Lesko et al. Respond to “The Importance of Descriptive Epidemiology”

In his thoughtful response (1) to our paper (2), Platt points out the utility of descriptive analyses beyond describing features of specific target populations. We agree there are steps in all epidemiologic investigations that are descriptive in nature, even if the research question is not descriptive. Platt rightly highlights the importance of descriptive statistics for investigating the feasibility of answering another (usually nondescriptive) research question. We also note the importance of descriptive statistics for data checking, prior to answering another research question. We intended our framework to help guide answering descriptive research questions, rather than generation of descriptive statistics in service of answering causal or predictive questions. However, many of our points (about measurement error and measures of incidence) apply to these descriptive exercises as well. In our experience, students tend to struggle with feasibility and data checks as much as with estimating parameters of interest because our curriculums may not cover data management explicitly and academic exercises often rely on precleaned and processed data. Thus, enhanced training in descriptive epidemiology supports the goals of causal and predictive epidemiology, but we caution that overemphasizing this aspect of descriptive epidemiology might give the impression that descriptive epidemiology is not important in its own right.
Source: American Journal of Epidemiology - Category: Epidemiology Source Type: research