Background
There has been a split between ‘Afro-pessimists’ and ‘Afro-optimists’, with regards to the potential spread of the Sars-CoV-2 coronavirus. However, since the diagnosis of the first African case in Egypt on 16 February 2020 and the subsequent announcement of the pandemic by the World Health Organization (WHO) on 11 March 2020, little work has so far been published in scientific journals on the situation in Africa [
1,
2]. As early as February 2020, initial modelling correctly estimated that the importation of the SARS-CoV-2 virus into Africa would first affect Egypt, as well as Algeria and South Africa [
3]. At that time, researchers predicted that Francophone West African countries were at low risk of virus importation due to limited air traffic with China [
3]. Africa was expecting to be well-prepared, according to members of the
Africa Centres for Disease Control and Prevention (CDC Africa) [
4], by the time the epidemic arrived. Four months after these estimates were made, it remains true that, compared to the rest of the world, the West African region does not seem to have suffered a major epidemic shock, especially if we compare the current situation with that experienced by some countries during the Ebola crisis [
5,
6]. The most recent models predict that the 22% of the population in the African continent could become infected with SARS-CoV-2 during the first year of the pandemic, with approximately 150,000 deaths [
7] and with peaks of contamination varying from one country to another [
8]. The same researchers predict that Francophone West African countries will have few deaths related to the virus during the year: fewer than 800 in Benin, 1000 in Burkina Faso and just over 2200 in Senegal [
7]. It has been estimated that the peak of cases in Senegal would occur between 28 May and 15 June 2020 [
8]. However, these are all estimates and, at the time of writing this article, we still have little data on the reality of the state of the epidemic in Africa [
9,
10] in contrast with the countries that were first affected [
11,
12] nor do we have ample data on the effect of public health measures, which have been adapted in various ways to their national contexts [
1,
4,
9,
13,
14]. As of 21 July 2020, CDC Africa estimates that there were 736,288 cases of COVID-19 in Africa and 15,418 deaths, representing only 5% of all reported cases worldwide. In Africa (54 countries), the case-fatality rate (CFR) has been reported at 2.1%, compared to a 4.2% average CFR in all countries where data are available (
n = 215). West Africa alone accounts for 14.8% of cases and 11.2% of deaths on the African continent [
15].
Faced with the academic divide between the Afro-optimists who believe that too much has been said about the fragility of health systems in Africa, and the Afro-pessimists who remind us of the disasters caused by the Ebola virus [
16,
17], we would like to propose a third way, that of Afro-realism. In order to do so, we describe and analyse the situation in seven Francophone West African countries where our team members are established. The governments of all these countries did not wait until they were overwhelmed by the pandemic, nor did they wait for the call for public health measures and physical distancing from the WHO on 7 April 2020, to react [
18]. Anticipating the arrival of SARS-CoV-2, each government quickly put public measures in place to counter the advance of the virus, even if their populations did not always fully support them. A survey carried out in early April 2020 in 20 major African cities showed that 30% of people were opposed to closing markets, 29% to stopping traffic between cities and 22% to closing places of worship [
19]. In Senegal, on the other hand, a survey conducted in early April 2020 showed that 72.5% of people were in favour of a two-week lockdown and 85.6% were very or rather confident in the government’s capacity to deal with the crisis [
20].
In addition, voices have been raised to question the lack of inclusion and equity in the governance bodies governing the management of the crisis and the public measures taken [
21]. Similarly, amid rumours of the exploitation of the virus for political purposes and the ineffectiveness of public health measures and/or treatment, these issues were also being hotly debated in the public sphere. In Niger, for example, some believed that the pandemic was used for financial reasons: “
politicians are manipulating data to present more positive cases in the hope of winning funding from donors” [
22]. In Cameroon, people also question the statistics “
The death numbers from COVID19 is wrong” [
22]. Although the situation differed between countries, and our previous analyses have showed that routine data could be valuable for evaluating public health interventions in West Africa [
23,
24], while the quality of health data in this region of the world is often debated and brought into question [
25]. Frequently side-lined by the debates regarding the epidemiological data, public action against COVID-19 remains understudied [
2]. In early June, a first study in Kenya with a sample of 213 people demonstrated the effectiveness of the policy package on the epidemic’s reproductive rate [
26]; however, there has been a lack of similar analysis in the Francophone West African region. The objective of this article is to describe and analyse the epidemiological evolution of COVID-19 in seven Francophone West African countries during the first 7 months of the pandemic, as well as the public measures taken to deal with it.
