What does a vulnerability index tell us about the COVID-19 pandemic in South Africa?
The most vulnerable provinces are seeing the fastest increase in new cases, but older South Africans are at risk regardless of where they live.
With over 400,000 reported cases of COVID-19, South Africa is the fifth hardest hit nation in the world and the worst affected on the African continent. Recent reports of a 60% rise in excess deaths indicate that the impact of COVID-19 may be greatly underestimated. The spread of the disease in South African communities has been magnified by health, socio-economic, and structural inequities. Understanding better how these factors interact will be key to slowing its further spread.
Inspired by our US COVID-19 Community Vulnerability Index, we created the Africa COVID-19 Community Vulnerability Index (Africa CCVI) to measure community ability to mitigate, treat, and delay transmission of COVID-19, and to reduce its economic and social impacts.
The Africa CCVI maps the vulnerability of each sub-national region in 48 countries. By pairing the CCVI vulnerability score with mobility data and estimated rates of hospitalization and mortality, we can arrive at a holistic picture of each region’s preparedness for COVID-19.
Compared with all African countries, South Africa is highly vulnerable to COVID-19 on the CCVI, due to four factors: high population density; high fragility (a term we use to describe factors like civil unrest or food insecurity); a higher proportion of people aged 65 and older; and a high prevalence of epidemiological risk factors such as HIV/AIDS.
But our index also allows us to drill down further to understand vulnerability at a more localized level. We can ask: which provinces of South Africa are more vulnerable, and why? How is the pandemic moving through these communities? And how are people there responding?
Here’s what we found when we dug deeper into the data:
1. COVID-19 hit less vulnerable communities first — but it’s now catching up in the most vulnerable regions.
In many countries, densely populated urban centers like Cape Town in South Africa or New York in the United States were among the first places to be hit hard by COVID-19. But these may not be the most vulnerable regions.
We grouped South Africa’s nine provinces into three categories of vulnerability: high (a score of 0.67–1.0 on the CCVI, where 1.0 is the highest possible score), medium (0.33–0.67), and low (0–0.33). Three provinces fell into each category. In the chart below we calculated the confirmed cases per 100,000 people and plotted those numbers against vulnerability level.
The orange line represents the three provinces with medium vulnerability — including Cape Town — and these saw a rise in confirmed cases from early May. Cases increased in high-vulnerability provinces only at the start of June — indicating a four-week delay in the spread or detection of COVID-19 in the provinces most at risk. Although medium vulnerability provinces currently have the most cases per capita, cases are rising more rapidly in regions hit later. Greater vulnerability may contribute to faster spread of the virus once a community is exposed, as we have seen in the US.
2. While several vulnerable regions are current hot spots, other vulnerable regions with low case counts should not be neglected.
Let’s take a closer look at where cases are located.
South Africa has the continent’s highest total number of cases and the second highest per capita case count, in part due to widespread testing. However, COVID-19 may still be spreading faster than is being tracked, given an increase in the cumulative testing positivity rate from June 6 (4.9%) to July 21 (28.2%). Regionally, cases are concentrated in Western Cape, Gauteng, and Eastern Cape (in descending order of cumulative cases).
Looking at the map below, we see that two out of the three regions with the most cases — Western Cape (CCVI=0.6) and Gauteng (0.9) — are highly or very highly vulnerable per the CCVI. The epicentre has shifted over time, from Gauteng (Pretoria/Johannesburg) to Western Cape (Cape Town) in mid-April, returning to Gauteng (Johannesburg) in early July. Cases in Gauteng have skyrocketed in the past two weeks. With the greatest fragility, worst health system, and highest population density of any province, Gauteng’s predicament may be attributable to these underlying drivers of vulnerability. Western Cape also has a weak health system and high population density. Eastern Cape, one of the least vulnerable provinces, has the third highest number of cases, resulting from superspreading events traced back to three burials. Vulnerabilities in this province are driven by an older population and household crowding.
While two of the most vulnerable provinces, KwaZulu-Natal and Mpumalanga, currently have lower case counts, they should be prioritized in preparedness and mitigation efforts. The weak health infrastructure of both provinces, coupled with the highest HIV rates in the country, make them disproportionately at risk of COVID-19.
3. Fatalities remain concentrated in older populations.
Despite the high number of COVID-19 cases, South Africa has a low case fatality rate (1.5%) compared with other African countries with large numbers of cases, such as Egypt (4.9%) and Nigeria (2.1%).
This may be due to South Africa having a stronger health system overall, as well as widespread testing. So who is getting infected? The majority of identified cases (80%) are in the population aged 50 and over, especially in Western Cape (where 79% of cases are among over-50s). This province also accounts for nearly half the fatalities nationally. In contrast, Gauteng has a much younger population and accounts for only 19% of COVID-19 deaths despite having over 50% of the country’s cases.
