2020, hindsight

Narrow interest groups are already pushing dishonest and revisionist histories of the current pandemic. We should not let them succeed.

Time in the pandemic is quicksand. Slow and fast. There are no days or weeks or months. It is the same day.

We forget things. We have forgotten things. This is a feature of how the human mind works, not a glitch; a function of decay and interference. It would be impossible to operate if we weren’t able to store important details, overwrite trivial memories or update others. But it’s an imperfect process.

We are born with muscle memories — reflexes — that share their origins with newborn tree-dwelling monkeys (our last common ancestor was six to seven million years ago). Infants who have never seen snakes or spiders still exhibit stress responses when shown pictures of the potentially deadly creatures. And yet we forget our own recent pasts with relative ease. We discard what is inconvenient, retain what is expedient.

In South Africa for example, the still-unresolved corruption of the Arms Deal, the fraud and deceit of Schabir Shaik have seamlessly been displaced by the actions of the Gupta family and the Gupta-aligned. And these news and corruption cycles have, in turn, conveniently covered over much older, still unresolved ones linked to decades of graft and profitable violence under the apartheid regime.

We remember what suits us and what serves us, and forget or discard what does not.

Sometimes our memories need a jog.

We only know what we know (March 2020)

We only know what we know.

At the start of the pandemic we knew very little. We are still learning about the novel coronavirus; about how it passes among us, about how to contain or control it, about how to treat people who become very ill, about why some people become ill and others do not, about the after-effects of infection and of illness.

The World Health Organisation declared a pandemic in March 2020. At that stage a little over 4,200 deaths worldwide had been confirmed as a result of the newly-named SARS-CoV-2. But that figure was about to change, rapidly, as the pandemic peaked in cities across Europe.

By the end of that month more than 900 people were dying each day in Italy; the same figure again in Spain. France was counting more than 1,000 Covid-related fatalities a day at the start of April. Similar numbers were reported in Britain, and again in America — where the numbers inched up until they were approaching 1,500 deaths a day in that country. News reports from New York described the relentless sound of ambulance sirens, 24 hours a day. There were reports of emergency rooms overwhelmed, photographs of doctors with the outline of masks branded onto the skin on their faces. There were images of refrigerated trucks being set up as temporary morgues to accommodate the sudden mass of excess bodies.

Even in these rich, well-resourced countries, it was clear medical facilities were insufficient: they did not have enough beds, not enough ventilators, not enough masks and PPE for the medical staff who, quickly, became sick too. The fatalities were growing so quickly that people started to represent trends using logarithmic graphs [not shown below] — which show exponential rather than linear growth.

Months later, these memories already seem distant. We have already forgotten how bad it was.

This graph below, sourced from Our World in Data, serves as a reminder as to what the world looked like at the time: a tidal wave of deaths.

And this was the context in which the decision was made, in South Africa, to impose and maintain a strict lockdown.

What we knew, then, was that there was a highly infectious virus that appeared to have an extremely high fatality rate in Europe, North America, and Britain (higher even than it had been reported in China, where the outbreak had started). There was, as yet, no data on how this same disease would proceed to infect and kill a population already heavily burdened by HIV and TB — both specifically and jointly problematic, in terms of lowered immune response and damaged respiratory function — as well as other chronic conditions like diabetes and hypertension. We did not know how this pandemic would affect countries where large sections of the population were poor and chronically malnourished, and where there were most certainly not enough beds and ventilators for everyone.

In March when the South African president signed the lockdown measures that would be implemented from the 27th of that month, the initial projections — based on the available data (from countries like Italy, China, and the US etc) — were that between 88,000 and 351,000 South Africans could die of Covid-19, depending on the level of the state’s response. [Author’s note: this wasn’t an entirely abstract exercise for South African epidemiologists. As this article in GroundUp notes, during the peak of the HIV-Aids crisis, as many as 770 people were dying a day. It is also worth pointing out that the ‘known uncertainty’ behind these figures (evident in the extremely wide range) and the potentially extremely high stakes meant that we needed to make sure the worst case didn’t happen. The greater the uncertainty, the more careful we needed to be.]

These initial figures were soon amended to a smaller range, between 120,000 and 150,000 deaths; and by 21 April, revised downwards again to around 45,000 expected deaths. I.e. Within three weeks of starting the lockdown, the state’s models were working off the basis of 45,000 possible deaths.

What do we know or learn from this? That the initial estimate or model outcome was changed because it was based on limited information — which was updated and amended once newer and better information became available. This is how estimates and modelling work, and there is nothing inherently ominous about a projection being updated or revised. I.e. it’s not an exclusive feature of dishonesty (although of course it can be used as such), but more a failure of trying to know what we do not yet know, and doing this using limited information. It’s a bit like trying to draw a lion when you have never seen one in real life.

[Segue] We only know what we know — Part Two (Who ‘knows’ anyway?)

[The initial] Lockdown was not emotionally or physically or fiscally comfortable. The poor bore the brunt, but the middle class complained the loudest.