Methods
This paper analyses quantitative and qualitative empirical data from several sources, mainly collected from a regional platform on COVID-19.
We launched
Covid19Africa.com, designed as an information platform to track the SARS-CoV-2 pandemic, on 31 March 2020. Focusing on West Africa (including non-Francophone countries) and Francophone countries throughout Africa, the site is designed to facilitate contextual readings and effective management of data. The objective of the site is to facilitate the open and free sharing of data with a comparative perspective. The platform relies on a network of 26 contributors spread across the countries. A total of 32 countries are subject to daily monitoring of epidemiological data published on WHO recognized sites, official government sites and situation reports (SITREPs). All these public data are compared in order to validate the recognized situation in the countries. International sites were not preferred for this analysis because they publish data that often differ from the actual situation in these countries. Indeed, these websites/databases often consider the official case announcement date as the date of diagnosis, while there may be several (up to 3) days of difference between a positive case diagnosis and the official announcement of it. This is particularly the case for data relating to Benin and Burkina Faso.
Study area
The analysis presented in this article focuses on seven West African countries (Benin, Burkina Faso, Côte d’Ivoire, Guinea, Mali, Niger and Senegal), which were chosen because they illustrate the varying dynamics of the pandemic and the taking of various government measures in the same geographical area, and because our team is sufficiently familiar with these contexts to analyse the respective national situations.
Quantitative analyses
The epidemiological data from this study comprise the cumulative number of daily cases of COVID-19, the daily number of deaths and the number of tests performed. The study period is from 28 February 2020, the date of the first case detected in west africa (Nigeria), to 1 October 2020. The epidemic curves were constructed to describe and compare the trends in each country and a 7-day moving average was applied.
The data are dynamically mapped to visualize the distribution of cases and deaths, using a
Json program developed by our team that allows the publication, in cartographic form, of country data entered into a shared database. Spatial analyses in the form of bivariate maps [
27] and spatiotemporal analyses complement the dynamic maps available at
Covid19Afrique.com. The bivariate maps combine the attack rate and the number of COVID-19 cases. The spatial and temporal analyses were performed using the spatial scan statistics implemented in SatScan (version 9.4) [
28]. This method detects regions with higher-than-expected disease incidence in time and space by assigning them a relative risk, producing as a result a list of spatiotemporal clusters that can be used to identify the epidemic phases in the study area.
We propose an analysis of spatiotemporal clusters only for the countries where lower administrative data are available: Burkina Faso and Senegal, in order to illustrate the concentration of COVID-19 cases within countries, as well as the intermittent emergences of clusters within countries. We used Kulldorff’s space-time scan statistical analysis to detect the temporal, spatial, and space-time community clusters of COVID-19 at infra-national scale to verify whether the geographic clustering [
28] of COVID-19 has been caused by random variation or not.
