Pandemics Data & Analytics (PANDA) is a multidisciplinary group seeking to inform policy. PANDA considers explanations that allow us to count the human costs of COVID-19 responses globally.
PANDA members work voluntarily, offering their skillset to contribute to informed policymaking and decisioning. PANDA is not aligned with any political entity and is funded by its members, speaking engagements and the public. PANDA quickly grew from its South African roots into a global organisation that has a presence in every corner of the world and currently counts over 140 members predominantly from healthcare fields (doctors, healthcare academics, geneticists, epidemiologists, paediatricians, virologists, psychologists etc) but also from other professions including data scientists, engineers, actuaries, lawyers and economists. PANDA’s approach is premised on the fact that COVID-19 is not just a medical challenge. We believe that most governments have failed to deal properly with the virus because they have failed to recognise the patient as a sum of its parts as medical ethics require, and have looked at the impact of one disease alone. PANDA is guided by a scientific advisory board that consists of some of the finest medical minds on the planet including professors of medicine at Harvard, Stanford, Johannes Gutenberg, Nottingham and Oxford Universities as well as a Nobel prize laureate and a former chief scientist and Vice President at Pfizer.
Our key principles
- Lockdowns run contrary to pre-COVID science (endorsed by the WHO and other medical bodies), which consistently advised against quarantining the healthy under any circumstances, particularly because they entail devastating collateral damage (lack of access to healthcare for other medical emergencies, psychological trauma due to lack of human interaction, lower tax revenues to provide social services and starvation and poverty from job losses to name a few). There is no evidence in the data (after nearly a year) to justify a change in the pre-lockdown science. Analysing the data, there is simply no correlation between the stringency of lockdowns and the number of deaths experienced.
- Imposing restrictions on healthy people, including mandatory mask wearing runs contrary to pre-COVID science. As recently as December 2020, the WHO stated “that there is limited and inconsistent evidence to support the effectiveness of masking healthy people.” There is also science suggesting that extended periods of mask wearing can be harmful due to oxygen deprivation and various other side-effects. Health and safety laws have traditionally limited the time during which masks can be mandated. Observational as well as the more rigorous RCT studies undertaken since 2020 have only added to the science recommending against mandatory masking. PANDA believes that there is sufficient doubt around the efficacy of masks and sufficient evidence that they can have harmful effects, that masks should not be mandated. However, PANDA supports each individual’s right to choose to wear a mask.
- Poor countries, such as South Africa, are the worst affected by lockdowns as they don’t have the financial resources or infrastructure to support the people who have lost their jobs due to the economic collapse caused by restricting the economy. Poverty is a key driver of poor health outcomes and deaths in developing countries and lockdowns significantly increase poverty.
- Within a society, poor people are the worst affected by lockdowns. Poor people tend to live in more densely populated environments, have jobs that are less conducive to remote working, lack the resources to educate their children remotely and the money to afford good healthcare. They are the first to lose their income when lockdowns are imposed. Lockdowns have resulted in the rich isolating at home and being served by the poor, with the wealth gap widening as a result.
- It is questionable if one can control a virus at all (unless you aggressively quarantine an island nation such as Singapore or New Zealand and deploy an intensive entry testing and isolation system). If successful, these measures mean that a population must remain isolated until the virus has disappeared. Although viruses are known to attenuate over time, it is unusual for a virus to disappear entirely. The Spanish flu continues to circulate today. Where a virus has a significant aged-based mortality component, efforts to ‘shelter’ should, in our view, be focused on the most vulnerable (elderly and people with co-morbidities in the case of SARS-CoV-2) whilst allowing the rest of society to continue as normal – which will reduce the economic (and thus social) trauma felt – increasing the resources to protect the vulnerable.
- By implementing focused protection, the natural outcome will be greater community immunity as the young and healthy develop immune responses (with little danger to them) to the virus, short-circuiting its spread to the most vulnerable. This is not a ‘strategy’, but rather an inevitable outcome of focused protection. Focused protection is not an alternative to vaccination. Effective and safe vaccination would enhance community immunity levels. Studies of previously infected COVID patients are showing what we knew already – the human immune system is complex and very effective.
Only open science, debate, and policy transparency allow us to understand the problem better and make wiser policy choices.
What did we say during the first wave?
As COVID-19 reached South Africa’s shores, various institutions began modelling the number of lives that would be lost. All models modelled only one wave of COVID-19, which was expected to complete in 2020. Initial models of the first wave produced for government suggested that between 89,000 and 350,000 lives would be lost during 2020. The virus was expected to attenuate (as viruses do) and modelling subsequent waves would therefore involve different assumptions.
PANDA’s principal observation was that before a decision could be taken to impose lockdown measures designed to save lives, government needed to know how many lives would be lost to the lockdown. No one was modelling the impact of the lockdown measures and there was little recognition that the measures would even have any impact on lives and livelihoods. Our initial report in early 2020 attempted to quantify the impact of increased poverty on life years lost and the results concluded that the total loss of life years from lockdown would dwarf the life years that would be lost to COVID-19. We did not try to redo the work of those who had modelled the COVID-19 deaths. We relied on their data. We commented though that the later outbreak in South Africa gave us the benefit of seeing what had happened elsewhere in the world and we noted what has now proven to be true – that the estimates of first wave COVID deaths were overwrought. We had calculated that around 20,000 deaths would be caused by the virus in the first wave and we mentioned this estimate in the very first paper we produced.
