On March 16 2020, French authorities ordered a large scale lockdown to counter the COVID-19 epidemic wave rising in the country, stopping non-essential economic, educational, and entertainment activities, maintaining mainly food retailers and healthcare institutions. One month later, the number of new hospitalizations and ICU admissions had reached a plateau and were beginning a slow descent. We developed a spatialized, deterministic, age-structured, and compartmental SARS-CoV-2 transmission model able to reproduce the pre-lockdown dynamic of the epidemic in each of the 13 French metropolitan regions. Thanks to this model, we estimate, at regional and national levels, the total number of hospitalizations, ICU admissions, hospital beds requirements (hospitalization and ICU), and hospital deaths which may have been prevented by this massive and unprecedented intervention in France. If no control measures had been set up, between March 19 and April 19 2020, our analysis shows that almost 23% of the French population would have been affected by COVID-19 (14.8 million individuals). Hence, the French lockdown prevented 587,730 hospitalizations and 140,320 ICU admissions at the national level. The total number of ICU beds required to treat patients in critical conditions would have been 104,550, far higher than the maximum French ICU capacity. This first month of lockdown also permitted to avoid 61,739 hospital deaths, corresponding to a 83.5% reduction of the total number of predicted deaths. Our analysis shows that in absence of any control measures, the COVID-19 epidemic would have had a critical morbidity and mortality burden in France, overwhelming in a matter of weeks French hospital capacities.

Europe is now considered as the epicenter of the SARS-CoV-2 pandemic, France being among the most impacted country. In France, there is an increasing concern regarding the capacity of the healthcare system to sustain the outbreak, especially regarding intensive care units (ICU). The aim of this study was to estimate the dynamics of the epidemic in France, and to assess its impact on healthcare resources for each French metropolitan Region. We developed a deterministic, age-structured, Susceptible-Exposed-Infectious-Removed (SEIR) model based on catchment areas of each COVID-19 referral hospitals. We performed one month ahead predictions (up to April 14, 2020) for three different scenarios (R 0 = 1.5, R 0 = 2.25, R 0 = 3), where we estimated the daily number of COVID-19 cases, hospitalizations and deaths, the needs in ICU beds per Region and the reaching date of ICU capacity limits. At the national level, the total number of infected cases is expected to range from 22,872 in the best case (R 0 = 1.5) to 161,832 in the worst case (R 0 = 3), while the total number of deaths would vary from 1,021 to 11,032, respectively. At the regional level, all ICU capacities may be overrun in the worst scenario. Only seven Regions may lack ICU beds in the mild scenario (R 0 = 2.25) and only one in the best case. In the three scenarios, Corse may be the first Region to see its ICU capacities overrun. The two other Regions, whose capacity will be overrun shortly after are Grand-Est and Bourgogne-Franche-Comté. Our analysis shows that, even in the best case scenario, the French healthcare system will very soon be overwhelmed. While drastic social distancing measures may temper our results, a massive reorganization leading to an expansion of French ICU capacities seems to be necessary to manage the coming wave of critically affected COVID-19 patients.

The concept of care pathways is increasingly being used to enhance the quality of care and to optimize the use of resources for health care. We here propose an innovative method in epidemiology that is derived from social sciences: state sequence analysis (SSA). This method takes into account the chronology of care consumption and allows for identification of specific patterns. A process for using SSA in the health area is proposed and discussed. The main steps are: data coding, measurement of dissimilarities between sequences (focusing on optimal matching methods and the choice of related costs), and application of a clustering method to obtain a typology of sequence patterns. As an example of its use in the health area, SSA was employed to analyse care pathways of a random sample of patients with multiple sclerosis. This sample has been selected from the main French healthcare database covering the period 2007 to 2013 ( n = 1 000). A five-cluster typology was obtained which allowed distinction of care consumption groups. Overall, about half of the patients had low care consumption, about one quarter had medium to high consumption, and another quarter had high consumption. We conclude that state sequence analysis is an innovative and flexible methodology that is worth considering in health care research.