Five gaps in Covid projections
The World Health Organisation declared coronavirus a pandemic on March 11.
By June, the coronavirus disease (Covid-19) discovered in China had claimed over 475 000 lives from over 9 million cases.
The Malawi Ministry of Health’s Kuunika modeling projected the potential impact of Covid-19 in Malawi given several control measures, including a nationwide lockdown human rights defenders opposed.
The Kuunika report has not been shared publicly though Malawians desperately need to understand why and how Covid could kill 50 000 people as it projects.
The mathematical modelling captures the process by which Covid-19 cases and deaths data are generated in Malawi, but its projections remain far from the Malawi’s experience.
All models are wrong, but some are useful.
Here is why the Kuunika model leaves a lot to be desired.
(1) The projections poorly match observed deaths and hospitalised cases.
Between April 2 and June 10, the modellers projected over 30 000 deaths and over 100 000 hospitalised casesif there were no control measure. They also forecast about 2 000 deaths and 5 000 hospitalised cases under countrywide lockdown.
On the contrary, the country confirmed four deaths from 455 cases. Only six were hospitalised, zero needed intensive care and 55 recovered.
In fact, the nation may be said to be under no control scenario given widespread political gatherings.
Clearly, projections mismatch observed data beyond acceptable limits.
(2) The model does not acknowledge that asymptomatic individuals can recover without being mildly or critically infected.
The pandemic in Malawi will largely comprise asymptomatic and mild infections.
Without making a reasonable assumption about some individuals who may recover without showing signs and symptoms, the model not only violates Covid-19 transmission trends but also over-projects mortality and cases likely to be hospitalised.
(3) Covid-19 is projected to spread in rural and urban districts without the basis of social contacts.
The virus spreads through droplets during physical or close verbal contacts.
The Kuunika model does not acknowledge that the contact behaviour among Malawians is not random.
The degree to which children contact children is different from how adults contact adults or children contact adults in rural or urban settings.
This information gap gives rise to wrong projections of higher infections in highly populated districts though the intensity of social contacts may be low.
Covid-19 transmission depends on how frequent different ages contact each other.
(4) Besides, the continued importation of cases are excluded in this model.
However, most of the country’s confirmed cases are returnees from South Africa and other Covid-19 hotspots.
The model projects unrealistic control measures for Malawi. Instead of persistently projecting unfeasible scenarios like country-wide lockdown, Kuunika may focus on other important projections e.g. In a worst case scenario, which we are not currently experiencing anyway, a plausibility of locking down a district due to substantially high infections while other districts operate normally, shielding high-risk individuals, whether the current testing, case isolation and contact tracing is indeed working, and more importantly, projecting the size of Covid-19 pandemic based on locally available data, despite limited testing. Put it in Charles Babbage words “Errors using inadequate data are much less than those using no data at all.”
(5) The long-term projections of coronavirus transmission are unreliable.
Since the new disease is not well understood, several factors could influence its long-term spread.
The flu season may potentially increase the spread may reduce or increase infections.
The potential for a second wave transmission currently reported in Iran may result in resurgence due to international connectedness.
Kuunika provides insights into the dangers of Covid-19 despite its shortcomings.
The elderly and those with predisposing conditions are at high-risk of dying from Covid-19, so it is still vital for all to take precautionary measures, including washing hands, social distancing and case isolation.