Using a model that integrates epidemiological, health systems and health economic outcomes, we projected the health service and economic resources needed during an unmitigated COVID-19 epidemic in five cities in Africa and South Asia. We find that demand for ICU beds and healthcare workers will exceed current capacity by orders of magnitude, but the capacity gap for general hospital beds is narrower. Costs falling on both the healthcare sector and on households will also be large, and be close to normal annual healthcare expenditure due to all causes in all the cities except Johannesburg.

With optimistic assumptions about disease severity, the gap between demand and capacity for general hospital beds is closer in some, but not all the cities. If lengths of stay can be reduced and existing beds can be made available for COVID-19 patients, the gap may be small enough to be bridged by a combination of different mitigation measures that flatten the epidemic curve (and hence reduce peak demand for beds). The impact of such interventions on health outcomes in resource-constrained settings has been explored elsewhere¹⁹. Effective measures include physical distancing, shielding of vulnerable patients (such as older adults with comorbidities) most likely to require hospital admission and construction of field hospitals. Conversely, the shortfall in critical care capacity is likely too large to bridge within the available to prepare for the epidemic peak (and possibly even future waves).

Our modelling takes into account the demographic age profile of the different cities. Hence Nairobi and Addis Ababa are projected to have less severe health care demand shocks compared to the older populations in Johannesburg, Delhi and Karachi more susceptible to severe disease. However, all these cities contain large floating populations such as migrant, informal and casual workers who are not fully in census or even satellite imagery data used by WorldPop to construct city populations⁸. These populations may also bear the brunt of the economic cost of physical distancing and other epidemic mitigation measures. For instance, the lockdown in India (since March 25, extended to May 3) has caused a significant economic shock to millions of migrant and daily wage workers in the informal sector²⁰. Hence extremely difficult decisions are needed to trade off the costs of an unmitigated epidemic that we present with the costs of mitigation measures, with both costs falling on both healthcare systems and households.

Several studies have projected the number of cases, hospital admissions and ICU admissions that a COVID-19 epidemic would cause in countries around the world, including Africa and South Asia³⁴. Our findings move beyond these studies by (i) considering also the capacity of health systems in key cities to cope with this demand, (ii) considering the healthcare staff resources needed, since these fall short of demand by greater margins than hospital beds, and (iii) considering economic costs to health services and households. These costs include a variety of resources, and hence highlight the inadequacy of simply providing beds, oxygen tanks or ventilators without supporting equipment and staff. For example, costs associated with treatment of severe pneumonia include high-flow oxygen, pulse oximeters and health care staff with training in respiratory medicine in order to effectively recognise and manage hypoxemic COVID-19 patients before they reach a critical condition.

There are a number of important caveats to our findings. The first is that we are examining health care worker capacity considering only doctors, nursing and midwifery staff who are currently available. We do not consider the possibility that staff with less formal training might receive on-the-job training to manage COVID-19 patients. However, we also do not consider that health care workers may themselves be unavailable due to COVID-19 or responsibilities to family members with COVID-19, as has been reported in many countries facing COVID-19 epidemics²¹. Second, our capacity figures show resources available for all healthcare needs, but not all of these can be allocated to COVID-19 patients even if elective procedures are postponed for as long as possible. Indeed, most hospital beds in these cities are near full occupancy, and many health care workers work outside the hospitals. Further work needs to be done to further examine and understand real capacity constraints and the opportunity costs of transferring capacity to COVID-19, where the proportion of hospital admissions that are acute may be higher in high-income settings. Third, our cost results suggest that substantial additional financing is required. However, even where financing may be made available, we do not account for the absorption capacity to use this financing quickly to purchase commodities, construct new infrastructure, train new staff and institute substantial service delivery changes, which will challenge many LMICs, as it has done in high-income countries. Lastly, our economic costs assume that all symptomatic cases would be isolated for the full 7 days recommended by WHO. In practice, this recommendation may not be adhered to by patients with mild disease. Similarly, not all households will necessarily lose income while ill, since some ill individuals may continue to work.