Gaëtan Texier1, Liliane Pellegrin2, Claire Vignal3, Jean-Baptiste Meynard4, Xavier Deparis5, Hervé Chaudet6. 1. Centre Pasteur du Cameroun, BP 1274, Yaoundé, Cameroon; UMR 912/SESSTIM - INSERM/IRD/Aix-Marseille University/Faculty of Medicine, 27 Bd Jean Moulin, 13385 Marseille, France. Electronic address: gaetex1@gmail.com. 2. UMR 912/SESSTIM - INSERM/IRD/Aix-Marseille University/Faculty of Medicine, 27 Bd Jean Moulin, 13385 Marseille, France; French Armed Forces Center for Epidemiology and Public Health (CESPA), Camp de Sainte Marthe, 13568 Marseille, France. Electronic address: liliane.pellegrin_chaudet@univ-amu.fr. 3. Center for Research in the Psychology of Cognition, Language and Emotion (PsyCLE), 29 avenue Robert Schuman, 13621 Aix-en-Provence, France. Electronic address: claire.vignal@orange.fr. 4. UMR 912/SESSTIM - INSERM/IRD/Aix-Marseille University/Faculty of Medicine, 27 Bd Jean Moulin, 13385 Marseille, France; French Armed Forces Center for Epidemiology and Public Health (CESPA), Camp de Sainte Marthe, 13568 Marseille, France. Electronic address: jb.meynard@wanadoo.fr. 5. UMR 912/SESSTIM - INSERM/IRD/Aix-Marseille University/Faculty of Medicine, 27 Bd Jean Moulin, 13385 Marseille, France; French Armed Forces Center for Epidemiology and Public Health (CESPA), Camp de Sainte Marthe, 13568 Marseille, France. Electronic address: xavier.deparis@wanadoo.fr. 6. UMR 912/SESSTIM - INSERM/IRD/Aix-Marseille University/Faculty of Medicine, 27 Bd Jean Moulin, 13385 Marseille, France. Electronic address: herve.chaudet@univ-amu.fr.
Abstract
INTRODUCTION: Epidemiologists manage outbreak identification and confirmation by means of a "situation diagnosis", which involves validating (or invalidating) an alarm (signal identified as abnormal) as an alert (a real, characterized outbreak) and proposing the first countermeasures. This work investigates how uncertainty is materialized during this stage, and how experts develop strategies to address this uncertainty with the help of an early warning system. METHODS: We built an experiment using a simulation platform with a scenario involving both a natural and an intentional outbreak. Observations of expert activities were recorded and formalised using a specific task analysis method. These formatted data were then categorized by applying RAWFS (Reduction- Assumption - Weighing - Forestalling- Suppression) heuristics. RESULTS: We quantified uncertainty and the mechanisms involved. During the situation diagnosis, two sorts of uncertainty were characterized: practice-imposed uncertainty and situation-imposed uncertainty. We did not find either weighing pros and cons or suppression strategies in this area of expertise, but highlight the predominance of coping strategies that relied on reduction (66,4%) and assumption-based reasoning. We observed a predominance of the phone (89%) to cope with uncertainty and among electronic tools, the surveillance system plays a major role (69% of cases) and is mainly used in reduction strategies. We detail tools and systems used to support experts in their coping strategy. CONCLUSION: We confirmed that a surveillance system must include different features that provide relevant information to help users reduce uncertainty and thus support their decision making. In that perspective, the flow diagram and proposal presented in this study can help prioritize the necessary changes to surveillance system design.
INTRODUCTION: Epidemiologists manage outbreak identification and confirmation by means of a "situation diagnosis", which involves validating (or invalidating) an alarm (signal identified as abnormal) as an alert (a real, characterized outbreak) and proposing the first countermeasures. This work investigates how uncertainty is materialized during this stage, and how experts develop strategies to address this uncertainty with the help of an early warning system. METHODS: We built an experiment using a simulation platform with a scenario involving both a natural and an intentional outbreak. Observations of expert activities were recorded and formalised using a specific task analysis method. These formatted data were then categorized by applying RAWFS (Reduction- Assumption - Weighing - Forestalling- Suppression) heuristics. RESULTS: We quantified uncertainty and the mechanisms involved. During the situation diagnosis, two sorts of uncertainty were characterized: practice-imposed uncertainty and situation-imposed uncertainty. We did not find either weighing pros and cons or suppression strategies in this area of expertise, but highlight the predominance of coping strategies that relied on reduction (66,4%) and assumption-based reasoning. We observed a predominance of the phone (89%) to cope with uncertainty and among electronic tools, the surveillance system plays a major role (69% of cases) and is mainly used in reduction strategies. We detail tools and systems used to support experts in their coping strategy. CONCLUSION: We confirmed that a surveillance system must include different features that provide relevant information to help users reduce uncertainty and thus support their decision making. In that perspective, the flow diagram and proposal presented in this study can help prioritize the necessary changes to surveillance system design.