Category

Data Literacy

wearing a mask for protection against corona

Evidence-Based Responses to the Coronavirus Chaos

By | Current Affairs, Data Literacy

Whether we should have seen a disaster like COVID-19 coming, is up for debate. It’s too late now to pause on that point for too long. But we should at least try to learn from this so that we are better prepared in future. Not only for a similar pandemic but for the looming climate crisis.

Of more immediate relevance is how South Africa is using data to respond to the crisis, and what we can learn from other countries as we deal with a fast-changing complex problem.

What does the data tell us?

Is the way in which the government uses data to respond to the Coronavirus chaos an indication that South Africa will see better evidence-based policies in the future?  

Many South Africans are pleasantly surprised by the extent to which the government response to the Coronavirus is informed by evidence. More specifically, journalists and particularly health journalists have not always had good words for South African government responses to important health issues.

Now everything has changed. At least for the moment.

Unlikely cheers from the media and activists

Earlier in April, journalists wrote about how different the relationship with the government is now, during the COVID-19 crisis, compared to what it was like during the HIV crisis in the ’90s. They feel that the  South African government is now treating them with respect and as allies in the fight against the Coronavirus.

Many journalists have heaped praise on the government’s management of the crisis, to the extent that they themselves acknowledged that they are singing Kumbaya. Mia Malan[1] (Bhekisisa Centre for Health Journalism’s editor-in-chief and executive director) related how Professor Salim Abdool Karim (the calmly confident ally of South African Health Minister Zweli Mkhize, and chairperson of government’s advisory committee on Covid-19), was chastised, labelled as “poisonous” and told to “shut up and listen”, by the Mbeki-era Health Minister Manto Thsabalala Msimang at the height of the HIV-crisis.

His “sin” at the time, is now his virtue: using scientific facts as a basis for giving advice to the government, and for making public statements.

In the same article, Malan quotes Karim on how different the current relationship with the Minister of Health is. “With the coronavirus, our experience with government is exactly the opposite [of what we endured during the Mbeki and Tshabalala-Msimang era].

The minister has been contacting us, he wants to involve us, he is seeking the opposite of what Mbeki and Tshabalala-Msimang wanted… With HIV we were so slack with taking things up, we delayed mother-to-child-prevention of HIV and access to antiretroviral treatment. But with Covid-19 we’re proactive and we’re acting early…”[2].

Malan also says the same activists who have fought government previously on the HIV issue, are now “supporting and praising Cyril Ramaphosa’s early, evidence-based interventions.”[3] The source of the more constructive relationship and a new-found respect for government lies in transparency and access to data.

With HIV, government was seen as blocking access to data, while now, journalists are commending the government for facilitating a free flow of information. The health ministry has even set up a WhatsApp group to share the latest figures directly to journalist’s phones, and government’s actions thus far are regarded as an indication that they are heeding the advice of scientists and academics, to take the best possible decisions to ensure that COVID-19 can be managed as best as possible.

The dominant role of the “scientists and the health activists in the Covid-19 National Command Council”, was evident, according to Ferial Haffajee[4], who also noted that at least initially, “South Africa had bucked the trend of worst-case community transmission…and is on course to flatten the rate of infection and lower its peak by September”.

cyril ramaphosa

Appreciation for the quality of leadership

The applause for Ramaphosa and Mkhize’s leadership was also extended to the premiers of key provinces. Lester Kievit[5] of the Mail and Guardian pointed out: “The premiers of the three worst-affected provinces, Gauteng, the Western Cape and KwaZulu-Natal, have not wasted any time in showing they’re capable of leading and being the public face of the pandemic response in their provinces.”

Kievit continued to quote Sanusha Naidu, political analyst and lecturer at the University of Cape Town, who said that “the roles and importance of provincial government are coming to the fore during the pandemic in a manner that has not yet been seen in South Africa’s history”, even though it has unfortunately taken “a pandemic of this nature and this proportion to see this kind of leadership to emerge in South Africa.

Bill Gates also joined the praise singers in his interview with Trevor Noah on 4 April 2020[6]: “I was talking to President Ramaphosa today, who is not only [the] president of South Africa, he’s also the head of African Union… He is a very strong voice there, encouraging African countries to act quickly when the number of cases is still fairly low, which is true throughout sub-Saharan Africa right now.”

