Modelers' Perception of Mathematical Modeling in Epidemiology: A Web-Based Survey

Gilles Hejblum,Sophie Lesieur,Alain-Jacques Valleron,Laura Temime,Michel Setbon

doi:10.1371/journal.pone.0016531

Abstract

BackgroundMathematical modeling in epidemiology (MME) is being used increasingly. However, there are many uncertainties in terms of definitions, uses and quality features of MME.Methodology/Principal FindingsTo delineate the current status of these models, a 10-item questionnaire on MME was devised. Proposed via an anonymous internet-based survey, the questionnaire was completed by 189 scientists who had published in the domain of MME. A small minority (18%) of respondents claimed to have in mind a concise definition of MME. Some techniques were identified by the researchers as characterizing MME (e.g. Markov models), while others–at the same level of sophistication in terms of mathematics–were not (e.g. Cox regression). The researchers' opinions were also contrasted about the potential applications of MME, perceived as higly relevant for providing insight into complex mechanisms and less relevant for identifying causal factors. The quality criteria were those of good science and were not related to the size and the nature of the public health problems addressed.Conclusions/SignificanceThis study shows that perceptions on the nature, uses and quality criteria of MME are contrasted, even among the very community of published authors in this domain. Nevertheless, MME is an emerging discipline in epidemiology and this study underlines that it is associated with specific areas of application and methods. The development of this discipline is likely to deserve a framework providing recommendations and guidance at various steps of the studies, from design to report.

Highlights

The increased use of mathematical modeling in epidemiology (MME) is widely acknowledged [1]
The heterogeneity of the methods used in Mathematical modeling in epidemiology (MME), as well as the diversity of the problems addressed raises several questions: is there a simple shared definition for this emerging scientific discipline? Is MME only mainly aimed to answer questions of decision makers, or is it a scientific discipline of its own? What are the criteria of good science in MME? To answer these questions, we chose to collect the opinions of the scientists who do MME
The first cluster groups items (Monte-Carlo simulation, differential equations, Markov model, basic reproduction number R0) for which most respondents considered that the use of the technique in the analysis qualifies the study as belonging to the MME field

Summary

Introduction

The increased use of mathematical modeling in epidemiology (MME) is widely acknowledged [1]. When data are not there, or not yet there, MME provides rationales in Public Health problems to support decisions in Public Health, and this constitutes one of the reasons for the increased use of MME, For example, some models have been proposed for estimating non observable putative risks of importance in terms of public health, such as the risk of cancer after exposure to diagnostic radiations [2], the residual infectious risks in blood transfusion [3], or the future size of a the epidemic of an emergent disease [4]. The above list of key public health domains attests for the diversity and interest of problems for which MME might be involved in supporting public decision. There are many uncertainties in terms of definitions, uses and quality features of MME

Objectives

Methods

Results

Conclusion