©2019 by Cassandra Capital LLC.

The information provided on this website is general in nature only and does not constitute personal financial advice. The information has been prepared without taking into account your personal objectives, financial situation or needs. Before acting on any information on this website you should consider the appropriateness of the information having regard to your objectives, financial situation and needs.

COVID-19 (Coronavirus) Outbreak  Tracker

This page tracks several important factors of the ongoing coronavirus outbreak.

 

There are several important factors to determine for the novel coronavirus to allow us to predict how the outbreak will unfold. The case fatality rate (CFR), the  R-nought (R0) value of the virus, the hopitilization rate, and the percentage of the population expected to be infected.

The Case Fatality Rate (CFR)

NOTE - As the epidemic has spread, several countries with have reached the limits of their testing capabilities. In the US, the CDC shipped faulty test kits to the states; Italy will now only test "at-risk people showing symptoms of COVID-19"; And France has retained strict limits on who can be tested. Because of this restriction in testing, severe cases are more likely to be discovered, and recoveries missed skewing the CFR higher.

 

Therefore, I have briefly paused calculations while I consider how to further restrict the dataset.

There are several different methodologies for calculating the CFR of an emerging epidemic. However, all of the calculations I make here are based on cases from countries who scored at least 50 out of 100 on the 2019 Global Health Security Index's measure of their ability to detect and report emerging epidemics. This excludes data from mainland China and Iran along with many other countries. Although this excludes a significant amount of cases, it is arguably more accurate given that Chinese data is particularly unreliable  and the the Iranian healthcare system is almost certainly missing recovered cases. This exclusion also allows us to factor in a greater proportion of mild and asymptomatic cases that the nations with highly ranked health surveillance are best able to detect. This keeps the CFR from being skewed higher.

Methodologies
In Methods for Estimating the Case Fatality Ratio for a Novel, Emerging Infectious Disease several methodologies are for calcualting the CFR of an emerging outbreak are we are currently experiencing.
They found that two methods were the most reliable and accurate.
  • The resolved cases method (simple estimate 2 in the paper) with the formula e2(s)=D(s)/{D(s)+R(s)} where, D(s) and R(s) denote the cumulative number of deaths and recoveries.

  • And the Kaplan-Meier method

I want to made a quick note about a flawed methodoly that should not be relied on  - particuarly for this novel coronavirus. The so-called Naïve CFR (simple estimate 1 from Methods for Estimating the Case Fatality Ratio for a Novel, Emerging Infectious Disease). Not only does our original study show that it can grossly underestimates the CFR while an outbreak is ongoing - a recent study specifically on the current novel coronavirus warns:

The time from the illness onset to death is also comparable to SARS [15], and the 15–20-day mean delay indicates that a crude estimation of the ratio of the cumulative number of deaths to that of confirmed cases will tend to result in an underestimation of the case fatality risk, especially during the early stage of epidemic spread. [emphasis added]

Current Estimates

Using the resolved cases methold:

14.14%(13.42%- 14.86% Confidence Interval: 95%) Current as of 2/29/20 6:00am EST

Using the Kaplan-Meier method:

*Estimate using this methodolgy coming soon.

Quick Points:

One of the hopes of people watching China’s coronavirus outbreak was that the alarming picture of its lethality is probably exaggerated because a lot of mild cases are likely being missed. But on Tuesday, a World Health Organization expert suggested that does not appear to be the case...

“So I know everybody’s been out there saying, ‘Whoa, this thing is spreading everywhere and we just can’t see it, tip of the iceberg.’ But the data that we do have don’t support that,” Aylward said during a briefing for journalists at WHO’s Geneva headquarters.

That being said, there is still an very good chance that there is a bias skewing the CFR higher because of preferential ascertainment of severe cases - just not at the order of magnitude that we hoped. See Potential Biases in Estimating Absolute and Relative Case-Fatality Risks during Outbreaks

  • Why has the CFR been rising lately? The data coming in has not been good! Optimistically, perhaps as the virus spreads to a new country the first thing they notice is deaths - missing recoveries and temporarily skewing the number higher? Pessimistically, I note that the estimates for SARS initially underestimated the CFR and rose to the true higher value with more data.  See the graph below from Methods for Estimating the Case Fatality Ratio for a Novel, Emerging Infectious Disease which examined how different methodologies for calculating CFR preformed during the SARS outbreak.

The R-nought (R0)

The R0 esimtate is the weighted average from collected pre-print studies (studies with calculations based on newer data are weighted higher).More details is provided in the Sources and Calculations section further down on the page.

 

Current Estimate:

R0: 3.85(Lowest range from study R2, Highest Range R6.6)  Current as of 2/29/20 6:00am EST

  • Why is this estimate so high - what are other sources calculating for CFR? Excellent question. Firstly I'll point out that when I began estimating CFR using this methodology I was getting about ~3.5% for quite some time. My estimate has risen and I suspect other estimates (at least thoses baseed on data from outside mainland China) would rise as well based on today's current data.

