key: cord-284965-6hss53nu authors: Evans, M. V.; Garchitorena, A. V.; Rakotonanahary, R. J.; Drake, J. M.; Rajaonarifara, E.; Ngonghala, C. N.; ROCHE, B.; Bonds, M. H.; Rakotonirina, J. title: Reconciling model predictions with low reported cases of COVID-19 in Sub-Saharan Africa: Insights from Madagascar date: 2020-07-17 journal: nan DOI: 10.1101/2020.07.15.20149195 sha: doc_id: 284965 cord_uid: 6hss53nu The COVID-19 pandemic has wreaked havoc globally, and there has been a particular concern for sub-Saharan Africa (SSA), where models suggest that the majority of the population will become infected. Conventional wisdom suggests that the continent will bear a higher burden of COVID-19 for the same reasons it suffers high burdens of other infectious diseases: ecology, socio-economic conditions, lack of water and sanitation infrastructure, and weak health systems. However, so far SSA has reported lower incidence and fatalities compared to the predictions of standard models and the experience of other regions of the world. There are three leading explanations, each with very different implications for the final epidemic burden: (1) low case detection, (2) differences in COVID-19 epidemiology (e.g. low R0), and (3) policy interventions. The low number of cases to date have led some SSA governments to relax these policy interventions. Will this result in a resurgence of cases? To understand how to interpret the lower-than-expected COVID-19 case data in Madagascar, we use a simple age-structured model to explore each of these explanations and predict the epidemic impact associated with them. We show that the current incidence of COVID-19 cases can be explained by any combination of the late introduction of first imported cases, early implementation of non-pharmaceutical interventions (NPIs), and low case detection rates. This analysis reinforces that Madagascar, along with other countries in SSA, remains at risk of an impending health crisis. If NPIs remain enforced, up to 50,000 lives may be saved. Even with NPIs, without vaccines and new therapies, COVID-19 could infect up to 30% of the population, making it the largest public health threat in Madagascar until early 2021. The COVID-19 pandemic has wreaked havoc globally, and there has been a particular concern 20 for sub-Saharan Africa (SSA), where models suggest that the majority of the population will 21 become infected. Conventional wisdom suggests that the continent will bear a higher burden of 22 COVID-19 for the same reasons it suffers high burdens of other infectious diseases: ecology, 23 socio-economic conditions, lack of water and sanitation infrastructure, and weak health systems. 24 However, so far SSA has reported lower incidence and fatalities compared to the predictions of 25 standard models and the experience of other regions of the world. There are three leading 26 explanations, each with very different implications for the final epidemic burden: (1) low case 27 detection, (2) differences in COVID-19 epidemiology (e.g. low R0), and (3) policy interventions. 28 The low number of cases to date have led some SSA governments to relax these policy 29 interventions. Will this result in a resurgence of cases? To understand how to interpret the lower-30 than-expected COVID-19 case data in Madagascar, we use a simple age-structured model to 31 explore each of these explanations and predict the epidemic impact associated with them. We 32 show that the current incidence of COVID-19 cases can be explained by any combination of the 33 late introduction of first imported cases, early implementation of non-pharmaceutical interventions 34 The COVID-19 pandemic has killed hundreds of thousands of people, collapsing health systems 78 and economies around the world. Most models predict that without intervention, the majority of 79 the global population will become infected and tens of millions will die as a result of the pandemic 80 [1] . There have been particular concerns for sub-Saharan Africa (SSA) [2-4], as the major factors 81 that drive high burdens of other infectious diseases, such as the environmental and socio-82 economic conditions, lack of water and sanitation infrastructure, and weak health systems, are 83 equally relevant to the threat of COVID-19. However, so far, the perceived burden of COVID-19 84 in SSA is low compared to expectations both from epidemiological models and from epidemic 85 patterns in other regions of the world [5,6]. Though SSA comprises 11% of the global population, 86 it has only 3.6% of the total global COVID-19 incidence, much of which is due to case reports 87 from South Africa [7] . As of July 2020, most SSA countries are reporting fewer than 100 new 88 cases daily [8] . There are three leading potential explanations for the lower observed burden of 89 COVID-19 in SSA: 1) low case detection, 2) region-specific epidemiology (e.g., different R0), and 90 3) early implementation of effective policy interventions. The important difference among these 91 alternative explanations is that explanations based on low case detection and effective 92 interventions imply that there will be a major resurgence if interventions are relaxed, while 93 explanations based on region-specific epidemiology allow for a safe reopening. 94 The lower-than-expected number of reported cases may be due to low detection and 95 reporting rates. RT-PCR laboratory capacity in SSA is limited [9] and many countries have among 96 the lowest testing rates in the world [8] . Moreover, health care access for fever and respiratory 97 infections is low [10], which means that many symptomatic cases will not be detected, and the 98 stigma associated with COVID-19 could further reduce health-seeking behaviors [11] . 