As Covid-19 infections started spreading throughout the world many of us found ourselves anxiously scouring the web for information to help gauge our feelings and thoughts. Dashboards like the Johns Hopkins CSSE have given people a way to think about how the pandemic might spread in their home countries. The popularity of dashboards like this illustrates how people naturally consider numbers as evidence: if it is quantified, it must be true; if it contains decimals it must be accurate. For many, numbers are the closest authorities we can turn to during this period of uncertainty.

Covid-19 data is dangerous territory.

Differences between country findings illustrate that there is still much we do not know about the virus and how it plays out in different environments. In an effort to captivate readers or to be the first to share information with our social circles, numbers are at risk of being cherry-picked, worst-case scenarios are highlighted, and data presented without context or validation. As infections spread throughout the African continent, consumers and producers of data should be cautious to avoid these pitfalls.

A glancing read of differences in Covid-19 fatality rates highlights how important context is. Spain – one of the countries worst hit by the pandemic – has a case fatality rate of 10%. This rate is calculated as the ratio of deaths to total confirmed cases to date. On the other hand, Germany has a more manageable rate of 3%.

These figures are facts in their own right but can be misleading if presented without context:

• Spain’s focused testing of high-risk groups such as the old and hospitalised have biased detection towards those more likely to die. This inflates case fatality rates.
• Germany’s extensive testing has detected more cases outside of high-risk groups. This deflates case fatality rates.

The more testing we do, the less biased our information and more accurate our view of how lethal and prevalent this virus is. This will go far in understanding how lockdowns should be approached in Africa. Understanding this at sub-national levels may help policy-makers decide whether an area is ready for softening lockdown restrictions or at risk of breaching the capacity of the local medical system. Unfortunately, extensive testing is uncommon in Africa and few counties systematically release testing data.

Iceland – having tested nearly 12% of the population – has reliable data for estimating the number of true cases and suggests more than 50% of cases are likely asymptomatic. An analysis that applied Iceland’s testing findings to Sweden’s infection data suggested Sweden is likely to have 15 times as many cases as it has detected. However, this conclusion would assume that both countries have comparable demographics, that both borders are equally porous, and that both governments responded similarly. Drawing conclusions for Africa based on Iceland’s data would require at least as much cognisance of these and other assumptions.

As we prepare for the spread of the pandemic in Africa, we must remain pragmatic and honest in how we approach information, and cautious when drawing conclusions from comparisons. In a data-scarce environment and without conclusive evidence from the northern hemisphere, clarity around how this pandemic will evolve is yet to come and conclusions around what the best government response is are still developing.

Being irresponsible with data can have the same detrimental consequences as other mediums of fake news. We therefore have a responsibility to ourselves and our communities in how we engage with and communicate data. Three principles for both consumers and producers of data may help support this:

• Embrace uncertainty and the limitations of the science. This could mean not shying away from the terms “might” or “may” and guarding against letting beliefs pollute interpretation.
• Provide context and the tools needed for interpretation. Highlighting data sources, how key metrics such as cases are defined and should be read, and providing a view of both sides of a hypothesis are critical. This also requires an understanding of whether the data is relevant in an absolute or relative sense.
• Be front-footed in identifying and sharing limitations of the numbers. This relates to understanding assumptions, investigating whether sample sizes are representative and understanding that even governments and the health sector face challenges in collecting and distributing data.

Equipping ourselves and our communities for interpreting and sharing data will help us make better-informed decisions. A data-first mindset may be one of Africa’s best weapons for mitigating the forthcoming challenges of this pandemic. DM

Korstiaan Wapenaar is a  senior associate with the Digital Economy Team at Genesis Analytics. Genesis Analytics has developed a user-friendly dashboard to track the spread of the coronavirus in Africa. Please use this link to access the full dashboard.