The 2020 Novel Coronavirus Outbreak

South Korea’s Fatality Rate of COVID-19 Is Not A “More Accurate” Estimate

A March 5 article suggests that the case-fatality ratio of coronavirus is closer to 0.6% based off data from South Korea. However, the calculation ignores the huge problem of time lag.

The Article

A March 5 SCMP article writes:

Coronavirus: South Korea’s aggressive testing gives clues to true fatality rate

With 140,000 people tested, the country’s mortality rate is just over 0.6 per cent compared to the 3.4 per cent global average reported by the WHO

Various factors can influence this percentage, but scientists agree that all things being equal, it is more accurate when more people are tested

The key points being:

  • South Korea is detecting the milder cases through widespread testing, while other countries are missing these.
  • Therefore, South Korea’s 0.6% case-fatality rate is more accurate than the 3.4% reported by the WHO.

And to fully clarify, we look at the way this is calculated:

Yet in South Korea, where the country’s Centres for Disease Control and Prevention on Thursday reported 6,088 cases and 40 deaths, the mortality rate appears to be hovering around 0.65 per cent.

Where dividing the total deaths by the total cases for March 5 gives 40/6088 = 0.65%.

The Problem

Confirmed, Recovered And Death Count of COVID-19 cases in South Korea. Source: JHU

The case-fatality rate tells us the proportion of deaths among infected people. It splits the confirmed cases into two groups — those who recover, and those who don’t.

Now look at the data for March 5: 6,088 confirmed cases, 41 deaths and 35 recoveries. The problem is that at March 5, there are 6,012 cases (99% of all confirmed cases at the time) without a known outcome. You cannot calculate a case-fatality rate based off these figures. Doing this assumes that all active cases will eventually lead to recovery, but this underestimates the true case-fatality rate.

It is true that South Korea has better surveillance than other countries. To get a better idea of the fatality rate, we would need to wait a few weeks, until the cases have a known outcome. Since many of the cases were detected after February 19, and it takes 3 weeks on average for the outcome of a case to be known, we’d need to look at the figures of only these 6,088 earlier cases, at a time around late March.

Who knows what the true case-fatality rate is? It might turn out to actually be closer to 0.6%. But this can’t be concluded based off the methodology presented in the article.

March 9 Update: Why Is South Korea’s Rate Lower Than Italy’s?

Others have claimed that South Korea’s lower fatality rate is a result of good surveillance detecting mild cases. I offered the alternative explanation that South Korea’s confirmed cases are more recent, thus patients have not reached an outcome yet.

Why does Italy have a higher case-fatality rate of 3.8% then?

  • Are they failing to detect mild cases? (Detection issues)
  • Are the cases they detect merely older, thus reaching their outcomes? (Outbreak started earlier)
Confirmed, Recovered And Death Count of COVID-19 cases in Italy. Source: JHU
Confirmed, Recovered And Death Count of COVID-19 cases in South Korea. Source: JHU

Comparing daily increases in South Korea and Italy, we see greater daily increases in deaths and recoveries for Italy in the period around late Feb to early March, which suggests an older outbreak. However, there is something strange. For example, there is a spike of 42 recoveries on Feb 2, but around 3 weeks prior (Feb 6), there are only 2 confirmed cases. It is clear that surveillance in Italy is poor, and the number of confirmed cases is too low, thus inflating the fatality rate. Unlike Italy, South Korea doesn’t seem to have this issue as much.

But it is also worth seeing how the outbreak actually evolved in South Korea.

South Korea has seen a surge in the number of infections — adding more than 4,000 cases in less than two weeks — with around 60 percent of the national total linked to the Shincheonji Church of Jesus.

Source: Yahoo (March 2)

Most of South Korea’s confirmed cases came after the super-spreader event at the Shincheonji Church of Jesus, which was around ~Feb 19. Since it takes 3 weeks until an outcome is reached, this means that many outcomes should be determined at around March 10~14.

Fatality rate = deaths / confirmed cases.

It is true that in general, proactive testing plays a part in lower fatality rates. It seems like Italy is not detecting the true number of cases, thus inflating the fatality rate.

However, I still think it is an error to say that proactive testing in South Korea has revealed the “true” fatality rate. For the moment, it has lowered the fatality rate, as it uncovered the cases around the Shincheonji super-spreader event, which would have otherwise been hidden by their secretive members. However, we don’t know if these cases are truly mild, or if these cases will be more severe but have simply not reached their outcomes yet.

We will get a better idea in the coming days.

April 9 Update: As I Predicted

On March 5, the case-fatality rate was estimated to be 0.65%, which I criticized, due to the poor methodology (99% of cases still active, no known outcome.)

One month later, on April 9, I checked the dashboard again.
204/10,423 ~= 1.95%.

It is 3 times greater than what was earlier stated in a media. Importantly, 10,219 / 10,423 = 98% of cases have a known outcome, so the data is far more reliable here.

Was South Korea overwhelmed? Not nearly as much as other countries. And they achieved 1.95%.

I’m concerned.

My other article: how the case-fatality rate is miscalculated, and why it is not important to understanding the severity of the outbreak.

Math, stats, data. Influenced by the complex systems perspective. I prefer to take the critical view.