Goodmorning everyone. My training is mainly clinical, and I would be grateful that I can be helped. I assessed in a multilevel survival model the contribution of various clusters to the effectiveness of a preventive medicine intervention pertinent to the interception of new diagnoses of atrial fibrillation based on a screening program. Each cluster is identified by a primary care center and is categorized by the number of nurses working in the center. The “zero cluster” identifies a primary care center without nurses. My intention is to compare the contribution made to the diagnosis of atrial fibrillation by professional organizations provided by nurses with the cluster zero. The outcome of the analysis is the hazard of atrial fibrillation in the follow-up considered by the study. The multivariate regression model used is a multi-level parametric survival model that attributes an exponential distribution to basal hazard. I used the mestreg command for that analysis. I quantified the random component of the model relevant to each cluster (the estimated empirical Bayes residuals with their standard errors) using predict. The exponentialized difference between the point estimate of the residual pertaining to a cluster other than cluster zero with that relevant to cluster 0 should produce a cluster-specific Hazard Ratio. At this point I would also like to calculate the confidence intervals relevant to these Hazard Ratios and this is why I turn to your attention. Thanks for what you can suggest
0 Response to About Random Effects in a Multilevel Survival Model
Post a Comment