99 The epidemiology-based explanations for low COVID-19 cases are based on 100 considerations of well-established factors: warmer climates, younger age distributions, and lower 101 contact rates due to lower population density and transportation infrastructure in rural areas 102 [12, 13] . In addition, there is considerable interest in the potential immune-mediated 103 consequences from living in a system with greater exposure to other infectious diseases and 104 related prophylaxis and therapeutics [14] . For example, there are major trials underway on the 105 effects of trained immunity due to the BCG vaccine, which may increase innate immunity against 106 a range of respiratory infections [15] . However, many of these hypotheses have recently come 107 into doubt. The pandemic phase of COVID-19 is driven by high susceptibility, not climate [16] , 108 suggesting that warmer, humid climates will not decrease transmission at this time. Further, past 109 outbreaks of influenza, including the 1968 pandemic and 2009 H1N1 outbreak, spread throughout 110 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 17, 2020. It remains unknown whether SSA-specific conditions will result in different epidemic 124 dynamics in SSA than elsewhere, and whether the current lower-than-expected case burdens 125 can be explained solely by detection rates and policies. To explore these issues, we compare 126 COVID-19 reported case data with predictions from a simple SEIR compartmental model for 127 Madagascar that integrates age-structured social contact matrices and fatality rates, assuming Madagascar reported its first imported case relatively late, on March 20, 2020, and the 139 government implemented NPIs early in the epidemic (Table 1) . Madagascar instituted a national 140 lockdown on March 23, 2020, three days before its first case attributed to local transmission. 141 Testing practices are also similar to those in other SSA countries, initially focusing on screening 142 for imported cases and eventually expanding to test contacts of known cases for local 143 transmission. 144 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 17, 2020. . https://doi.org/10.1101/2020.07.15.20149195 doi: medRxiv preprint Our exercise shows that the current incidence of COVID-19 in Madagascar can be 145 explained by the early and effective implementation of NPIs and low case detection rates, both of 146 which are supported by strong anecdotal evidence. In contrast, arguments of regional-specific 147 epidemiology are based on correlational observations that have yet to be proven. This suggests 148 that the epidemic will grow in Madagascar, and similar countries in SSA, and that these 149 populations remain at risk of an impending health crisis. Our model indicates that, if NPIs remain 150 enforced at the level needed to explain current case burdens, nearly 50,000 lives could be saved. By July 2020, the simple forecast for an unmitigated epidemic predicts a daily incidence of 34,322 156 cases, which is nearly 500 times the reported daily incidence (Fig. 1A) . Simply accounting for 157 detection rates between 0.1 -1% results in predictions that closely approximate the reported daily 158 incidence of COVID-19 cases in Madagascar (Fig. 1B) . Are these low levels of case detection 159 reasonable? For countries where per capita testing is over 100-fold higher than in Madagascar 160 (currently 79.1/100,000 population), it is estimated that less than 10% of COVID-19 cases have 161 been detected [29] . Though the precise case detection rates for Madagascar cannot be discerned 162 from available data, there are a number of indicators suggesting that these are lower than the 163 already low rates of Europe or the US. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 17, 2020. A reduction in transmission rates of 30%, relative to an unmitigated scenario, can also explain the 193 daily case report rates of COVID-19 in Madagascar (Fig. 1C) . This reduction could be the result 194 of NPI policies put in place in Madagascar or of innate characteristics affecting the epidemiology 195 of COVID-19 (e.g. baseline contact patterns, climate, etc.). NPIs were implemented within three 196 days after the first confirmed imported case of COVID-19 in the country (Table 1) (Table 1) , and the targeted lockdown of these population centers could 206 have reduced spread to the rest of the country. While mobility data is not available for 207 Madagascar, other SSA countries have reported reductions in mobility ranging from 1.4% in 208 Zambia to 19% in Senegal compared to pre-NPI levels [36] . With a sparse road network that is 209 well regulated in Madagascar, 30% represents an obtainable reduction in contact rates. 210 Because NPIs were implemented early in the epidemic, their effects on transmission 211 cannot be disentangled from baseline contact patterns in the country, which may be lower than 212 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 17, 2020. . https://doi.org/10.1101/2020.07.15.20149195 doi: medRxiv preprint those of Europe or the US. Nearly half (47.73%) of the Malagasy population lives in rural areas, 213 and most of the country is over 3 hours from a population center with more than 50,000 people 214 [37]. Therefore, baseline contact patterns in the rural areas of Madagascar may be reducing 215 disease spread in a way that is unidentifiable from the effects of NPIs. 216 217 Which path is Madagascar on? 218 The evidence presented here provides no indication that the epidemiology (e.g. R0) of COVID-19 219 is fundamentally different in a fairly typical SSA country than elsewhere. We demonstrate that the 220 current trend in reported cases in Madagascar can be explained by its early stage in the epidemic, 221 combined with low detection rates and lower contact rates from NPIs ( Fig. 2A) . Understanding 222 how much of the discrepancy between predicted and reported case burdens is due to low 223 detection rates or NPIs has enormous implications for our expectations regarding the 'true' burden 224 of COVID-19 in Madagascar. For this, we explored different combinations of detection rates and 225 NPI efficacy that explain the observed trend in reported cases, together with associated 226 predictions of epidemic morbidity and mortality burdens (Figure 2 ). If the low number of reported 227 cases is due primarily to a low detection rate, we predict over 13 million people will be infected 228 with the virus if NPIs are not in place (Fig. 2C,D) , imposing a huge burden on an already 229 weakened health system. On the other hand, if the low number of cases is due to a reduction in 230 contact patterns, the model predicts a lower total burden of approximately 8 million people 231 infected with the virus (Fig. 2C,D) . If NPIs are driving these contact patterns and are responsible 232 for the lower-than-expected case burden, the lifting of these restrictions is very likely to lead to 233 an uncontrolled outbreak. 