Introduction to EPITools

Introduction

Many practical problems in the analysis of public health data require programming or special software, and investigators in different locations may duplicate programming efforts done elsewhere. Often, simple analyses, such as the construction of confidence intervals, are not done and complicate appropriate statistical inferences for small geographic areas. There are many examples of simple and useful numerical tools that would enhance the work of epidemiologists at local health departments and yet are not readily available for the problem in front of them (e.g., exact confidence intervals for low incidence rates, statistical pooling methods for meta-analysis, lifetime risk calculations, etc.). The availability of these tools will encourage wider use of appropriate methods and promote evidence-based public health practices. Therefore, the objectives of EPI Tools are the following:

• An interactive website with useful tools for the analysis of epidemiologic data, and which facilitates the sharing of problem-solving tools. The target audience for this website is local health department epidemiologists and graduate students in epidemiology.
• An expandable website so that epidemiology tools can be contributed from users around the world who are developing tools to solve their unique local problems. This will create both a supply and demand for numerical tools employed on this website, and also the opportunity to learn from the experiences of public health colleagues at distant sites.
• An educational website that helps to train epidemiology graduate students in the methods of community health epidemiology and population health statistics, and that promotes evidence-based public health practices.

Background

The World Wide Web provides the ability to deliver documents and provide interactive content to anywhere in the world, and to any computer platform with Internet access and browser software. In analyzing public health data, the need to perform programming tasks is often a barrier to complete data analysis. While much software has been written, programs may be excessively costly given the intended use, or may even be unavailable on the user's platform. Programming skills are often required, for instance, even for simple tasks such as computing confidence intervals for count data using standard formulas. Finally, investigators in different locations will frequently duplicate the same effort; data analysts in local health departments have no way to easily share such programs. The EPI Tools Website will address these problems, by making simple interactive tools available through the World Wide Web for the most common of such data-analytic tasks---based on the needs of local health departments and graduate students in public health, and will develop an infrastructure for individuals to contribute tools and programs they have developed.

EPI Tools consists three components. First, Rweb is a web-available version of R which is a free program very similar to S or S-Plus. In fact, any of these analyses can be conducted in either R, S, or S-Plus. R is powerful and versatile program for exploring data and conducting epidemiologic and statistical analyses. R is also a powerful programming language for creating functions to carry out tailored analyses. Second, we will compile a library of R functions that were created, developed, and tested for real world problems facing practicing epidemiologists. As EPI Tools users become comfortable writing their own R functions, they can contribute their functions to the EPI Tools library. Third, user-friendly interface modules will be created for the most popular epidemiologic tools that might not be readily accessible to novice users who are unfamiliar with R.