Some knew that a disaster like this was coming

Bill Gates was one of the few who saw this coming. In his 2015 TED Talk titled “The next outbreak? We’re not ready”, Bill Gates made a chilling prediction: “If anything kills over 10 million people in the next few decades, it’s most likely to be a highly infectious virus rather than a war – not missiles but microbes… We have invested a huge amount in nuclear deterrents, but we’ve actually invested very little in a system to stop an epidemic. We’re not ready for the next epidemic.”[7]

Despite this warning, enough had not been invested in the global health care system to prevent the next pandemic. According to Gates, “the Ebola outbreak exposed a shortage of specialists who were well trained to deal with the epidemic. This should have been used as a case study to prepare for the next one…The issue was not that we didn’t have a system that worked well enough, the problem was we didn’t have a system at all. We didn’t have a group of epidemiologists who would have gone, seen what the disease was and how far it had spread.”[8]

Gates says he hoped to raise awareness that would help the United States to get ready for “the next epidemic”, by implementing various preparations, which would have helped to respond quicker, more effectively and more efficiently to the current pandemic. He acknowledges that despite having epidemics like Ebola and Zika, the current situation is unprecedented: “…a respiratory pandemic that’s very widespread, really, we haven’t seen anything like this for 100 years.”[9]

The scale of the disaster is overwhelming

The reality was that no one was ready for COVID-19, says Professor Salim Abdool Karim. “Nobody ever thought that we would need to deal with something like this. We all thought we were going to deal with something like [the] flu.” And indeed, “nobody saw this coming” was a common sentiment amongst ordinary citizens in the days after the Coronavirus lockdown became the new reality in South Africa.

While “dozens of cases of pneumonia of unknown cause” were being treated in China by the 31st of December 2019, and a “new virus” was identified by researchers in China days later[10], nobody would have believed anyone who predicted that the world would be on lockdown three months later. In fact, for most people, a “closed off” country was a totally foreign concept.

Now it has become everyday life.

The acute reality of the extent to which our daily lives have become surreal in a matter of weeks, was evident by the 1st of April when there seemed to be general agreement that no April’s fool joke could upstage the bizarre reality of the world. And that April Fools’ Day jokes would simply not be appropriate.

The nature of the beast

It is not that the world did not plan for disaster, or that it did not foresee an epidemic – it is the nature, scale and the unpredictability of the current disaster that is problematic. It is airborne, highly contagious, not localised or contained to a country or certain geographic area – it is a tsunami that engulfs the world.

Disasters can take many shapes and forms, and often when thinking about disasters, we tend to think about natural disasters, like earthquakes, tsunamis, or disasters related to climate change.  Now we are faced with a medical disaster, which poses particularly wicked problems for scientists, economists, politicians and officials to solve.

The impossible choice of this crisis is one where decision-makers have to weigh up the cost of lives against the cost of containment measures to the economy.  The available data is incomplete, and there is not even a quick and simple way to determine how to attach weight to and weigh the choices that need to be weighed. Ultimately, the simple question “What are we willing to sacrifice economically to save a life?”[11], has no simple answer.

The well-known Cynefin model[12] comes to mind, and the descriptions of “unknowable unknowns” and “extremely volatile and urgent situation” in the “Chaotic” quadrant shows where South Africa and the world is at this time. We may be faced with a level of complexity that has not been experienced ever before.

It is both apt and ironic, that the proposed action in the chaotic quadrant is to “take swift and decisive action” and to “apply novel practices”. Novel practices indeed, to fight a novel virus, now known as the Coronavirus.

graph on coronavirusSource: Agility 11

The level of complexity that has to be taken into account in the management of the Coronavirus crisis is unprecedented.

Complexity in action

With the Coronavirus, a salient factor is that there is a continuous shift in what is known and what is not yet known, and importantly, what should be known. Adding to the level of complexity is that while there are commonalities, patterns and trends that emerge globally, countries in the world are vastly different in terms of many factors, a few of which are mentioned here:

  • Population size, density, and age profile;
  • Health system characteristics, coverage and access;
  • Health profile of the population in terms of the incidence of conditions such as diabetes, hypertension, HIV and Tuberculosis (TB);
  • Infrastructure aspects such as housing and transport systems
  • Economic system characteristics, resilience and pre-COVID-19 economic status

Countries also have varying levels of expertise in dealing with similar types of viruses.