To get a sense of which provinces may yet be hardest hit by the health effects of COVID-19, we estimated the percentage of people infected with COVID-19 that would require hospitalization (the Infection Hospitalization Rate, or IHR), and the percentage of infected people that would die as a result of the disease (Infection Fatality Rate, or IFR). Both rates were calculated by combining age- and gender-stratified estimates developed for France by Salje et al. with data on African subnational age and sex structures.
The map below depicts the predicted IFR by province. While the predicted IFR range — from 0.7% to 0.9% — may seem small, the 28% increase in potential fatalities that it represents is substantial in a country with a limited ICU capacity of just 3,300 beds.
The three most vulnerable provinces according to our CCVI — Gauteng, Mpumalanga, and KwaZulu-Natal — have the lowest estimated IFRs. It is the provinces with the oldest populations — Western Cape, Eastern Cape, and Limpopo — that have the highest predicted IFRs, based on our analysis. While case counts are still low in Limpopo, the high expected IFR is critical to note, especially since the province was socially distancing the least even as cases continued to rise over the past two weeks.
4. Vulnerability and government policy notably impact social distancing.
To understand how the population has responded to government policies, we analyzed Google mobility data.
South Africa decreased its mobility the most during the pandemic compared with all other countries across Africa that have available Google mobility data. Across the continent, the average decrease in mobility is 11%, but in South Africa it was more than twice that rate, at 28%. As we can see in the map below, the three most vulnerable provinces in South Africa (Western Cape, Gauteng, and KwaZulu-Natal) decreased their mobility the most.
We believe the reason for higher overall vulnerability coinciding with more social distancing is a combination of case prevalence and high socioeconomic status. Gauteng (Pretoria/Johannesburg) and Western Cape (Cape Town) have been the epicenters of COVID cases since mid-April, and the larger burden in these provinces likely caused the population to respond by reducing their mobility more drastically. Despite their vulnerability, these provinces have a higher socioeconomic status than others, which may have made it easier for their populations to social distance more.
Looking more closely at changes in mobility in relation to government policy, we found that military enforcement of South Africa’s lockdown, which sparked much social unrest, did little to further decrease mobility. Our analysis suggests that government announcements alone encouraged populations in all provinces to stay home and social distance, resulting in a 50% decrease in mobility in the two weeks between the national disaster declaration and the militarily enforced lockdown. After a month of lockdown, it only just three days from when the government announced plans to reopen (April 23rd) for there to be a sharp increase in mobility (April 26th), even though the re-opening campaign did not begin officially until May 1st.
Bringing together insights on provincial vulnerability and government policy on mobility, this data suggests that governments should be responding to people’s concerns, particularly around food insecurity, rather than on legal enforcement measures to see greater adoption of social distancing measures.
With case numbers continuing to climb, South Africa’s COVID-19 predicament looks dire, but with continued measures and careful consideration of underlying vulnerabilities, it may still be possible to reduce the number of deaths and mitigate the socioeconomic impacts of the pandemic.
First, while the government may be tempted to focus on hotspots of new cases, it must remain vigilant everywhere. Three neighboring vulnerable regions — KwaZulu-Natal, Limpopo, and Mpumalanga — have few confirmed cases, but new case counts have increased rapidly over the last two weeks. The socioeconomic characteristics of these regions, with household crowding and high poverty rates, leave them vulnerable to faster spread of COVID-19 and greater adverse impact of lockdown restrictions. To encourage social distancing and increase case detection, interventions could focus on providing cash transfers to food-insecure populations and deploying mobile testing centers in areas with inadequate health infrastructure.
Second, in Limpopo and KwaZulu-Natal — provinces with a relatively higher proportion of people over 65 — mitigation policies (such as earlier store opening hours and clear signage about social distancing requirements) could be targeted to the elderly. Quarantine sites and isolation units could be used to protect family members from those who may have been exposed to the virus. This is important because nearly a quarter of the population of each province live in multi-generational households.
By understanding vulnerability at a more localized level, South Africa can better prepare and respond to this unprecedented viral threat.
This work was made possible by everyone at the Surgo Foundation, including but not limited to (in alphabetical order): James Baer, Sofia Braunstein, Grace Charles, Bethany Hardy, Anubhuti Mishra, Sema Sgaier, Peter Smittenaar, Nick Stewart, and Staci Sutermaster.
Data updated as of July 14. Case and death data are from the COVID-19 South Africa Dashboard, which tracks daily confirmed cases from February 5 onwards. Mobility metrics were developed by Google and policy data was provided by Partnership for Evidence-based Response to COVID-19 (PERC). The IFR was modeled based on rates calculated from analyses by the Center for Global Development and Salje et al.