Evaluating the impact of curfews and other measures on SARS-CoV-2 transmission in French Guiana

Abstract

While general lockdowns have proven effective to control SARS-CoV-2 epidemics, they come with enormous costs for society. It is therefore essential to identify control strategies with lower social and economic impact. Here, we report and evaluate the control strategy implemented during a large SARS-CoV-2 epidemic in June-July 2020 in French Guiana that relied on curfews, targeted lockdowns, and other measures. We find that the combination of these interventions coincided with a reduction in the basic reproduction number of SARS-CoV-2 from 1.7 to 1.1, which was sufficient to avoid hospital saturation. We estimate that thanks to the young demographics, the risk of hospitalisation following infection was 0.3 times that of metropolitan France and that about 20% of the population was infected by July. Our model projections are consistent with a recent seroprevalence study. The study showcases how mathematical modelling can be used to support healthcare planning in a context of high uncertainty.

Fig. 1. Geography, demographics, and timeline of interventions.

a Map of French Guiana (Source: https://gadm.org/maps.html). b Population pyramids for Metropolitan France and French Guiana. c Age-specific probability of hospitalization given infection $p_H^{Age}$ in Metropolitan France (%). d Average probability of hospitalization given infection p_H in Metropolitan France and French Guiana (%). e Daily hospital admissions in French Guiana and timeline of interventions. In panels c and d, dots denote posterior means while bars denote 95% credible intervals.

Fig. 2. Analyses made on June 18th 2020 describing a scenario with no change in transmission rates and the impact of a short-term lockdown.

a, b Projections for the number of ICU and general ward beds required under different severity scenarios (baseline in red, low severity in green, high severity in blue). Solid lines indicate model posterior means while colour areas indicate 95% credible intervals. c and d Projections for the average number of ICU and general ward beds required under different territory-wide lockdown scenarios (black represents our baseline model, red represents lockdowns starting on June 27th, blue represents lockdowns starting on July 4th, solid lines correspond to lockdowns lasting for 10 days, while dashed lines correspond to lockdowns lasting for 15 days). In all panels, black dots indicate data used to calibrate the models, while empty circles denote data not available at the time of the analyses. The black dashed line in all panels indicates the date of the analyses (June 18th). The coloured dashed lines in panels c and d indicate the start of the simulated lockdown (red for June 27th and blue for July 4th).

Fig. 3. Analyses made on July 2nd 2020 evaluating the impact of control measures implemented in French Guiana on transmission and healthcare demand.

a Estimated reproduction number through time. The horizontal dashed line indicates a reproduction number R = 1. b, d Projections for the number of daily hospital admissions (b) and ICU (c) and general ward (d) beds. Solid lines indicate model posterior means while colour areas indicate 95% credible intervals. Red is used for model M1 (one change point for the transmission rate), while blue is used for model M2 (two change points for the transmission rate). In all panels black dots indicate data used to calibrate the models, while empty circles denote data not available at the time of the analyses. The dashed line panels b–d indicates the date of the analyses (July 2nd).

Fig. 4. Analyses made on 25 August 2020.

a Projections for the seroprevalence measured with the Euroimmun assay. Solid lines indicate model posterior means while colour areas indicate 95% credible intervals. The black dot and vertical bar indicate seroprevalence estimates from between 15 and 23 July 2020 (dashed lines). b Projections for the proportion infected. Solid lines indicate model posterior means while colour areas indicate 95% credible intervals. c Age distribution of hospitalized cases in Metropolitan France (brown) and French Guiana (green). d Simulated numbers of hospital admissions in the Cher department (Metropolitan France, brown) compared to those obtained for French Guiana (green). Solid lines indicate model posterior means while colour areas indicate 95% credible intervals.