In a recent presentation to the South African Minister of Health, Karim referred to South Africa’s unique epidemic trajectory (see Figure 1). The curve had turned quicker than even “best-performer countries like South Korea and Singapore and that community infection (the third stage after imported and local transmissions) are lower than expected.

But there is no time to get smug or complacent, because South Africa is not out of the woods yet. We may have bought time to spread out the disease over more months, and to prevent “small flames” from becoming “raging fires”[13].

graph of coronaFigure 1: South Africa: number of cumulative cases and number of days since the 100th case

Source: Daily Maverick/Business Maverick

Another factor that adds to the complexity, is that within countries, cases are not evenly spread. This is clear from the high number of cases in Wuhan, Lombardy, Madrid and New York[14], to name a few. In South Africa, the highest number of cases are in Gauteng and Western Cape provinces, and within these provinces, the concentration of cases is clearly higher in some suburbs than others, as is evident from the two maps below.

Just remember, that this is data that was presented to the Parliamentary Health Portfolio Committee on 10 April 2020, so by the time you read this, the patterns may have changed. It is also possible that clusters of cases occur in different types of populations, which may lead to different patterns in mortality rates.

covid cases in gauteng

Figure 2: COVID-19 cases in Gauteng on 10 April 2020

Source: National Department of Health[15]

covid cases in western cape south africaFigure 3: COVID-19 cases in the Western Cape on 10 April 2020

Source: National Department of Health[16] 

Learning and adapting on the trot

While there is agreement that there are no clear answers at the moment and that answers are changing all the time, there is some disagreement on what evidence should be used to plan responses, and also how this evidence should be used. Emerging data is used for decision-making, and one of the ways in which it is used is modelling.

Early in April, the National Institute for Communicable Diseases (NICD) confirmed that significant progress was made by groups commissioned to model and project the spread of the Coronavirus in South Africa[17]. Modelling was also used by other countries in the world to make projections that informed governments’ responses, regarding “procurement of emergency medical supplies as well as the building of additional hospital bed capacities in countries around the world”[18].

Modelling, however, is not foolproof, and some critics pointed out that it is simply not known how the virus may spread in circumstances unique to South Africa. Alex Welte (research professor at, and the former director of, the Centre of Excellence for Epidemiological Modelling and Analysis at Stellenbosch University) cautioned in a “GroundUp” article that “there is more we don’t know than we do”[19].

As is highlighted above, the issue of relevant and complete evidence is pertinent to ensure good decision-making. That is a luxury that does not exist at the moment. Whether the decisions made at the moment are the correct ones, will only be established later.

Paving the way for evidence-based policy making in future

Currently, decision-making on the Coronavirus crisis rests heavily on evidence produced by health scientists. There are, however, other groups that may gain more clout as the situation unfolds. For example, the business lobby will strengthen its appeals to the government to also protect the economy, and undoubtedly this lobby will also rely on evidence to give weight to their arguments.

DPME quoteEvidence decision-making can be a complex affair, and the same applies to evidence-based policy making. Although South Africa is committed to evidence-based policy-making (see the text box below), in reality, party-political ideology inevitably plays a role in policy-making.

DPME. Overview Paper, What is Evidence-Based Policy-Making and Implementation? Evidence-Based Policy-Making and Implementation October 2014[20]

It should be noted that the “evidence-policy” gap in the health sector is not unique to South Africa. Scholars[21] who researched the phenomenon has found that scientists could do more to provide “better evidence to reduce policy-maker uncertainty”, and concluded that successful evidence-based policymaking required “pragmatism, combining scientific evidence with governance principles, and persuasion to translate complex evidence into simple stories” (Cairney and Oliver, 2017).

Despite the limitations of the current situation, documented literature on the “evidence-policy gap”, and some level of cynicism regarding South Africa’s track record with evidence-based policy-making, it is encouraging that emergent evidence is being used to craft South Africa’s response to the Coronavirus crisis. It may give the South African government the confidence to rely more on evidence-based policy making in future.

By Fia van Rensburg

[1] Malan, M. 2020. The facts beat the quacks: Our #Covid19SA vs. our #HIV response. Bhekisisa Centre for Health Journalism Article

[2] Ibid

[3] Ibid

[4]Haffajee, F. 2020. The new ‘Doctors Pact’ that could help flatten the Covid-19 curve. Daily Maverick. 14 April 2020.

