Fig. 1: Estimates of under-ascertainment of deaths in Damascus.

In (a) the daily model-predicted COVID-19 deaths in Damascus are shown using the default model parameters and a range of values for the under-ascertainment of COVID-19 deaths. These are compared to the excess mortality reported by the government during 25 July–1 August, which is shown with point ranges, with estimates assuming a baseline of 32 and 64 deaths per day shown with a circle and triangle respectively. Trajectories with 0.05 and 0.1% of deaths reported are not shown above 200 daily deaths to ease viewing of other curves. In (b) the log likelihood for each level of under-ascertainment is shown for different assumed baseline mortality. Model log likelihoods presented reflect the mean model log likelihood across the full sensitivity analysis conducted (n = 14,000), suggesting that under-ascertainment is likely between 1 and 3% when viewed across both baseline mortality estimates. The box represents the median and interquartile range (IQR) and whiskers represent the IQR ± 1.5*IQR with points depicting values outside this range. In (c) the model-predicted hospital occupancy for the simulations in (a) are shown. The hospital capacity for Damascus is shown with a dashed horizontal line, with the 2-week period in which hospitals were reported to be first at capacity shown shaded in red. In both (a, c) the median of 100 draws from the posterior parameter space are presented for visualisation purposes.