The Heterogeneous Severity of COVID-19 in African Countries: A Modeling Approach

Abstract

The COVID-19 pandemic has had a considerable impact on global health and economics. The impact in African countries has not been investigated thoroughly via fitting epidemic models to the reported COVID-19 deaths. We downloaded the data for the 12 most-affected countries with the highest cumulative COVID-19 deaths to estimate the time-varying basic reproductive number ([Formula: see text]) and infection attack rate. We develop a simple epidemic model and fitted it to reported COVID-19 deaths in 12 African countries using iterated filtering and allowing a flexible transmission rate. We observe high heterogeneity in the case-fatality rate across the countries, which may be due to different reporting or testing efforts. South Africa, Tunisia, and Libya were most affected, exhibiting a relatively higher [Formula: see text] and infection attack rate. Thus, to effectively control the spread of COVID-19 epidemics in Africa, there is a need to consider other mitigation strategies (such as improvements in socioeconomic well-being, healthcare systems, the water supply, and awareness campaigns).

Keywords: Attack rate; Pandemic; Reproduction number; SARS-CoV-2; Seroprevalence.

Fig. 1

Daily confirmed cases (in black triangles) and deaths (in red triangles) of COVID-19 in 12 African countries with the most cases of deaths from COVID-19 (population standardized, cases and deaths per 1 million people)

Fig. 2

Schematic diagram of COVID-19 model. The compartments in green represent the uninfected populations, the orange compartments represent the infected populations, and the dark orange class denotes the death proportion

Fig. 3

Time series fitting results of weekly confirmed COVID-19 deaths (in red circles) in nine of the 12 African countries with the most COVID-19 deaths, which were hit milder (with low reporting rates). Deaths are population standardized (i.e., deaths per 1 million people). The median of the simulation is represented by the black curve, and the time-varying basic reproductive number ($R_0(t)$) is denoted by the blue dashed curve. The 95% confidence interval of the simulation is shown by the shaded (gray) region. The estimated IAR is displayed in the title of each panel

Fig. 4

Time series fitting results of weekly confirmed COVID-19 deaths (in red circles) in three of the 12 African countries with most COVID-19 deaths, which were hit harder (or with a relatively higher reporting effort): a Libya, b South Africa, c Tunisia and d South Africa with excess deaths. Deaths are population standardized (i.e., deaths per 1 million people). The median of the simulation is represented by the black curve, and the time-varying basic reproductive number ($R_0(t)$) is denoted by the blue dashed curve. The 95% confidence interval of the simulation is shown by the shaded (gray) region. The estimated IAR is portrayed in the title of each panel