20) | 6.49 (5.42) | 1.38 (1.83) |
| INLA models | |||
| Model 8 | 87.04 (33.60) | 5.92 (5.37) | 1.44 (1.91) |
| Model 9 | 92.44 (26.44) | 6.36 (5.41) | 1.39 (1.83) |
| Empirical Bayes approach | 84.15 (36.54) | 5.58 (5.18) | 1.39 (1.87) |
| Generalized linear mixed model | 85.70 (35.02) | 6.01 (5.38) | 1.39 (1.84) |
| Poisson approximation | |||
| Exact | 95.19 (21.42) | 9.68 (5.48) | 1.78 (1.94) |
| Byar’s | 97.70 (14.98) | 8.67 (5.51) | 1.78 (1.94) |
Abbreviations: CI, confidence interval; CrI, credible interval; INLA, integrated nested Laplace approximation; RARECARENet, Information Network on Rare Cancers; RMSE, root mean squared error.
a Coverage and width of the 95% CI or CrI and RMSE between the observed and predicted numbers of cases across the 1,000 simulated data sets for 2 cancer sites with small case counts in all areas.
b WinBUGS models: model 1, |$\sigma\!\!\sim\!\! \mathrm{Uniform}(0,500)$|; model 2, |$\sigma\!\!\sim\!\! \mathrm{Uniform}(0.1,500)$|; model 3, |$\sigma\!\!\sim\!\!\mathrm{Uniform}(0.2,500)$|; model 4, σ |$\sim \mathrm{Uniform}(0.3,500)$|; model 5, σ |$\sim \mathrm{Uniform}(0.5,500)$|; model 6, σ |$\sim \mathrm{Uniform}(1,500)$|; model 7, σ |$\sim \mathrm{Uniform}(a,b)$|.
c INLA models: model 8, |$\tau \sim \gamma (0.00001,0.00001)$|; model 9, σ |$\sim \mathrm{Uniform}(0,\infty)$|.
This PDF is available to Subscribers Only
View Article Abstract & Purchase OptionsFor full access to this pdf, sign in to an existing account, or purchase an annual subscription.
