What’s up with SA’s Covid-19 infection numbers?

There is something very peculiar about SA’s Covid-19 numbers.

From March 21 to March 24, there were 352 new cases, an average daily increase of 27.9.%. In the four days after that, the number of new cases increased on average by 21%. But in the four days ending on Wednesday April 1 there were 193 new cases, a daily average increase of 3.84%.

What on earth is going on?

The most likely answer is that the numbers are incorrect and that SA is missing many cases. Health minister Zweli Mkhize, for one, seems to realise this, as he warned on Wednesday: "The small growth we are currently experiencing is coming before a devastating storm".

By Wednesday night the number of Covid-19 cases had risen to 1,380.

Mkhize knows this increase — just 27 on the previous day — is low. Which is why he said the government’s plan is "to try to identify [what] the real situation is" by finding cases in poorer and undertested areas.

Ominously, he suggested, "the disease is spreading silently".

He explained that not enough poorer people were being tested. He was referring to people who had not travelled abroad and who had mild symptoms, particularly those who, "when sick … do not seek assistance immediately," he said.

Of the 44,202 tests conducted so far, only 6,000 have been done in the state sector. This suggests that at least 80% of all tests have been on middle-class people who can afford a private lab test and that not enough tests have been conducted on the 84% of state patients who are without medical aid.

Experts spoken to by the FM this week say SA shouldn’t think that the spread of infections have somehow halted. A more likely explanation is that we’re missing confirmed cases and that the reported numbers aren’t reflecting reality at this point.

Still, the number of infection is below the original predictions, which is good news. A team of experts had modelled an expected 4,000 cases by April 2 and an expected 7,000 without a current lockdown. We’re beating those projections.

However, the strange numbers make it much harder to predict future cases and to project the number of hospital beds that will be needed.

It also makes any diagnosis of the success of the lockdown tricky. If, for example, we see a marked increase in cases next week, it may lead people (incorrectly) to think the lockdown isn’t working.

In fact, the increase could be marked. For one thing, testing is becoming both more ubiquitous and more comprehensive. Mkhize says the government will begin adding a serology test (essentially, a blood test) to detect Covid-19 antibodies. This could indicate a previous infection, which can be misleading and lead to unnecessary quarantines, whereas the more time-consuming lab tests detect a current infection.

Also, the National Health Laboratory Service (NHLS) is re-purposing GeneXpert machines that have traditionally done tuberculosis testing in this country so they can be used to test for the coronavirus. And the results are available in 45 minutes.

Here, SA has an advantage: we have 180 of these GeneXpert machines, the highest number in the world.

The machine was designed in the US, but the test to check that the machine is correctly performing tuberculosis tests has a proudly South African connection. Let’s explain: to roll out testing and machines all over the country, one needs a test to check that the machine is giving correct results, rather than false negatives. So that test needs a test – a validation, in other words.

The simple, portable and cheap test to validate that a GeneXpert machine in far-flung Eastern Cape is giving correct TB results was designed by Johannesburg professor Lesley Scott. Scott and her team are now looking at a way to design a test to validate the GeneXpert results for Covid-19.

As soon as the validation test is ready (which Mkhizwe says will be early next week) it can be used. The NHLS has bought 60 mobile labs, which look like fancy minibus taxis, each to be kitted out with four small GeneXpert machines. The labs can drive to where communities live and test thousands of people a day.

But more tests mean more cases, which risks making the lockdown appear ineffective.

American nightmare

The consequences of not testing early enough have become nightmarishly clear in the US. At last count, there were 190,000 cases in that country, and 4,127 people have died from the virus.

But if SA is struggling to understand what’s happening with its numbers right now, statisticians in the US have been equally wrong-footed.

Each week infectious disease experts and epidemiology gurus from the top US universities have been asked by statisticians for their predictions on the number of Covid-19 cases in the country.

Two weeks ago, bio-statisticians at the University of Massachusetts, Thomas McAndrew and Nicholas Reich, asked infectious disease experts and epidemiology gurus from Ivy League US universities — including Harvard and Johns Hopkins — how many Americans they expected to be infected by March 29.

Their answer was 19,000, according to Nate Silver’s website FiveThirtyEight. In reality, that number was more than 136,000. The best in the business were wrong by many multiples.

This wasn’t a one-off either. On March 16, McAndrew asked the epidemiologists what their models suggested the numbers would be by March 23, and their average answer was 10,567. But the number of cases at that stage was more than 40,000.

It shows how the models being used to predict the spread of Covid-19 are often flawed. But this is through no fault of the researchers; there are, as some have said, just too many unknown unknowns to plug into models.

It may explain why US president Donald Trump, widely derided for his Covid-19 response, initially said life could return to normal by Easter (April 12). He later shifted the expectation to June.

The US’s problem, in the critical first few weeks, was a lack of testing. It’s a mistake that Mkhize doesn’t want to repeat, and this is why he’s adding the serology tests and harnessing the GeneXpert machinery.

While this is indisputably a good thing, there are unintended consequences to this that Mkhize will need to manage deftly: as the increased number of tests is likely to make confirmed cases soar in the next few weeks it could make it seem, incorrectly, as if the lockdown isn’t working. And South Africans, some of whom are already angry about the sacrifice, won’t want to believe they’re sitting at home for nothing.

But president Cyril Ramaphosa will have to stand firm if that happens. At the moment, the only way to control the disease is through the measures SA has implemented.