Between April and May there were a number of public squabbles between South African public health experts, and these were shamelessly incited and exploited by the media perhaps in the hope of generating a few extra clicks. This also exposed the limitations of knowledge and experience in our own media corps: people who could not spot basic errors in health reporting, who treated disease numbers as if they were inflation rates; and who, significantly, chose to prominently and repeatedly include dissonant voices purely because they disagreed with government policy, and not because they actually had any new insight or relevant expertise to add to the issue.

It is important that people are allowed to criticise the government. And there is usually no harm in people being allowed to have or air their opinions, whether or not they are informed on a matter [the exceptions to this of course are expressions that fall under hate speech, which may incite harm, or which are malicious and false etc]. But there is a problem when commercial media positions these ‘strongly held opinions’ as having some sort of professional weight in public health decisions, even when the opinion-holder has no relevant qualifications. This is a common example of false equivalence: where non-experts and interest groups are positioned as equivalent to actual experts. It is irresponsible, and often, in the case of public health outcomes, potentially dangerous.

There is a growing tendency in some circles to deride the notion of ‘experts’ and ‘authority’, as if bowing down to someone’s superior knowledge is a sign of weakness and submission, rather than acceptance that some people may know more about a particular subject than you do.

An expert is not only someone who knows a lot about a particular subject, but also a person who has been assessed — either by a group of (expert) peers, or an industry body — as meeting a set of minimum standards required in order to be a qualified and credible expert in that field. Many professions require not only the completion of a university degree or equivalent, but also entry or admission or board exams to be certified in specific fields. I am sure nobody reading this would want to go to a dentist who had not completed dental school and who had not been admitted and certified by the Health Professions Council of South Africa (or whatever other statutory body exists in your region). Expertise also exists outside of conventional Western institutions and learning, but still requires a form of community-based ‘peer review’ in order to be considered credible and to have earned expertise. For example, one cannot simply declare oneself to be a herbalist, or isangoma without having proceeded through various training and initiation and acceptance protocols.

There is a marked tendency in some quarters (perhaps it was always there, just less publicly shared) to deride ‘experts’ on the one hand, while also assuming that being ‘clever’ or having expertise in one area of life qualifies that same person to have confident positions on all other areas of life.

Two hundred years ago, it was common for experts to be sort of polymaths: Darwin was a biologist and a geologist for example. These days this is not impossible, but it is less common. This isn’t a limitation but an acknowledgement: an accountant and a physicist who both study numbers, study very different numbers. Understanding economics, or how economies work, does not mean you can simply apply the same metrics to demography, criminology, or public health. And this is why each discipline has a specialised learning path and even degree outcome in some cases. No data exists without context (even physics assumes this in the form of physical constants), and without understanding the context you may make or compound qualitative or quantitative errors, which will affect and reduce the integrity or validity or relevance of your outcomes.

The circumstances of the Covid-19 pandemic involve and require particular types of expertise to make sense of what was happening: not just medical doctors, but more specifically doctors and scientists who have specific expertise in immunology, virology, public health, epidemiology, infectious diseases, bioinformatics, demography, disease modelling, and so on. South Africa has quite a high concentration in all of these, because of our history with HIV and TB. So: we are considered global experts, although this was at a high and terrible cost.

The nature of humans, of the internet and social media, and the nature of the pandemic, meant that pretty much from the start, a great and growing number of new non-experts attempted to wade in with unfounded but confident opinions about often quite technical data or concepts or processes: whether this was understanding the infection rate, the mortality rate, the impact of sunshine or BcG vaccines… When I observed this phenomenon in the early pandemic weeks, I noticed first that people’s ability to read and draw graphs was markedly poor, and that this poverty of understanding was easily exploited by people who wanted to drive specific agendas. This has not changed much and is something of a global problem. The nice people at The Conversation have published some advice on how to read graphs.

From my own perspective, although I studied science and human biology as an undergraduate and have reported on health and public health in South Africa for over two decades (I also teach data literacy, including how to read graphs and how to read the context of graphs or data visualisations), I know that I am not an expert on most of the topics around Covid-19. The skill I do have is how to find, contact and engage with people who are vastly better informed than me, and who are also accredited experts in their respective fields.

I feel no shame, nor inhibition, in saying that I do not know many things, and I even enjoy seeking knowledge or information from others who do know about those things. I do not personally understand why a series of (mostly men) lawyers, economists, actuaries, designers, etc, feel repeatedly entitled to go head to head with public health experts — or critique graphs and data they plainly do not understand — as if they were the dean of the faculty. Where does such undeserved confidence come from? This is not to say professions or professionals must stick to their silos, but rather: they should acknowledge and take into account their own lack of expertise, because it is as important a factor as missing numbers and data. We only know what we know.

The point is, acknowledging that there are experts in different and specialised fields does not make you a ‘wuss’ or a submissive sheep. It makes you smart.

We only know what we know — Part Three (What changed about what we know about Covid-19?)

Why did the South African state update its fatality models downward in April? Why were we told not to wear masks, and then told we had to wear masks? There are many small and large inconsistencies in the short story that is Covid-19 so far. This is because we are learning new things all the time, and also because we have made mistakes as we learned — as individuals, and as states. Also, running a government and a country is not like running a household. It requires usually large amounts of bureaucracy, which are typically less efficient and also provide frequent opportunities for error and oversight (deliberate or accidental).

