Fig. 5: Characterising the dynamics of the second wave in Damascus.

In (a) we show the number of obituary notifications uploaded daily to the community maintained ‘Damascus Mortality’ Facebook page. In (b), the daily inferred excess death notifications are shown, with error bars showing the 95% CI and the blue line indicating the weekly average excess deaths (n = 100). In (c) model fits (purple, median and 95% CI in shaded region) to the notification excess death data (vertical green bars) are shown under the assumption that 40% of total COVID-19 deaths are ascertained by the notification excess deaths, which correctly captures the excess mortality reported by the Syrian government between 25 July and 1 August (black dots). The model-predicted COVID-19 deaths in (c) are significantly greater than the official reported COVID-19 deaths (blue bars), with the reporting fraction (official deaths divided by model-predicted deaths) since July shown in (d). In (d) the 2-week rolling average reporting fraction is shown, with the 95% CI shown between the dashed lines.