PANDA was lambasted for making this prediction because most “experts” were predicting five to fifteen times more casualties. Our critics noted that modelling is difficult and criticising models is easy. PANDA only got into modelling because we were told to disclose our model if we wanted to criticise others. Some of the same critics who say that modelling is difficult to excuse SACEMA’s 89,000-351,000 numbers criticise PANDA for mentioning estimates between 10,000 to 20,000 in the initial stages. We adjusted our model as more data became available and we published our model online in real time. From the end of July 2020, our estimate was between 16,000 and 29,000 confirmed COVID-19 deaths in the first wave. By 5 September 2020, the South African COVID-19 Modelling Consortium (“SACMC”) had revised their model down to around 16,000 confirmed COVID-19 hospital deaths for the first wave, which is equivalent to 20,000 confirmed COVID-19 deaths inside and outside hospitals. Our projections were based on the daily, confirmed COVID-19 deaths published by the National Department of Health on a daily basis, and we reported these numbers against our projections on a daily basis in a transparent way. Confirmed COVID-19 hospital deaths is a subset of confirmed COVID-19 deaths, which is a subset of all COVIDCOVID-19 deaths which is a subset of excess deaths. It would be unfair to compare a projection for one variable (say confirmed COVID-19 deaths), against the actuals of another variable (excess deaths).
Set out below is a list of projections for the first wave:
- 16,300 COVID-19 deaths in hospitals (SACMC 5 Sep projection)
- 20,000 confirmed COVID-19 deaths (PANDA first estimate 5 May and PANDA 1 Sep projection)
- 37,000 total COVID-19 deaths (SACMC 5 Sep projection) – validated against 80% of the MRC excess deaths number.
All mortality models turned out to be wrong. By the onset of the second wave towards the end of October 2020, just under 20,000 confirmed COVID-19 were reported by the National Department of Health, far less than the 351,000 government had initially modelled and slightly less than the deaths PANDA had initially modelled.
PANDA’s May prediction turned out to be the most accurate prediction made at the time, but the discussion about accuracy of mortality models is not one that PANDA ever sought to have. Even if government had been correct that 89,000 – 351,000 people would die of COVID in the first wave, PANDA’s analysis showed that more damage would be done through lockdowns. Government has still not produced any lockdown impact assessment and there has been no analysis to challenge PANDA’s. On the contrary, every analysis done since we produced our initial paper has confirmed that lockdowns do more harm than the virus. We have already seen, notwithstanding the belated recognition that the impact on lives and livelihoods must be balanced, the short-term economic damage wrought, as millions of jobs have been lost. The longer-term effects are only beginning and will be felt for years to come.
Where we are now
It should be noted that no one modelled a second wave for South Africa and that by the end of the year, even government was estimating around 20,000 confirmed COVID-19 deaths for the country. In the rest of the world, we saw no increase in infections during summer, despite lockdown stringency being eased. With the onset of winter, there was an increase in positive test numbers that was not matched by an increase in deaths. PANDA predicted that the second wave would not kick off until winter. We were clearly wrong.
The emergence of a second wave, which we (and all the other modelling consortiums) did not expect to be as dramatic or to come so soon, and an increase in deaths is getting an understandably large amount of attention, particularly from our critics. South Africa is a significant outlier both in regard to the timing of the second wave and to the size of the wave, making it difficult for anyone to have predicted the second wave. We have attempted to explain the South African anomaly but with the paucity of accurate data available, we are not in a position to offer a definitive explanation at this time. Certainly, the theory seems plausible that a more virulent variant that no one had anticipated is responsible. We do not know enough about the new variant to be able to say for certain if it is more contagious and if so, how much more contagious. We are still unable to separate out deaths with COVID and deaths from COVID. Given the inaccuracies in the rt-PCR testing regime, we still do not have accurate infection or death numbers. More data is required and this data will likely only be produced by government in two years’ time.
Regardless, none of this changes our broad principles, that lockdowns are not the answer to the challenges that this virus poses, regardless of the exact number of tragic deaths or waves experienced. As is demonstrated by all of the attempts to model the virus, predicting the exact timing and severity of these waves, associated strains etc. is nigh impossible. It distracts us (and many others) from the key question of how we should respond in an environment of unknowns. PANDA’s message remains unchanged. In order to ensure that we abide by tried and tested principles of public health, we must first do no harm. We must not focus on one virus alone and we cannot implement measures until we understand the collateral damage such measures will cause. The damage caused by lockdowns is simply too large and the evidence that these lockdowns make any difference is too weak to justify the current approach.
There is no evidence that lockdowns have achieved what they were intended to achieve in South Africa. In both the first and the second wave, the imposition and lifting of lockdown restrictions had no impact on positive test numbers. Given the time it takes for measures to reflect in infection and death figures, it is evident that even the latest round of lockdowns implemented on 29 December 2020, had no impact. The measures were instituted after the second wave peak in the Western Cape, and only days before the countrywide peak.
Where to next?
Rather than simply criticising the response, we have come up with a practical and pragmatic set of policy recommendations that we hope will encourage debate and find a solution to the current predicament that the world finds itself in.
For more info, see PANDA’s Protocol For Reopening Society.
Public health policy is inherently one of trade-offs and PANDA continues to question whether the correct trade-offs are being made. We will continue to call out what we see as bad science and ill-conceived policy. We will continue to ask the difficult questions. PANDA is a group of passionate people wanting the right decisions to be made for the sake of the people of our country – particularly the poor, on whom the current measures impact the most. This does not mean that we have all the answers and won’t make mistakes. Only open science, debate, and policy transparency allow us to understand the problem better and make wiser policy choices.
Photo by Jeremy Bishop on Unsplash