In the very early days, civilians were told it was not necessary to wear surgical masks. This is largely because, at that time, there was a global shortage of medical-grade protective equipment (and in fact medical equipment) needed to protect frontline workers, most specifically healthcare workers, and — as many might recall — most such items were indeed stocked out in South Africa for several weeks.

At that stage the benefit of homemade cloth masks had not been widely proven or agreed. But it was very clear that frontline workers were at high risk, and that they needed the PPE.

In a fairly short period of time, evidence emerged that almost any form of face covering could play a role in reducing transmission of Covid-19, after which many countries including South Africa began recommending, and eventually regulated, wearing of cloth masks in all public places. Over time, the supply chain issues relating to basic PPE also resolved, making masks of all types more widely available.

This is not purely a case of the ‘authorities’ blowing hot and cold: it was mostly a case of initial shortages of PPE (and a real and demonstrated concern that middle class shoppers and countries would buy up the world’s stock, leaving those most in need without!), and subsequent better information that informed policy changes. I have a vivid memory of going grocery shopping in early lockdown and seeing a man wearing a [then as rare as gold and almost as expensive] disposable N-95 mask sideways with the nose section over his cheek, and a 2cm gap between his face and the mask, making the entire thing not just ineffective but potentially a hazard to others. He might as well not have been wearing a mask; and, because the mask he wore was single-use/disposable, he essentially wasted the item. Regulating this kind of confident foolishness remains a challenge.

In the South African context, there were many other peripheral regulations that altered the nature of how people experienced or perceived the lockdown and perceived the ‘legitimacy’ of the lockdown — specifically the prohibition against purchasing alcohol or tobacco; and, in the early weeks, the closure of retail shops and so on. I suppose this is also human nature, that South Africans could obsess about not being able to buy open-toed shoes while people were dying. Yes there were inconsistencies and certainly errors, and many were extremely petty and not medically or socially necessary. And yet, overall, our policy achieved critically important outcomes (see next section). My personal point of view is that if you judge the merits of lockdown by whether or not you could buy sandals, then you are a fool.

One other very important change to ‘what we knew’ that took place between March and June/July, and which is relevant to this discussion, was the massively improved treatment protocols (and new/approved drug regimens including the use of dexamethasone and Remdesivir), which have seen Covid-19 case fatalities drop significantly in many countries around the world. So, even as we see an increase in infections in many regions, this is not being accompanied by as high a number of deaths as previously seen i.e. the outcomes for those patients who become ill is significantly better now than it was five months ago.

This is critical for understanding the South African context too. By delaying the peak of infections in this country, we almost certainly managed to greatly reduce the number of deaths that would have occurred. (This is a difficult claim to make because, as before, we only know what we know. We will never know how many lives were saved through these interventions, we can only try and count the lives that were lost. We cannot prove a negative, what might happen if we had not taken the actions that we did.)

We only know what we know — Part Four (Did lockdown work?)

At the start of South Africa’s lockdown, the buzz-phrase was ‘flattening the curve’. The idea behind this was, specifically and overtly, to delay new infections so that the disease peak would happen at a later stage. At the time, this was not just necessary but essential because South Africa (like many other countries) needed time to set up and resource its healthcare infrastructure, to cope with what was expected to be a wave of seriously ill and dying patients that would need not only hospital beds but special care, ventilators, oxygen, and so on. ‘Flattening the curve’ would buy us time.

What happened in real terms is that flattening the curve also appears to have brought us lives. Because, by delaying the disease’s peak, when we did hit peak infections the medical response to Covid-19 infections was exponentially better — as were patient outcomes — than it had been only weeks before (as above). We couldn’t have known this at the time, but we still benefited from this.

In mid-May, government projections were estimating that, depending on the impact of and compliance with lockdown regulations, the Covid-19 active cases disease peak could be pushed out from [what was projected to be] early June, without intervention, to either mid-July (pessimistic intervention view) or mid-late August (optimistic intervention view).

It’s interesting to view the graph (taken from an NICD presentation on ‘Estimating cases for COVID-19 in South Africa’ shared on 19 May) to see how this fits with what we now know, which is that our active cases peaked late July to early August.

‘Active cases’ is a tricky metric though, because confirmed/active cases only show us what we test for. If we don’t test, we don’t see more cases. And we already know that testing capacity was limited long before our ‘peak’ because of limited tests and laboratory processing facilities. However, when we look at weekly deaths in South Africa, data provided by the South African Medical Research Council, below, clearly shows a notable peak in deaths around 15 July, which more clearly focuses our timeline.

Graph taken from SAMRC Report on weekly deaths in South Africa.

Reading the available data and models provided by the NICD and MRC and others, what we can say we ‘know’ is that a projected ‘no intervention’ disease peak of early June was pushed out to at least mid-July, if we go by peak deaths.

This in itself would already suggest that the lockdown restrictions worked, as much as they pushed the curve out by several critical weeks. This can also be confirmed by looking at the length of time that elapsed between the country’s first death, in late March, and the arrival of the peak, in mid-July.

