Table 4 Parameters varied through probabilistic sensitivity analysis
Parameter | Simulation | Source |
|---|---|---|
Clinical effectiveness | The log odds of response were simulated from a Normal distribution with mean and standard deviations derived from the CIs of the network meta-analysis (transformed to a log odds scale). | Indirect analysis results. |
Association between mortality and HAQ-DI score | The relative risk was simulated from a Lognormal distribution with parameters implied by the point estimate (1.330) and its confidence interval (1.099 to 1.610) | Wolfe et al. [35] |
Age (years) | Normal distribution defined by the mean (52.165) and the standard error (51.893 to 52.4378) | RAPID 1, RAPID 2 and FAST4WARD |
Gender | Beta distribution defined by N (1821) and n(1506) | RAPID 1, RAPID 2 and FAST4WARD |
Weight | A cumulative distribution function derived from CZP-related data. | RAPID 1, RAPID 2 and FAST4WARD |
Baseline HAQ-DI score | Normal distribution defined by the mean (1.624) and its confidence interval (1.610 to 1.638) | RAPID 1, RAPID 2 and FAST4WARD |
Number of previous DMARD | Normal distribution defined by the mean (2.258) and its confidence interval (2.207 and 2.308) | RAPID 1, RAPID 2 and FAST4WARD |
Disease duration | Normal distribution defined by the mean (6.557) and its confidence interval (6.351 and 6.763) | RAPID 1, RAPID 2 and FAST4WARD |
Anti-CCP antibody positive | Normal distribution defined by the mean (1.676) and its confidence interval (1.611 and 1.741) | RAPID 1, RAPID 2 and FAST4WARD |
Anti-CCP antibody negative | Normal distribution defined by the mean (1.621) and its confidence interval (1.606 and 1.635) | RAPID 1, RAPID 2 and FAST4WARD |
Utility weight | Sampled from a randomized percentage of population (mean 0.380 and confidence interval 0.372 and 0.388) | RAPID 1 and RAPID 2 |