[5] Kiewit, L. 2020. Fighting Covid-19: The rise of the premiers. Mail and Guardian 16 April 2020.

[6] Head, T. 2020. Bill Gates praises Ramaphosa – but billionaire bats away false claims. The South African News 5 April 2020.

[7] Carras, C. 2020. In 2015, Bill Gates predicted an epidemic would kill millions. Here’s what he says now. The Star 14 April 2020.

[8] Bhengu, C. 2020. In 2015, Bill Gates said the world was not ready for the next major virus.

[9] Carras, 2020.

[10] Taylor, D.B. 2020. A timeline of the Coronavirus Pandemic. New York Times. (14 April 2020).

[11] Hartford, T. How do we value a statistical life? The Financial Times Limited.

[12] The Cynefin model (pronounced kuh-NEH-vin), created by Dave Snowden, is a useful model of complexity. It defines four domains: Obvious (simple), Complicated, Complex and Chaotic.

[13]Haffajee, F. 2020. The new ‘Doctors Pact’ that could help flatten the curve. 14 April 2020.

[14]Klasa, A. (Ed.). 2020.Coronavirus tracked: the latest figures as the pandemic spreads. The Financial Times Limited. Accessed on 17 April 2020 at https://www.ft.com/coronavirus-latest

[15] Gauteng cases mapped. Source – National Department of Health presentation to Parliamentary health portfolio committee, 10 April 2020.

[16] Western Cape cases mapped. Source – National Department of Health presentation to Parliamentary health portfolio committee, 10 April 2020.

[17] Cowan, K. 2020. Coronavirus in SA: ‘Significant progress’ on modelling – but, for now, we will not know what it shows. News24. 1 April 2020. https://www.news24.com/SouthAfrica/News/coronavirus-in-sa-significant-progress-on-modelling-but-for-now-we-wont-know-what-it-shows-20200401

[18] Ibid.

[19] Ibid.

[20] DPME

[21] Cairney, P., and Oliver, K. 2017. Evidence-based policymaking is not like evidence-based medicine, so how far should you go to bridge the divide between evidence and policy?. Health Res Policy Sys 15, 35 (2017).

graph on coronavirus

Looking beyond daily updates on the number of COVID-19 cases

By | Data Literacy

This blog is about data. Specifically, what we do with it, and what we should think about to ensure that we make sense of it and interpreting in the context of the bigger picture.

COVID-19 data is topical at the moment, and we use a specific dataset to illustrate key points about interpreting data.

This article was compiled in early April and data figures have changed significantly since being published.

The most commonly heard data-based story on COVID-19

Most people are following updates on COVID-19 cases closely. We often find ourselves discussing the increase in the number of cases and deaths when the statistics are updated. These discussions usually involve comparisons between countries.

For example, based on the first few columns in the table below, the story could be as follows:

“America now has the highest COVID-19 infection rate, with a total of 435,160 cases. Other countries with more cases than China are Spain (148,220 cases), Italy (139,422 cases), Germany (113,296 cases) and France (112,950 cases).  

China is the country with the sixth-highest number of cases (81,865).”

And the story could continue as follows:

“The highest number of deaths are recorded in Italy (17,669), followed by the USA (14,797) and Spain (14,792).”

We can also compare death rates with infection rates, and so on.

Numbers give us some sense of security because they give the impression that we can pin-point something – that we can know exactly what the situation is. However, if we do not contextualise data, it may not give an accurate account of a situation at all.

Regarding the COVID-19 data, we also need to look at other factors. In the case of COVID-19 data, key questions would be:

  • What is the proportion of cases in relation to the population of the country?
  • What is the number of people actually tested?
  • What proportion of the population has been tested?

There may be other stories too

Once we start asking these questions, the story changes, and we gain new insights. For example:

“Germany and France have almost the same number of COVID-19 cases (respectively 113,296 and 112,950), but the incidence per 1million of the population is slightly higher for France (1,730) than for Germany (1,352), and the number of deaths per 1million of the population for France is markedly higher (167) than for Germany (28).”

Similarly, we would be able to say the following:

“While the actual number of COVID-19 deaths in the USA (14,797) is almost the same as in Spain (14,792), the death rate per 1million of the population is much higher in Spain (316) than in the USA (45).”

What is important about this, is that if we look at China, which we previously saw as the country with the 6th highest infection rate, we can see now that their infection rate and death rate per 1 million of the population was very low, at respectively 57 and 2.

