Statistical analysis of teleconsultations in the context of the COVID-19 epidemic

Abstract

We consider the spatial and temporal distribution of teleconsultations associated with the COVID-19 epidemic in Martinique, French West Indies from March to May 2020. Statistical tools for the detection of high-frequency areas are presented. The mathematical modeling underlying the so-called scanning methods are discussed taking into account the influence of covariates on teleconsultation occurrences and their evolution over time. Some tools available in the R programming environment and the SaTScan software are presented. The spatio-temporal statistical analysis of COVID-19 teleconsultations is performed. Areas for which the frequencies of people using teleconsultations are significantly higher than elsewhere are presented and these results are discussed with respect to covariates providing relevant information on specific characteristics of Martinique island.

KEYWORDS:

• Counting measure
• COVID-19
• likelihood ratio test
• Poisson process
• Cox process
• scan statistics
• R environment

Acknowledgments

We would like to thank the Management of the Caisse Générale de Sécurité Sociale de la Martinique for access to data on teleconsultations and frailty indices.