The system uses a spatio-temporal scan statistic in the form of a circle/cylindrical scan [
29]. Space-time scan statistics complement baseline disease rate maps, and, using a variety of data models, can be used to determine whether observed space-time patterns of a disease are due to chance or randomly distributed. Scan statistics detect clusters that are outliers (a cluster not observed under baseline conditions). The statistics use cylinder (scan windows) that are centred on grid points and move (scan) systematically across a study area to identify clusters of cases (each window counts the number of cases aggregated by geographical unit) [
30]. A retrospective analysis was carried out to identify all significant clustering events (epidemic phase) that occurred up until the time of writing this paper. The permutation model of the scanning statistics tests only used case data within each candidate cylinder and calculated the ratio between the number of observed cases (we considered that at least 5 cases of COVID-19 were needed to be considered) and the number of expected cases under the null hypothesis that the observed cases are randomly distributed in space and time. The expected number of cases was calculated as the sum of all observed cases multiplied by the size of the scanning window and divided by the size of the entire study area [
28]. The observed/expected ratio (Tables
3 and
4) was used to estimate the probability that a candidate cylinder represents a true significant clustering event of COVID-19 cases. The window with the maximum likelihood is defined as the most likely cluster area, and other clusters with statistically significant log-likelihood ratios (LLR) were defined as the secondary potential clusters. The
P-values of LLR were estimated through 999 Monte Carlo simulations. We have mapped the attack rates of COVID-19 per commune in order to display on the same map both the global situation of the epidemic and the clusters detected by this analysis. The cartographic documents thus summarize the epidemic situation in the outliers.
Some spatial and temporal analyses have already been carried out in Africa on the influence of meteorological factors in promoting or hindering the spread of the aerosol pathogen COVID-19 in Africa [
31] or on the early spatial and temporal dynamics of COVID-19 in the first 62 days of the disease outbreak on the African continent [
32]. This analysis adds to those already published by Adekunle [
31] and Gavawam [
32] but this paper is one of the first spatio-temporal analysis carried out in African countries at lower administrative level, but is part of a trend including other similar studies carried out in the USA at the beginning of the pandemic [
30] and a study in Kuwait targeting a specific population within the capital [
33].
These analyses are, however, dependent on the implementation of the tests in the regions and on the reporting modalities, but they do make it possible to highlight situations where the local incidence is high for all localities, especially smaller ones. The reporting process is identical in Burkina Faso and Senegal, which makes the analyses comparable.
Spatial cluster analyses were performed using SaTScan® v9.4.4 (Martin Kulldorff, Harvard Medical School, Boston, MA, USA and Information Management Services Inc., Silver Spring, MD, USA. The maps were generated using Quantum-GIS® v3.10 (Open Source Geospatial Foundation Project, Beaverton, OR, USA).
Qualitative analyses
A documentary analysis based on situation reports from country ministries, scientific articles, reports from the WHO, CDC Africa, and the national press was compiled to enable the recording and tracking of events and government measures to produce the synthesis of information presented in this article. A qualitative analysis of the content of these documents was carried out in addition to a situational analysis carried out by the researchers present in each of the seven countries. A transversal analysis of the content of these studies was carried out and validated by all the authors of this article.
Discussion
Supported by our team of multidisciplinary collaborators from the COVID-19 data platform, we have undertaken this project to describe and analyse the epidemiological evolution in seven Francophone West African countries during the first 7 months of the pandemic, and the public measures taken to deal with it. Our epidemiological analysis demonstrated the diverse nature of COVID-19 outbreaks depending on the context of its spread, and has highlighted the delicate issue of case detection.
However, despite the diversity of contexts and epidemiological situations in the countries [
2], we have noticed a certain similarity in the reactions to the arrival of the pandemic among them. It is true that each country was able to take advantage of the relatively late arrival of the virus in the sub-region, compared to the Asian and European continents, in order to prepare and even anticipate certain health measures. It has been hypothesized that this was also due to the use of evidence, including advice from WHO and the Africa Centre for Disease Control [
34]. Conversely, other measures seem to have been taken in haste and without consultation, which have led to misunderstandings, frustration and protests. Studies have been undertaken on the social acceptability of the measures [
35] to be carried out, but the mistrust often encountered and the sometimes violent demonstrations (Côte d’Ivoire, Guinea, Mali, Niger, Senegal) show that the decisions behind the measures and their content have not always been understood and integrated into policy. A certain form of inconsistency, as elsewhere in the world, has also been noted regarding the lifting of certain restrictive measures, the reasons for which are probably other than health concerns. This lack of consistency between epidemiological curves and public health measures has thus sometimes led to scepticism about the very existence of COVID-19 (not as world pandemic, but as a reality in the specific case of African countries discussed here), as was sometimes the case with Ebola [
36]. However, anticipation and preparation are precisely at the heart of epidemic management as the case of Ebola has clearly shown in the region [
37] and trust and governance are essential elements of good pandemic preparedness [
37].