Another way we can measure the beneficial impact of South Africa’s lockdown is by looking at the progression of the disease in other countries in the global south (with health profiles more similar to our own), and who experienced their first confirmed cases at around the same time as we did. Here I look at Brazil and Mexico, which confirmed their first cases at the end of February; and Argentina and Peru which, like us, confirmed their first cases in the first few days of March. The graph (below) is again provided by Our World in Data, which is based at the Oxford Martin Programme on Global Development at the University of Oxford and provides good quality and transparent data about the pandemic.

What the graph shows is that South Africa’s Covid-19 rise and peak (in deaths which, as I explained, provide perhaps a more concrete measure than only infections) began to happen at a much later stage than in Brazil, Mexico, or Peru — again, this is critical because the later peak meant much better patient outcomes, and a lower or flatter peak (than without intervention).

Argentina’s trajectory shows a similar slope to ours. Perhaps it is worth pointing out that Argentina also imposed a strict lockdown on its population, and that the lockdown was extended for several weeks, and that many people in the country accepted the lockdown and many others complained. Also worth noting is that while many finance publications are now writing (with a sense of glee?) about how lockdown has ‘ruined’ the country’s already fragile economy, others note — with seeming obliviousness — that the country delayed the virus peak by something approaching four months. Surely a delay of the virus’s peak is exactly what we were all aiming for?

This prompts the repeated debate: if we say a lockdown isn’t working, the question is: for who?

Peru of course also imposed a lockdown, but saw rapid infections and growing deaths in spite of this. As a result it has become the occasional poster child for the ‘Swedish model’ defendants, who wish to prove lockdown is an unmitigated disaster. However, if any of them bothered to read a little about the situation in Peru, they would learn that the reasons for the country’s lockdown failure were largely because it failed to lock down the open markets where most of the country did their grocery shopping. I.e. The lockdown was too porous to have been effective.

The people who think they know what they don’t know

It is easy to make the mistake (and I include myself here) of confusing newspaper headlines and social media outrage as representative of the broader world. But media and social media ‘storms’ can have real-world consequences. Real or manipulated social media outrage is regularly picked up on by journalists as being news-worthy. And, because the media presents certain issues as important, people eventually perceive them as important. When the media repeatedly presents certain unsupported or less-credible voices or viewpoints as credible, then these become normalised and even accepted as credible even if they are not (or should not be).

One example of this is the extant narrative around climate change. There is unambiguous scientific consensus that climate change is: real, man-made, and happening super-fast. Climate change deniers and climate rationalisers have been positioned as a sort of important ‘dissenting voice’, the other side in an issue that has a ‘both sides’ requirement as if it were marriage counselling. But this is not the case. There is no dissent (not among people who are qualified as experts to have scientific opinions on the matter). The climate change deniers are a fringe at best, and they are a minority with dangerous and permanent consequences for the rest of us. The news media (my own community) has spent the better part of the last few decades playing at ‘being impartial’ about the issue, which has resulted in it frequently giving equal weight and space to discredited climate deniers and minimisers. This in turn has allowed these same deniers to demand media space, and enabled governments and corporations to continue with policies and practices that are wreaking permanent damage on our environment because ‘nobody really knows’ (but we do know).

The strength or persistence of a group’s opinion or the loudness of their voice on social media should not be confused for its relevance or credibility. Not every story has two sides. Some have one. Others, multitudes.

There have been a number of groups of people, particularly on social media, who have made very loud complaints about lockdown and about the government’s handling of its pandemic response. I will repeat myself here, and say that criticism of the government is a good and healthy thing. But it should not be confused with ‘insight’ unless it comes from credible sources.

On social media in particular, a certain prominence has been given to clusters or pods of predominantly white males, with expertise in mostly non-medical areas of ‘business’ (because being a CEO qualifies you to do anything apparently). The fact that middle class white males have poorly informed and self-important opinions is nothing new on this planet, but the fact that many in the media have confused these same men as Covid-19 experts is troubling.

One particularly loud group, made up of mostly white males, have have named themselves after the national animal of China; a choice made more explicitly ironic because the group’s leader posts Sinophobic tweets. The group itself comprises people who describe themselves as investors and actuaries and economists. There are also two with medical degrees listed as members of the team as well as a geophysicist and a statistician. Four of the ‘team’ described on the group’s website are women. Only one of the ‘team’ is a person of colour. Not letting accuracy stand in their way, they have, on their own website, chosen an image of a black woman in a mask to represent them.

This is the main image on the group’s ‘about us’ page. None of the team members are, as far as I can tell, black women.

What this group has done is regularly publish and promote statements criticising the government’s modelling and the lockdown, and ridiculing the government’s projections while promoting their own allegedly superior product and insight, and using the latter to call for an end to lockdowns everywhere. Except their loudest claims are repeatedly incorrect.

They repeatedly claimed there would be no more than 10,000 deaths due to Covid-19 in South Africa. That figure has already been substantially exceeded

In May, the group wrote that they were “left wondering why anybody in their right minds would be talking up a story that involves anything more than 10,000 deaths for South Africa, with or without lockdown.”

