From: Meta-analysis of incidence rate data in the presence of zero events
Assigned values | |
---|---|
Methods evaluated | |
Inverse-variance method | |
Inverse-variance method with continuity correction | |
Poisson regression model with fixed intervention effects | |
Poisson regression model with random intervention effects | |
Bayesian Poisson regression model with random intervention effects using inverse-gamma priors for Ï„2 | |
Bayesian Poisson regression model with random intervention effects using half-Cauchy priors for Ï„ | |
Fixed parameters | |
Incidence rate ratio, exp(βint) | 0.2 |
Time, intervention group, t1 | t1∼Uniform(2,5) |
Time, control group, t0 | t0∼Uniform(2,10) |
Number of simulated datasets per scenario, B s | 500 |
Varied parameters | |
Percent of zero counts in the intervention group, Poisson(β1∗0.2) | 0.09%, 5%, 14%, 37%, 55%, 82% |
Heterogeneity: control and intervention groups, σ,τ | |
Scenario A | (0.1,0.5) |
Scenario B | (0.1,2.5) |
Scenario C | (1.0,0.5) |
Scenario D | (1.0,2.5) |
Scenario E | (0.1×β1,2.5) |
Number of studies, k | 5, 10, 20 |