Also, and despite the respite offered by the gradual advance of the pandemic, we have found that countries took a number of relatively similar - if not identical - health measures, which more fundamentally raises the question of the appropriateness of these measures to their national contexts and reawakens the myth of a “turnkey” response applicable to all and at all times. Yet all the scientific literature on public health interventions, including in Africa [
38,
39] and on COVID-19 [
13] affirms the importance of taking contexts into account in order for measures to be effective [
40]. This need to contextualize the health response also requires taking into account the specificities of the disease. Although the state of knowledge on this virus is still limited and constantly evolving, evidence of the effectiveness or the processes to be used to develop or organise specific actions is already available. For example, regarding containment measures, a scoping review synthesising relevant knowledge published in 2018 highlights the importance of community involvement for their effectiveness [
41]. A study on Ebola in 2014–2016 in the region similarly showed the need for community involvement in disease control interventions that take into account local dynamics [
39]. The analysis of the situation in five African countries at the beginning of the pandemic also showed the importance of community involvement [
10]. However, communities are still far from the process of reflection and formulation of health measures to be introduced within the context of COVID-19. This unfortunate observation has also been made for COVID-19 in Europe and elsewhere [
21,
42]. However, community engagement will be undoubtedly essential when testing the eventual vaccine, as was the case for Ebola [
43].
The observation drawn here is also valid for the management committees which, in the seven countries analysed in this paper, as elsewhere in the world [
21], have effectively neglected to involve representatives of users, patients or NGOs. As everywhere, the power of these committees remains inexorably in the hands of clinicians, as the interdisciplinary, intersectoral or health promotion approach has been totally ignored [
21]. Similarly, the presence of women has been completely side-lined, here as elsewhere, yet, as Bali et al. noted in a recent paper “
women are not only a vulnerable population, they can serve as agents of change whose contributions can improve epidemic response and recovery” [
44]. However, this situation of exclusion is deeply rooted in the region and the pandemic has therefore not been able to change this state of path dependency. The paradigm shift in public health approaches that this pandemic has shown to be indispensable is still far off [
45].
As in most countries across the world, politics has also been widely invoked in the management of the pandemic in the countries of the region. This has been prevalent in countries where elections were held during the crisis (Benin, Guinea, Mali) but also in countries where political movements have taken advantage of some of the challenges faced by governments and political parties in power, in order to attack them in the face of upcoming elections (Burkina Faso, Côte d’Ivoire, Senegal). In Côte d’Ivoire, funds distributed by Deloitte to help large companies to counterbalance the economic crisis precipitated by the pandemic were offered primarily to companies sympathetic to the ruling party. As a result of such actions, several civil society organizations in many countries have sometimes denounced the state of “management in total uncertainty”, such as the National Coalition for Health and Social Action in Senegal. The same is true for religious bodies where, in some countries such as Senegal or Mali, they have taken a prominent place in the debates and had a major influence on political decisions concerning certain measures, particularly, but not exclusively, concerning the closure of places of worship (especially for Muslims in the context of the Ramadan period). The pandemic has thus sometimes again highlighted the close links between the religious and political spheres in the region.