 

That being said, there is a wide range of CFRs which have been calculated:

4.8%  Clinical characteristics of 50466 patients with 2019-nCoV infection

4.6% or 7.7% Real time estimation of the risk of death from novel coronavirus (2019-nCoV) infection: Inference using exported cases

1.37% Estimating the Case Fatality Risk of COVID-19 using Cases from Outside China

2.3% Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China

7% Estimating the cure rate and case fatality rate of the ongoing epidemic COVID-19

20% in Wuhan (with breakdown of healthcare system) 1% elsewhere Estimating the risk of 2019 Novel Coronavirus death during the course of the outbreak in China, 2020

18% in Hubei (with breakdown of healthcare system) 5.1% 5.6% or 1.2% Outside of mainland China depending on methodology Severity of 2019-novel coronavirus (nCoV)

 

However, some of these lower estimates are using an obviously flawed methodology. For example, the largest study, from the Chinese CDC which has been widely reported on in the press, Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China

It calculates the CFR by taking the existing number of confirmed cases 44,672 and divides it by the number of deaths 1,023. Giving them a value of 2.3%  However, this methodology grossly underestimates the CFR (See study showing this exact fact) by assuming that no additional deaths will occur within the currently infected. From a study on the epidemiological characteristics of the virus:

The time from the illness onset to death is also comparable to SARS [15], and the 15–20-day mean delay indicates that a crude estimation of the ratio of the cumulative number of deaths to that of confirmed cases will tend to result in an underestimation of the case fatality risk, especially during the early stage of epidemic spread

Eventually we will get enough good data to be able to reach a consensus estimate for the CFR. It is still early in the outbreak.

I will continue to update this page because I have yet to see another source estimating the CFR as the current data comes in - and excluding the somewhat dubious numbers from mainland China and Iran.

 

The Hospitalization Rate

The hospitalization rate is perhaps the most concerning metric of the outbreak that we have so far. With several reputable sources estimating that somewhere between 15%-20% of case require hospitalization and ~5% of cases require an ICU. Currently 9% of Italian cases are in the ICU.

It is troubling the consider what the CFR might rise to if healthcare systems become overwhelmed and patients can't receive the level of care required.

One only has to look at the difference in CFR's calculated for Hubei province in China where the healthcare system has become overloaded with cases. 18%-20% CFR.

 

Current Estimate:

18-20% Hospitalized

5% ICU

Another useful data point is the trend in confirmed cases detected outside of mainland China. We can see that this trend is still increasing.

Updated 2/28/20

It is also worthwhile to watch the level of quarantine and lockdowns in China. As of February 15th it was estimated that some 760 million people in China are under some form of lockdown.This seems to have decreased somewhat as of (2/26) although there is increasing skepticism around China's numbers. Check out the excellent map below (Source) which is the best visualization I've seen of the extent of the lockdowns in China.

 

Please feel free to contact me with any new relevant studies I've missed or thoughts on my approach!

Sources and Calculations

R0 Values:

The weighted average  R0 value is the average R0 values of the studies below, with each study weighted to lose 2% for each day earlier than the latest published study. E.g. a study based on data from 1/22/20  would be weighted 30% less than a study with data up to February 5th (15 days)

Preliminary estimation of the basic reproduction number of novel coronavirus (2019-nCoV) in China, from 2019 to 2020

Novel coronavirus 2019-nCoV: early estimation of epidemiological parameters and epidemic predictions

Transmission dynamics of 2019 novel coronavirus (2019-nCoV)

Reporting, Epidemic Growth, and Reproduction Numbers for the 2019 Novel Coronavirus (2019-nCoV) Epidemic

Report 3: Transmissibility of 2019-nCoV

Early Transmissibility Assessment of a Novel Coronavirus in Wuhan, China

The Novel Coronavirus, 2019-nCoV, is Highly Contagious and More Infectious Than Initially Estimated

Epidemiological and clinical features of the 2019 novel coronavirus outbreak in China

Analysis of the epidemic growth of the early 2019-nCoV outbreak using internationally confirmed cases

CFR Calculation:

CFR calculation is calculated from cases from countries who scored at least 50 out of 100 on the 2019 Global Heath Security Index's measure of their ability to detect and report emerging epidemics. Source for recoveries and deaths. This leads us to exclude data from mainland China (although I have included Hong Kong and Taiwan) and also leads us to reject the recent fatalities from Iran in our calculations.

In Methods for Estimating the Case Fatality Ratio for a Novel, Emerging Infectious Disease several methodologies are compared. Fortuitously, these methodologies are compared by how they performed during the 2003 SARS epidemic which, given SARSCoV2’s close genetic relationship to SARS, makes the calculations especially relevant.

Among the various methodologies, I believe there is a clear choice.

The formula: e2(s)=D(s)/{D(s)+R(s)} where, D(s) and R(s) denote the cumulative number of deaths and recoveries.

Why use this methodology?

Use of this formula performed well compared to the other methodologies during the various stages of the SARS outbreak. This methodology was “reasonable at most points in the epidemic.” For copyright reasons won’t show the graph here — but I urge you to check out Figure 3 from the paper which compares the success of the various methodologies across the timeline of the epidemic.

This methodology also has the added benefit that we do not need to know the date when symptoms began for the fatal cases — information which is not readily available for all cases at this time.

Confidence Interval and Margin of Error calculated by:

n = [z2 * p * (1 - p) / e2] / [1 + (z2 * p * (1 - p) / (e2 * N))]

Where:

n  is the sample size,

z  is the z-score associated with a level of confidence,

p  is the sample proportion, expressed as a decimal,

e  is the margin of error, expressed as a decimal,

N is the population size.

 

Daily and Cummulate Infected Numbers from the John Hopkins CSSE Tracker

Additional Interesting Studies:

The role of absolute humidity on transmission rates of the COVID-19 outbreak

Estimating case fatality ratio of COVID-19 from observed cases outside China

Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus–Infected Pneumonia in Wuhan, China

Reconciling early-outbreak estimates of the basic reproductivenumber and its uncertainty: framework and applications to thenovel coronavirus (2019-nCoV) outbreak

Estimating underdetection of internationally imported COVID-19 cases