Another document published on their website on 11 May states as follows:

Applying our age-based fatality and population estimates from above, South Africa should expect one third of Spain’s rate of deaths, or 200 per million. This would imply around 12,000 deaths, and this works off one of the worst country experiences in the world. Performing the same calculation on ‘no-lockdown-Sweden’, also past its peak, we would estimate fewer than 8,000 deaths for South Africa. Using past-peak Iran as a proxy we arrive at 5,000. This adds further weight to our suspicion that projected deaths are still being wildly overestimated. Combined with an observation that no country’s epidemic has ever manifested as exponential, we tentatively surmise that the attack rates employed in these models are way too high and that the entire model class being applied is inappropriate. If we are correct, then our analysis of lockdown’s ratio of harms to benefits will move from overwhelming to infinite.

In early June, this group doubled down on their claim of 10,000 deaths, saying “we are challenged to understand how any model for South Africa could reasonably produce a death forecast of more than 10,000”.

In late July, the group leader once again cited their figure of 10,000 [deaths], and said ‘we haven’t had to change it [the figure]’.

At some point (records on their own website point to some time in July at the very earliest but this was not consistent with their other statements as above), the group quietly updated their estimated mortality figures to ‘between 10,000 and 20,00 deaths’, and have now started retrospectively claiming that they had claimed this all along.

One could call it revisionist, but I prefer: dishonest. Note the date this tweet was posted, below.

The group also derided the government’s model, saying that ‘for the 40,000 [sic] figure to be accurate, it would entail “age-based mortality that is 12 times worse than has been observed anywhere in the world”.

On the group’s own website (click to donate), they stated again that the ‘the model that [the government] figure is based on has now been shown to be massively divorced from reality because far fewer actual deaths have occurred than the model predicts.’

Reader: As of writing, there have already been over 15,000 official deaths attributed to Covid-19. When excess deaths are taken into account (this issue is discussed below), the figure is incrementally higher and indeed fits the government’s model.

Excess deaths

The latest Report on Weekly Deaths in South Africa from the Medical Research Council indicates over 42,000 excess deaths in South Africa between early May and the start of September. The profile of the pattern of these deaths (i.e. their peak and wane) very closely matches that of our Covid-19 infections, and strongly suggests that there is a direct link between the two, which implies that many of the excess deaths are probably Covid-19 related even if this was not officially captured in the cause of death form. This is discussed again, later in this text.

The ‘strong opinion’ group also made a number of incorrect projections about when they thought the peak would happen. The group’s leader (a ‘professional investor’ with no experience in infectious disease or medicine) stated confidently that ‘his research demonstrated that of 66 countries worldwide whose infection curves are believed to have already peaked, 56 peaked within 80 days of the first infection in that country’ and that ‘[…] To date, no nation we have noticed has seen peak daily deaths outside the range of 30–50 days from first death.” [itals and bold mine]

What do the facts tell us? South Africa’s first recorded Covid-19 death was on 27 March. The peak number of reported daily deaths was reached more than 100 days later. This alone should be seen as an indication that the lockdown was indeed effective — not at stopping the disease, but pushing out the peak to a later time, when we would be better equipped to cope with it. It also, again, shows that the premise on which the cherry-pickers based their projections were not, in fact, correct (nor did they issue any correction, they simply revise and hope no-one will notice?)

Because South Africa’s death figures provide such a strong rebuttal to this group’s claims, the group of non-experts have now tried to argue that the fault lies with how deaths and causes of deaths are recorded — as if the limitations that exist during Covid-19 are not, always, part and parcel of the bureaucracy of such things — and suggested that, among other things, deaths due to comorbidities (for example having HIV) are being ‘misleadingly’ recorded as Covid-19 deaths. They also appear not to understand how mortality data and collection typically works, and seem to confuse this lack of understanding with hints that there is some squelchy numbers cover-up being manifested by all the terrible professional demographers and medical professionals who have worked with state and population data for the last two decades.

Questions about mortality data are actually good and useful questions, when they are not framed as accusations, and incredibly South Africa has done rather a large amount of work on exactly this issue — because we faced and face similar problems because of the HIV epidemic, and widespread TB. The short answer is: this ‘problem’ is not a new problem, and has been quite robustly answered by specialists (experts! Remember those!) working in relevant areas. In fact, specialists are quite good at explaining, you know, the things they actually know about!

American virologist Dr Angie Rasmussusen explains the issue here: ‘Just because AIDS patients die from secondary pneumonias and cancer does not mean that HIV infection didn’t cause their death […]’

Something else that needs to be added: if you watch a lot of Hollywood movies, you may think everyone who dies gets autopsied. You may be relieved or disappointed to learn this is not the case, sometimes not even if a person died in unnatural circumstances. This is because it is expensive and also invasive to cut everyone up. In many situations, if there is a medical history that strongly indicates and agrees with how a patient died, the cause of death may be written (by a qualified doctor — sadly they do not allow ‘people who read an article’ to make the determination) without invasive procedures being required. This is also standard practice outside of South Africa, again because it is expensive and invasive. So while we assume most of the officially recorded Covid-19 deaths on the nightly figures released by the government are based on fatalities that took place where the deceased had tested positive for Covid-19, it is not only likely but probable there are cases where people presented symptoms and complications that fit the profile of Covid-19 but who died before they could be tested — and where a doctor may determine Covid-19 as the cause of death.