Surely this would significantly change the way we interpret the data?

worldometer graph

(Accessed on 9 April 2020 online at https://www.worldometers.info/coronavirus/#news)

Seeing the numbers in the context of the bigger picture

While the actual numbers are important, and while every life matters, a more accurate picture of the scale of the epidemic per country will be obtained if we sort the columns in this table according to statistics that show the proportion of cases in relation to 1 million of the population.

In the table below, we can see that countries that do not even feature in headline reports on the epidemic, actually have the highest incidence of cases in relation to 1 million of the population. At rates much higher than the countries which are hotspots, merely because of the sheer number of cases.

worldometer 2

(Accessed on 9 April 2020 online at https://www.worldometers.info/coronavirus/#news)

This table tells a different story:

microstates

Credit: https://en.wikipedia.org/wiki/European_microstates

“The number of COVID-19 cases in relation to 1 million of the population in three of Europe’s six micro-states are staggeringly high. Vatican City has 9,988 cases per 1million of the population; San Marino 9,077 and Andorra 7,300).”

According to this analysis Spain and Italy have respectively the 8th and 10th highest number of cases per 1million of the population. In this comparison China is not near to featuring on the list of countries with the highest number of cases.

Other contextual factors

An analysis of death rates per 1 million of the population will show similar results. Another factor that needs to be taken into account in interpreting the numbers is the trajectory of the disease. For example, we know that the peak has been reached in China, and that Italy is slightly ahead of Spain and the USA.

This will influence how we interpret the currently reported incidence and death rates. We could also inform our analysis by keeping an eye on the number of new cases reported daily (see the first table). And if this is steadily or exponentially increasing. We will be able to anticipate in what direction the numbers will move in the next few days.

Another factor that could influence the numbers is testing. We should consider the following questions: “Do countries conduct the same tests?” and “What is the reliability level of the tests?”

An issue that is coming up in the media is whether all COVID-19 deaths are actually reported as such. This may vary from country to country, and may also link up with the extent to which testing is done. If more people are tested, reporting of COVID-related deaths is likely to be more accurate.

It should also be considered if social distancing measures are or have been implemented in a country, and if so, what the nature of such measures are or were.

A shifting picture

It is clear that the COVID-19 statistics on any given day are merely a snapshot in time, which should be considered within the context of many factors. It tells only part of a story that is still unfolding.

It is also important to note that different countries are at different stages in the disease. This means that while numbers may give us a sense of “knowing exactly”, and even a sense of security, we have to engage with numbers critically to understand them better.

South African numbers

Compared to other countries, South Africa still has relatively low numbers (see the table below).

table of data

(Source: DWC. Derived from data from https://www.worldometers.info/coronavirus/#news on 11 April, 2020. Numbers have subsequently risen)

south african map

 

When interpreting provincial statistics in South Africa, it is necessary to at least consider the number of cases per province in relation to the size of the population in that province.

data from south africa

 (Source: DWC. Derived from data from https://en.wikipedia.org/wiki/2020_coronavirus_pandemic_in_South_Africa        and https://en.wikipedia.org/wiki/List_of_South_African_provinces_by_population)

As on 9 April 2020, the percentage of confirmed cases in Gauteng and the Western Cape is higher than the percentage of the proportion of the population resident in those provinces. Confirmed cases in KZN and Free State are on par with the percentage of the population resident there.

And in the other provinces, which typically have large rural populations, the incidence is significantly lower than the percentage of the population resident in those provinces. This picture will probably change as testing in communities is rolled out.

Improving data literacy

The examples we have used here relates to the current situation with COVID 19, but the principles mentioned here apply to other datasets and contexts as well. We encourage our readers to engage critically with data, to discover the multiple stories that can be told by a single dataset.

Even though we are engaging with numbers, which often seem to tell a very precise story, in reality, the truth is often more nuanced than what we see at face value.

Postscript

We have used COVID-19 data to illustrate some points regarding data literacy. We do want to emphasise that we are acutely aware that these numbers are not only cold facts, but that they relate to real people, real hardship and untold sadness of those who have lost loved ones in this time.

To express our sympathy with the loss of many families in this time, and to emphasise that we know that numbers on their own cannot fathom the depth of people’s lived experiences, we would like to share this poem with you.

By Fia van Rensburg

untitled poem