⚠️ Disclaimer: I am not a medical professional or epidemiologist. This analysis is not necessarily reliable.See the official NICD and SA Coronavirus websites for reliable information.
This analysis of public data about the 2020 coronavirus pandemic in South Africa tries to answer two questions:
-
Where are we now? Is the infection slowing down? Are our efforts working?
-
What's ahead? When will the infection peak, and when might it be over?
ℹ️ Note: You can click on charts below to see bigger interactive versions on Google Sheets, along with the underlying data and calculations.
We can expect that infections will roughly follow a logistic growth curve, so we can try to visualise where we are relative to that curve's inflection point, or midpoint.
💡️ Video explanation: "Exponential growth and epidemics" by 3Blue1Brown
The first way to look at this is to visualise the growth factor, which is the second derivative (rate of change of rate of change) of the total cases over time, or the ratio between new cases each day.
- Growth factor > 1 means we are before the inflection point, in the exponential growth phase.
- Growth factor = 1 means we are crossing inflection point, and growth is stabilising. Ideally, this should be the half-way mark, at roughly half the number of cases that the total will eventually grow to, and half the number of days from the first infection to the peak.
- Growth factor < 1 means we have passed the inflection point, and growth is slowing down.
The data is noisy, so this chart includes several layers of smoothing over multiple days, to make the overall trend clearer:
As of the start of the national lockdown (~27 March), the growth rate is dipping below 1, but this is likely a temporary artifact of reporting, rather than a true inflection point.
💡️ Video explanation: "How To Tell If We're Beating COVID-19" by MinutePhysics
The second way to look at this is to visualise the trajectory of new cases versus total cases. This makes the exponential growth phase look linear, and clearly highlights when the curve begins to decelerate.
This shows the same temporary (?) dip after the start of the national lockdown.
We can estimate the progression of the infection over time by fitting a logistic growth curve to the data, along with an estimate of the total infected population.
SACEMA and NICD present the following estimates for how many of South Africa's population will become infected:
- 10% infected is an optimistic estimate, if early mitigation measures are effective.
- 20% infected is typical for the seasonal flu (19%).
- 40% infected is a more pessimistic estimate, if mitigation is slow or ineffective.
📰️ Source: "The terrifying coronavirus projections that pushed govt into lockdown", News24, 19 March 2020
We can estimate the number of active cases by assuming an average recovery (or fatality) period, and subtracting this from the infected cases. Most sources cite 2.5 week average recovery period, so we'll use that.
As of 6 April 2020, this gives us these reference points:
-
Inflection point: 26 Jun – 10 Jul, with 2.9 – 11.8 million infections. This is when the growth rate peaks, and begins to slow down.
-
Active case peak: 5 Jul – 19 Jul, with 2.4 – 9.6 million active cases. This is when the recovery (or fatality) rate overtakes the growth rate, and the number of cases begins to go down.
Zooming in on the shorter term:
On 23 March, South Africa declared a 3 week national lockdown from 26 March to 16 April 2020.
Projections for this lockdown period, as of 27 Mar 2020 (with cases confirmed and projected):
| Date | Lockdown | Cases |
|---|---|---|
| 26 Mar | Week 0 | 927 |
| 2 Apr | Week 1 | 1,462 |
| 9 Apr | Week 2 | 2,743 – 2,744 |
| 16 Apr | Week 3 | 5,478 – 5,480 |
-
Data Repository and Dashboard by the Data Science for Social Impact Research Group @ University of Pretoria
-
Dashboard by Wits University, iThemba LABS, and DataConvergence.
-
Dashboard by The Awareness Company
-
Google Sheet by Jon Luies
- Covid Trends by Aatish Bhatia and MinutePhysics