Dr Saadiq Kariem, the Western Cape provincial health department chief of operations, explained as much in a news report, saying: “[…] someone who, for example, comes in and dies of an acute respiratory illness and where that is the confirmed diagnosis or clinical suspicion, then that we would say would be the cause of death due to COVID-19.”

Or, as is equally the case, a doctor may not list Covid-19, either because a test was not performed or the Covid-19 status was unknown or because a medical history of the deceased was not available (that this takes place quite often is strongly suggested by the national excess death profile compared with official Covid-19 deaths), or because the person writing up the cause of death is a human and may be subject to all sorts of other unknown stresses and pressures including a sudden wave of excess deaths putting pressure on a system that was already under-resourced.

I need to emphasise that none of these practices is remotely new or surprising to anyone who has ever worked with mortality data. Any suggestion that it is part of a deliberate Covid-19 cover-up is merely demonstrating that the person doesn’t know the subject matter well enough.

A related issue is that the ’cause of death’ for individuals is not made available at the time of death. This process, as Professor Tom Moultrie has repeatedly explained, can take up to two years (there are also a number of good reasons why this data is delayed, because it must be anonymised; can you imagine if, for example, an individual’s HIV status became public post mortem?) The Covid-19 mortality information released by the Department of Health and the NICD is not from the cause of death form, but from a parallel surveillance system that requires each confirmed Covid-19 death to be reported. GroundUp has written an excellent explainer on how the system currently works.

Again, this is not a new issue. It is one researchers have struggled with for years, and we would love to see more regular data being released more regularly. But, for now, it is not available — and its unavailability is also not a sign of a grand conspiracy theory, but simply the way demographic statistics work in South Africa. If you have a problem with it, join the queue.

The jolly misinformers are also participating in a concerted effort to discredit the excess mortality data, which is the only way their own model could still be considered viable, and have started claiming that the majority of excess deaths (which, whatever their cause, most certainly exist and need to be explained one way or another) are not due to Covid-19, but are the result of something else — such as people not getting medical treatment or chronic medication because of the real demon in the room: lockdown.

The problem with this is (and once again it betrays a real limitation in their ability to understand disease profiles — although it is amusing to see these individuals give a shit about HIV and TB for the first time in their lives) the evidence doesn’t support this to any large extent.

The initial lockdown period may indeed have a disruptive impact on some people seeking and receiving chronic medical care (even though such care was allowed at all levels of lockdown), possibly because people were afraid of going to hospitals or clinics for fear of getting infected, or because transport options may have been limited. A recent report by journalist Pontsho Pilane indicates that as much as 15% of patients are battling to access chronic medication as a result of the pandemic (note, this is a broader and more inclusive problem than ‘the lockdown’), and that HIV testing fell by some 57% during the lockdown period, compared to the same period in 2019. However, another study of clinic visits in rural KZN indicates that visits for adult healthcare (including for HIV care) remained remarkably robust during lockdown, although child health visits showed a decline during the same period. This would impact future health outcomes for children (especially vaccinations), but not immediate mortality.

The short answer to this is that there is no doubt the pandemic is disrupting access to essential chronic care; but we don’t yet know to what extent, nor what the effect of this will be. What we can tell is that, based on the types of conditions affected, it is most likely going to have a medium- to long-term negative impact on health outcomes but would not be responsible for a sudden and immediate spike in deaths.

Procedures such as cancer treatments and many elective surgeries were also postponed during lockdown — the former because certain cancer therapies make the patients extremely immune-compromised, meaning they would be unable to fight off any possible Covid-19 infection; the latter because a) hospital beds were being reserved for expected critical Covid-19 patients and b) there was a genuine higher risk of exposure to Covid-19 at a hospital. Once again, these postponements could indeed have negative outcomes for the wellbeing of the patients, but there is no data as yet that indicates this would somehow be responsible for a sudden and large short-time spike in mortality.

Without any evidence to support this, this particular group of actuaries and investors bullishly argues that these chronic factors and not Covid-19 contributed to the rapid deaths of a number of people. This is despite the fact that (as Prof. Moultrie has pointed out, referring to the work of his colleagues who specialise in these diseases), death from previously well-managed TB or HIV, even when a patient stopped taking medication for a couple of weeks, would take a substantially longer period to manifest. Again, we actually have a rich and extensive body of research and bodies of experts on these diseases. It is a pity the actuaries never saw fit to consult them.

And, as I and others have already mentioned, there is this quite noticeable consistency between the excess deaths pattern and our Covid-19 infections patterns (scroll down to see the graphs that demonstrate this relationship).

As a quick (re) explainer:

The Medical Research Council explains that the term ‘excess deaths’ is ‘used in epidemiology and public health to measure the mortality impact of a crisis when not all causes of death are known.’

