As Australia starts relaxing its social-distancing restrictions, the epicentre of the COVID-19 pandemic is shifting rapidly to low and middle-income countries (LMIC).

LMIC governments operate in a far more challenging environment than our own. Not only do they have fragile health systems and limited access to essential medical equipment – there are fewer than 2,000 working ventilators combined in 41 African countries – they also lack the testing data needed for evidence-based policy responses.

The World Health Organization considers it critical for countries to prioritise diagnostic testing to track COVID-19 transmission, understand it, and suppress its spread.

But the scale of the testing challenge in LMICs is enormous. While Australia alone has conducted over 1.65 million COVID-19 tests since January 22, some developing countries have conducted only 3 percent as many tests on a per-capita basis.

In the absence of comprehensive testing, there is an alternative pathway to stem what has the potential to be a catastrophic wave of coronavirus infections throughout low-and-middle-income countries across the globe.

Together with international colleagues, we have developed a method to indirectly infer COVID-19 risk and fill this data gap in testing.

Our method relies on the insight that return migration is an important vector through which the virus has spread to LMICs. It means pre-existing bilateral migration links with COVID-affected countries can be informative about how the disease spreads.

The need for a data-driven and targeted approach is particularly important in LMICs, where strict lockdowns of the type Australia has employed risk creating a "hunger pandemic". Strict lockdowns could result in hundreds of millions of people struggling to survive as the COVID-19 pandemic exacerbates their already brittle social and economic situation amid poverty, existing conflicts, and the effects of climate change.

Most workers in LMICs are in the informal sector and live at the margins of poverty. Their governments lack the funds and resources vital to provide the kind of economic relief that has successfully allowed many Australian workers to retain their jobs.

Given these difficult constraints, perhaps the only feasible approach in these countries is to implement more limited lockdowns targeting regions where the virus is prevalent and insulating them from the rest of the country. The data we produce provides an opportunity to identify these hotspots and implement targeted lockdowns aimed to stop a dire outcome.

Our approach builds on the following idea: a country like Bangladesh, which had over 98,000 migrants who were living in Italy in 2017, was more exposed to COVID-19 than a comparable country with fewer migrants in Italy. Indeed, COVID-19 is now spreading rapidly in Bangladesh, which is fast becoming one of the worst affected LMICs.

These differences in pre-existing migration levels along with the virus intensity in destination countries allow us to infer how exposed an LMIC is to COVID-19. Indeed, at the national level, our predicted exposure measure is highly correlated with actual COVID-related confirmed cases and deaths.

But the key value of our analysis is its ability to guide regionally-targeted policy responses within countries. For instance, we have used administrative data on emigration in Bangladesh and the Philippines to identify COVID-exposed hotspots within each country.

To test our predicted exposure measure, we conducted a phone-survey of 909 households across one Bangladeshi district. We found that respondents in communities where a migrant returned in the two weeks prior are 242 per cent more likely to report COVID-19 symptoms.

Over time, our analysis will also be informative of sub-national economic impacts. As developed countries go through economic downturns, regions within LMICs with large emigrant populations will likely suffer substantial declines in remittance income. Identifying these regions will allow decision makers to provide targeted economic relief.

Undoubtably, having comprehensive testing data would be ideal for LMICs. But the lack of resources, administrative capacity, and the race against time mean that such data are unlikely to be available.

Our method provides an alternative way for LMICs to infer crucial information about COVID-19 hotspots within their countries. We hope that this serves as "proof of concept" of an approach that can be applied in other LMICs to implement evidence-based policies in the absence of adequate testing data.

Banner: Homeless people in Dhaka, Bangladesh, wait in a queue to receive aid during the nationwide lockdown imposed as a measure to prevent the spread of COVID-19, April 4, 2020/Shutterstock