Creating Forest Plots from Pre-computed Data using PROC SGPLOT and Graph Template Language

Historically, forest plots graphically display the information from the individual studies that went into the meta-analysis, and more recently the results of observational studies (e.g. epidemiology). They show the estimates (e.g. odds ratios, hazard ratios, and log transformed hazard ratios) and the amount of variation (e.g. 95% confidence intervals) and in a case of meta-analysis an overall pooled estimate. Forest plots in various forms have been published for more than 20 years, but have gained identity and popularity in the past 15 years. During this time they have been improved but it is still not easy to draw them in most standard statistical packages. In this paper we introduce a custom approach in SAS PROC SGPLOT that creates a forest plot from precomputed data based on the logistic regression results. Further we present a dynamic graph template for a forest plot that can be applied by researchers on their data. Finally we provide a dynamic graph template for meta-analysis.