The World Health Organization defines ‘excess mortality’ as: ‘Mortality above what would be expected based on the non-crisis mortality rate in the population of interest. Excess mortality is thus mortality that is attributable to the crisis conditions. It can be expressed as a rate (the difference between observed and non-crisis mortality rates), or as a total number of excess deaths.’

For the last few months, the South African Medical Research Council has published regular reports on mortality in South Africa, which show reported deaths from all causes and allow us to also evaluate excess deaths, as above.

Within these reports, the figures distinguish ‘natural’ and ‘unnatural’ deaths (or ‘all causes’ when both are included): natural deaths mean deaths from natural causes. Heart attacks and disease, including Covid-19, are considered ‘natural’ causes. Traffic and other fatal accidents, suicides, homicides, drownings and so on are considered unnatural. These deaths are counted together (to give a total) and separately, so that we can observe each type of death separately.

Unnatural deaths: What we observed during lockdown was that, under the early (strict) lockdown, unnatural deaths showed noticeable decreases [when compared to a ‘maximum’, ‘minimum’ and ‘median’ baseline that was determined using mortality data from previous years]. The decrease in unnatural deaths was due to restrictions in terms of movement (traffic accidents are responsible for a lot of deaths), but also restricted access to alcohol, and the imposition of a curfew, which also reduced fatal interpersonal violence.

What the Medical Research Council found was that, after an increase that peaked around mid-July (for most provinces), deaths from natural causes turned [downard] during the week 21 -28 July.

National natural (excess) deaths and reported Covid-19 deaths
Western Cape natural (excess) deaths and reported Covid-19 deaths
Gauteng natural (excess) deaths and reported Covid-19 deaths

The same group refuse(s) to acknowledge the association: On 23 July, the ‘leader’ of the smug dissenters posted on Twitter: ‘When Covid declines, as it does everywhere, and these deaths persist, there will be a reckoning.’

Reader: the deaths have not persisted. And perhaps there should still be a reckoning, just not the one he thinks.

A note on flu deaths and why Covid-19 is not just like flu:

The NICD has previously stated that South Africa usually reports between 7,000 and 12,000 seasonal influenza-associated deaths each year. The Department of Health’s 2017–2021 National Influenza Policy and Strategic Plan writes that ‘it is estimated that nearly 10 000 deaths’ are due to influenza in South Africa each year. What is important to note is that these figures are estimates and they are constructed or based on modelling using laboratory surveillance data, like positive samples, together with estimations of excess mortality during influenza season (which is done by ‘subtracting a seasonally varying background rate of non‐influenza deaths from observed death rates’ [Gul, et al., 2018]). Estimated flu deaths does not represent exactly 10,000 individual cases of influenza confirmed with separate lab tests. This, again, is standard practice in studying disease prevalence and mortality in populations. (If you think this is unusual, go and see how we estimate HIV prevalence).

Based on current available data about both confirmed Covid-19 case fatalities and the strongly suggestive excess mortality data, it should be quite evident that Covid-19 is more than ‘just flu’, and that the novel coronavirus — for which there is as yet no vaccine — has caused at least four times the number of deaths annual flu strains (for which there are available vaccines) are associated with in South Africa. (More information can also be found on the WHO’s FluNet).

In addition this year it appears that, because of lockdown and the closure of international borders during the pandemic, South Africa has not had a flu season for the first time in nearly 40 years.

This information has been integrated into the weekly mortality models presented by the South African Medical Research Council — i.e. They have been adjusted to take into account the fact that the x-many thousand deaths expected during influenza season (South Africa’s winter months) have not happened, as well as the fact that other natural deaths also tracked below expected levels in most provinces (possibly owing to fewer other communicable diseases) after lockdown and before the emergence of the excess deaths.

Fin: Whose lives matter?

There is something about certain forms of anti-lockdown performative pseudo-activism that I find personally repulsive, beyond the bad science and the smug (and mistaken) confidence in their data. And this is because of what it reveals about the assumptions that underpin certain calls to end lockdown.

In May, the group of mostly white male actuaries I mentioned previously told Business Day that if the lockdown was not lifted it would ‘lead to 29 times more lives lost than the harm it seeks to prevent from Covid-19 in SA’.

To reach this astounding claim, they had calculated the estimated life-years that would be lost due to Covid-19 deaths using age mortality data from New York and overlaying this with South African age profiles to see who they predicted would die, and how long they would have lived without Covid-19. So for example if you were hypothetically 75 when you died of hypothetical Covid-19 but hypothetically would have lived to 80, you lost 5 years. They then compared this with the estimated life-years they estimated South Africans would lose from the estimated poverty that they estimated would result from the restrictions placed by lockdown. Based on this rather abstract process, these actuaries had ‘worked out’ that, because of the lockdown, ‘10% of South Africans will become poorer, and as a result, will lose a few months of their lives.’

