How to perform a meta-analysis on incidence rates? I have a number of studies with incidence rates of a disease in the population and I want to obtain a weighted average, also taking into account moderator variables; 
specifically (i) study design (ii) population size from which the incidence was drawn, and (iii) the number of years of data collection.
As a novice how do I go about doing that? 
Everything I read seems to be aimed at interventions and effect sizes. 
 A: There are different approaches available, but the basic underpinning of course is cumulating the data from separate studies.
First and foremost it depends all on whether you want to pool unadjusted or adjusted effect estimates.
If you want adjusted ones, then you probably need to pool log-transformed adjusted hazard ratios (each weighted inversely to its variance). This can be done in most packages quite easily (even with RevMan).
If you are happy with unadjusted risk estimates, then several methods and models are available in a number of statistical softwares, but I would recommend meta and metafor packages in R. Note that the actual incidence of events will be important (with rare events a Poisson model is likely going to be better), and whether you have 0 cells, requiring some kind of transformartion. In addition, remember that you will have to pool the ratio of events / patients at risk * period at risk (for instance 4 events out of 8 patients each followed for 2 years will yield 4 events out of 16 patients/years).