19 related deaths (1,500) . This study assumed that 240 the regional particularities of SSA will decrease disease transmission and fatality rates based on 241 country-specific proxies for these factors, such as climate, transportation networks, and contact 242 matrices. Importantly, this study only considered reductions in transmission via reduced risks of 243 exposure, with a maximum of 2.6% of the population of Madagascar at risk of exposure at any 244 one time. While socio-ecological context is necessary to understand disease transmission, our 245 exercise suggests that the difference between reported and predicted case burdens in SSA can 246 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 17, 2020. . https://doi.org/10.1101/2020.07.15.20149195 doi: medRxiv preprint be just as easily explained by accounting for low detection rates and NPIs that reduce 247 interpersonal contact. 248 249 We do not currently have enough evidence to suggest that the epidemiology of COVID-251 19 is different in Madagascar than elsewhere. The low number of reported cases can be explained 252 by low detection rates, late introduction, and early and effective implementation of NPIs. In 253 contrast to the theory of a salutary epidemiology, each of these explanations is supported by 254 strong anecdotal evidence ( Table 2 ). As lockdowns are gradually lifted, other NPIs, such as 255 handwashing and social distancing, should be implemented to avoid a rapid growth in cases. The 256 public health system should remain prepared for an outbreak, with a peak of infections expected 257 between August and December depending on the transmission scenario (Fig. 2D) CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 17, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 17, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 17, 2020. Early and effective NPIs --Lockdown in population centers implemented three days after first imported case --Limiting travel on fragmented paved road network can easily disrupt withincountry movement 306 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 17, 2020. . https://doi.org/10.1101/2020.07.15.20149195 doi: medRxiv preprint Global, regional, and national estimates of the 309 population at increased risk of severe COVID-19 due to underlying health conditions in 310 2020: a modelling study Africa in the Path of Covid-19 COVID-19 in Africa: the spread and 315 response Looming threat of COVID-19 infection in Africa: act 317 collectively, and fast COVID-19: Shining the Light on Africa Predicting COVID-19 spread and public health 322 needs to contain the pandemic in West-Africa Access to lifesaving medical resources for 330 African countries: COVID-19 testing and response, ethics, and politics Assessing coverage of interventions for reproductive, 333 maternal, newborn, child, and adolescent health and nutrition Lessons learned from HIV can inform our approach to COVID-19 stigma The potential effects of widespread community 338 transmission of SARS-CoV-2 infection in the World Health Organization African Region: a 339 predictive model The relatively young and rural population may limit the 341 spread and severity of COVID-19 in Africa: a modelling study Defining trained immunity and its role in 344 health and disease BCG-induced trained immunity: can it offer protection against 347 COVID-19? Susceptible supply limits the role of climate in the 349 early SARS-CoV-2 pandemic Influenza in Africa: Uncovering the Epidemiology of 352 a Long-Overlooked Disease What African Nations Are Teaching the West About Fighting the Coronavirus. The 354 New Yorker Leveraging Africa's preparedness towards the 357 next phase of the COVID-19 pandemic Africa's response to COVID-19 World Health Organization. WHO urges caution as countries in Africa ease lockdowns Optimal strategies for quarantine stopping in France General expected patterns of strategies focusing on contact between age groups Projecting social contact matrices in 152 countries using contact 368 surveys and demographic data Projecting the transmission dynamics of SARS-371 CoV-2 through the postpandemic period Projected early spread of COVID-19 in Africa 374 through 1 The impact of COVID-19 and strategies for 377 mitigation and suppression in low-and middle-income countries Madagascar can build stronger health 380 systems to fight plague and prevent the next epidemic Measles cases at highest point since 2006 as outbreaks continue to spread Seroprevalence of Antibodies to SARS-CoV-2 in Six 385 Sites in the United States Madagascar Ministry of Health. COVID-19 -Madagascar Johns Hopkins COVID-19 Testing Insights Initiative Variation in False-Negative Rate of Reverse 392 Transcriptase Polymerase Chain Reaction-Based SARS-CoV-2 Tests by Time Since 393 Prevalence of Asymptomatic SARS-CoV-2 Infection: A Narrative 395 Review Multiple Indicator Cluster Survey version 6 Estimating sources and sinks 399 of malaria parasites in Madagascar COVID-19 Community Mobility Reports A global map of travel time to cities to assess 404 inequalities in accessibility in 2015 Global, regional, and national age-sex-specific mortality 406 for 282 causes of death in 195 countries and territories, 1980-2017: a systematic analysis 407 for the Global Burden of Disease Study