Correction: in an earlier version of this piece, I assumed the life years lost in this scenario was calculated using the group’s projected mortality data of 10k deaths. However the ratio was calculated using a much higher (no-lockdown scenario) upper mortality figure of 88,000 deaths, which was the ‘worst case’ scenario they calculated. However, the rest of my original comment (in bold) remains valid: These actuaries’ calculations also only seem to consider fatalities, and ignore the equally large if not larger number of people who became seriously ill but did not die, or those who became relatively ill and continue to suffer mild but debilitating long-term effects often described as ‘long covid’, or the data that shows even people who became mildly ill may suffer long-term and even permanent effects such as damage to the heart, etc. The cost of contracting Covid-19 is not simply ‘death or absolutely fine thank you’. i.e. Their comparison ratio (deliberately?) only considered life years lost due to adverse economic impacts from lockdown but failed to qualify the significant and broad adverse economic impacts of the pandemic – which have been seen even in countries that have had much more relaxed lockdowns. So: the inputs for the one scenario are imbalanced, and incomplete. In addition, I have to note that one of the reasons their ‘life years lost’ to Covid-19 group is so much smaller (in years), even at the higher end of the mortality spectrum, is because the actuary group estimated most deaths would be elderly people and of course they have fewer ‘life years’ to lose and so are less ‘valuable’ in the group’s estimations.

On paper, the headline-making claim that lockdown would cause a cumulative loss of life 29 times worse than what would be caused by the virus alone makes their call to end lockdown sound almost noble, altruistic: if we don’t let a bunch of people die now, more people will live shorter lives later.

But it’s also completely awful. Societies don’t work like this (I mean, outside of dystopian novels). We don’t sacrifice our sick and our elderly, so that other people can have their nails done. We don’t say ‘people with diabetes are going to die anyway so I’m keeping this life boat for those without comorbidities’. The coup de grace of the movie Soylent Green wasn’t the protein content of the eponymous food supplement but that it was made of people, of human lives.

People can manage chronic conditions, and can rebuild from possible future poverty. Present death, however, is typically hard to reverse. And this brings up a key point: these dissenters, with their lack of insight and absence of experience, are not responsible and not accountable for decisions that are made for the public’s health. But the state is. Academic professionals, medical professionals, scientists, they are. They are held accountable, because it is their job. If an accountant did a bad job with calculations, they would be fired — or, if it was negligent accounting, they might even suffer worse consequences. But for these ‘men with big opinions’ online, there are no consequences for being wrong. They don’t even acknowledge their errors of understanding, never mind apologise. This by the way is extremely bad science. The scientific method does not state that you should simply use the whiteboard eraser so that nobody can see your errors.

The people who have called the loudest for lockdown to end, from the start, have also been those who had the least to fear from the pandemic, and the most to lose from lockdown, i.e. money and prestige and comfort. Like in pretty much every other country, in South Africa the pandemic has disproportionately killed people from lower-income communities, killed people of colour. The profile of the dead is the opposite of the colour and class of those posturing online. Maybe it is easier to treat Covid-19 as a minor inconvenience (and not nearly as bad as not being able to have a dop) when nobody you know has died? This is speculation on my part, of course.

When I review the data that is currently available, what it appears to show is that the lockdown in South Africa — although it came with flaws, and even human rights abuses (and these must be accounted for) — prevented a possibly great number of deaths that would have happened by now if not for the lockdown measures.

We will never know whose deaths were prevented, the same way I will never know which 400+ non-natural deaths were prevented each week there were restrictions on movement and alcohol consumption. I just know that many people who would have died this year, did not die this year.

What I can also ascertain is that the self-appointed actuary-led group of pandemic wisdom is not, in fact, wise. Their data-based claims have failed to hold water, and are proven not only incorrect but also dishonest in their attempted revisionism. Their societal critiques remain grossly disconnected from the reality of medicine and public health, and from the majority of South Africans. They are not even widely supported by other actuaries. They are of course entitled to hold and even share grossly irresponsible opinions, even those verging on conspiracy theories (evidence by the group’s recent ‘anti-fear’ PR attempts). But it is not responsible of the media to continue publishing and promoting such content; particularly when there are so many other responsible and accountable sources available.

A couple of weeks ago, the lead actuary-investor posted: ‘Herd immunity is close for SA. Probably already there for the W Cape. An anti-fear campaign will be required. People need the facts. The facts about Covid are not scary. It’s the lies and fear-mongering that have driven the fear.’

Between 100 and 150 South Africans are still dying of Covid-19 every day — and those are only the confirmed cases. These are the facts we need, not an ‘anti-fear’ campaign built out of a personal sense of indignance or self-rightness, or a reliance on false metrics developed to quantitise human society and human lifespans into profits and losses.

We need to remind ourselves that groups like these — and countless others online, calling the pandemic some kind of hoax, or demanding an end to lockdown while the most vulnerable are still not protected – do not have ‘truth’ or even the ‘facts’ on their side. The scientific method is about trying to prove yourself wrong, not trying to reverse-engineer things so that you, and only you, are always right.

// ends

Journalist, writer and academic. Works in crime + violence research, fact-checking, data-led research. Believed in dystopia before it was cool.

Journalist, writer and academic. Works in crime + violence research, fact-checking, data-led research. Believed in dystopia before it was cool.