Although some believe, as in South Africa, that the measures taken have made it possible to delay the peak of the epidemic [
14], we believe that, in the context of the seven countries concerned here, such an assessment is impossible to make given the current state of knowledge. Driven by the urgency to act, measures have been applied almost everywhere and without any real means of evaluating their effectiveness. Moreover, most of these interventions were stopped, or their scope was reduced, at the end of June 2020, leaving a very short time window for evaluation, not to mention the fact that few research organizations will be able to measure the degree of fidelity with which they were implemented, and the real application of these measures in these countries [
46]. A study in the Democratic Republic of Congo showed that the official recommendations to wear masks were not respected [
47]. The challenges for modelling the effectiveness of these interventions will be enormous. In addition, these interventions were in most cases so complex that it is doubtful whether their effectiveness can be studied systematically [
48,
49]. Yet we were warned a few years ago in the region that “
considering a public health measure with such dramatic social effects as containment, the transnational scientific community should engage rapidly in building evidence about the efficacy of containment in the Ebola outbreak” [
50]. Several countries in the region have implementing seroprevalence surveys, the results of which may shed more light on the circulation of the virus, the effectiveness of interventions on health outcome or simply the herd immunity.
Finally, our cross-sectional analysis confirms all the challenges related to data access and the importance of promoting open access to data, especially when, as is often the case in the region, access to documents and epidemiological data is particularly complicated and difficult. The COVID-19 pandemic has only confirmed the importance of this situation [
2,
51] where no country in the region has yet put its epidemiological data, apart from those communicated daily to the media, online. Something that the West African Health Organization (WAHO) should do. The organization of our collaborative platform has made it possible to create this dynamic of rapid information sharing, as international or sub-regional organizations have not been able to achieve this speed of response. However, our analysis also highlights the challenges of the quality of these data, particularly when, for example, deaths are not counted in communities in Guinea in a disaggregated manner, when RDTs (Rapid Diagnostic Tests) are used in Benin or when the number of tests is reduced in Senegal or Guinea. The magnitude of the pandemic is thus likely to be underestimated here [
52], as elsewhere [
49]. Access to epidemiological data, if available, will make it possible to assess excess mortality in the countries, at different territorial scales, and thus estimate whether the figures disseminated reflect the real situation [
47]. Similarly, the proposed infra-national analysis illustrates the association of the spread of the epidemic from the capitals to the secondary cities. The spatio-temporal analysis characterizes situations where the incidence is much higher than expected. It thus shows that on an infra-national scale, all territories and communities are impacted and exposed to the virus, but are also sometimes much more isolated from health care facilities than large urban centres. It is also important to consider here that the analysis is at the local level of the municipality where the tests are carried out. This is a less usual scale of analysis but one that allows even the smallest number of households to be identified with a high incidence. This type of analysis proves to be very useful in the epidemic phase for targeting affected areas, but requires good quality data that can be retrieved in real time, to do this, it is necessary to mobilize prospective studies in order to identify spatio-temporal clusters. In several aspects, COVID-19 thus demonstrates the fundamental need for credible data as a governance tool to identify and support populations, particularly the most vulnerable [
53].
Acknowledgements
In particular, we thank Oriane Bodson (Benin), Chiarella Mattern (Madagascar), Isidore Sieleunou (Cameroon), Abdouramane Coulibaly (Mali), Anne Bekelynck (RCI), Frédéric Le Marcis (Guinea), Flore-Apoline Roy (Senegal), Aloys Zongo (Niger), IRD Representatives in West and Central Africa, François Parenty (France), Fondation Paul Ango Ela (Cameroon), Marie Morelle (Cameroon), Ibrahim Sana (Burkina Faso), Cloudly Yours (France, Accommodation), Florence Fournet (Côte d’Ivoire), Florence Boyer (Niger), Marjorie Le Bars (Mali), Dahab Manoufi (Chad), Sebastien Segniagbeto (Togo), Ousmane Koita (Mali), Gilles Salaun (France), Fatoumata Binetou Diongue Lopes (Senegal), Philippe Tous (Mauritania), Séverine Carillon (Senegal). The authors would like to thank Jack Stennett for editing support.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.