InfoMullet: (3/24 Update!) COVID curiosity. Outbreak or containment and why #FlattenTheCurve matters.
TLDRUpFront: (3-28 UPDATE) Italy, Spain, and France are all set to exceed China in confirmed cases within the next week. And the United States will surpass even those countries in the next 1-2 weeks with the most confirmed cases globally. Death rates in Italy, Iran, Spain, and France continue to increase as the decline in the US halt, indicating a potential reversal and increase in the week ahead. We use WHO Daily Situation Reports to compare and monitor over 40 countries across 9 regions including breakouts of Nordic and GCC regions that were previously aggregated. Bookmark this page and revisit as we’ll be updating data every few days.
FullContextInTheBack: The last InfoMullet article on COVID-19 introduced two mental-models on COVID-19 contagion of “outbreak” and containment. These mental models are vital for understanding what #FlattenTheCurve means. An “outbreak” is a spike in cases that overwhelms local services before burning itself out. While “containment” is a suppression of cases over time in order to allow local resources to better handle.
With more complete data from a WHO investigation of the Chinese outbreak and additional research by the University of Washington we can now locate a threshold zone of risk for the outbreak mode. This allows us to chart countries by region for both total cases and daily rate of change in new cases, comparing them to the China “outbreak” behavior.
And we are now seeing in Italy the consequence of failure to #FlattenTheCurve as it’s death rate spikes above 7%. This is likely caused by where the outbreak began as a cluster – in suburban and outlying areas in northern Italy. The rapid spike in cases led to an inability of hospitals to tend to patients resulting in the higher death rates. This is similar to China – where the initial contagion was clustered in Wuhan city and Hubei province. But China’s death rate began at 3%, dropped, then rose to close to 4% over time. This could be because of measures or due to under-reporting as discussed in our previous article.
This clustering effect of geography has implications to #FlattenTheCruve for other areas where initial outbreaks were clustered such as Seattle, WA in the United States. In addition to flattening the curve over time, distributing patients to areas less-hard-hit is a way of flattening the curve over geography. This can be done by bringing resources from less-hard-hit places into the contagion cluster: like sending extra firefighters in during a bad fire season. It can also mean exporting patients out to areas with less burden on their system.
Flattening the curve through geography, rather than time, isn’t without risks. Critical patients don’t travel well and movements into or out of the region create new vectors of contagion. In addition, a panicked population listening to the wrong mental-model may resist supporting nearby areas to #FlattenTheCurve. This is why maintaining a sober, but rational, information distribution to the public is key. Even if stoking fears may create a short term benefit of provoking mass social-distancing, as seems to be the case in the United States, if those measures fail the residual fears are going to make #FlatteningTheCurve by geography a point of grievance, protests, and potential instability.
Understanding Confirmed Cases & Daily Rate of Change
This is a quick overview for a lay reader to understand the charts. Or you can skip reading this, go to a chart, post it on Twitter claiming it’s proof of the zombie apocalypse and my sister will get mad at me. So please don’t do that. For a fuller technical description of how we calculated thresholds of risk see “Even More Context in the Way Back” section after the regional data.)
For each Region we’ve displayed two charts for select countries: total cases and the daily rate of change. The charts are aligned across the horizontal axis so they all begin at the first day a case was confirmed. This helps compare the behavior of a contagion to another country even if they happened at different times. For example asking “How is my country faring in response or containment at this point in time versus China?” This isn’t just for curiosity sake – countries that fare better containing outbreaks as they happen such as Singapore and South Korea probably have methods we should look at replicating. Plus China appears to have gone through a full outbreak cycle of growth and collapse having now contained the spread. This took about ~65 days and gives us a lifecycle baseline to compare other countries too. Finally pay attention that in the total cases chart, the vertical axis is a logarithmic scale of base ten, so that countries with cases ranging from under 10 to nearly 100,000 can be displayed on the same chart. But don’t misread a line “half-way” up the chart as being halfway to 100,000 – it’s closer to 1,000. What this means is you could still add all the confirmed cases world wide and they probably wouldn’t equal half of what China has already experienced, which is an important sanity check to keep in mind.
Whether a contagion is being contained or faces an outbreak, or whether an outbreak is being contained can be understood by several means. First, the steepness of the line as it passes through the threshold of risk. If it is a steep line – as seen in China, Italy, and Iran – then chances of an outbreak that are difficult to contain are high. But if the growth line is mostly flat as it enters the threshold of risk or never reaches the threshold it indicates there may be containment. Both Japan and Singapore show this example.
Even if a country has had an outbreak behavior, when we begin to see the growth line in total cases flatten out and the daily rate of change drop, such as in South Korea, that’s a good indication the contagion has been contained.
Finally, although total cases tell us where we’ve been, the daily rate of change tells us where we’re going. If the daily rate of change is increasing upward that’s a sign the contagion is accelerating. But if it’s declining, that’s a sign the contagion is being contained. This data however can be very ‘noisy’ in the early stages when there are few cases, so a growth rate of 100-200% might occur if the country got its testing act together.
Understanding the Death Rate Chart (3/14 update)
This chart was added in the 3/14 update. It takes the WHO Daily Situation Report of reported deaths due to COVID-19 and divides them into the total confirmed cases for that day. This provides a dynamic measure over time of fatality rates related to COVID19. The purpose of this is to show that what is being described as a “death rate” in a static way is actually a dynamic number that adjusts due to local factors: susceptibility of local population, the ability of healthcare infrastructure to support surge in cases etc.
To read the chart notice that the vertical access is a percentage of the total confirmed cases that died by that day of the outbreak. It is set to only display 0-10% for greater clarity. The horizontal access is the days since the first confirmed case. Like the charts above this allows contagions to be relatively compared in different countries, even if they started at different times. Also to avoid noise the death rate calculation was only applied once a country had passed it’s 20th death related to the infection. Keep in mind the ‘death rate’ is of confirmed cases only and should not be taken as applying to the total population in a country.
Caveat Lector the Deuce
Same caveats apply as the last article. This is intended to improve mental models and not for peer review. Specifically we are only charting the data the WHO provides in it’s Daily Situation Reports.(1) This means if a country is actively suppressing information it may not look bad until it is bad. Lack of available testing kits may also make some countries look “better” than others that are testing more people, and catching more cases. This data should be treated as notional and illustrative. And keep in mind that even in the cases of an outbreak we know from China what that looks like and have conducted forecasts within ambiguity for the United States, an “outbreak” does not mean the zombie apocalypse has arrived. Just that containing the contagion is going to prove very difficult until it burns itself out.
COVID-19 by Region
North America (3/28 Update)
(3/28) The BLUF of 3/28 is that the Fire Break intervention in the US is showing a decline from the 30%-40% growth of last week down towards 20%. The problem is with 100k+ cases as of today, even a 20% growth rate means we will have 200K cases by Sunday/Monday, 400k cases by Wednesday/Thursday and possibly 800k cases by next Sunday. Assuming we can even test that much and there isn’t a steeper decline from the shift to Mobilization & Lockdown that now impacts over 240M Americans. Canada is also in trouble for having failed to seriously implement it’s own Fire Break, let alone a lock down model. At current rates of growth their case count is likely to begin to cause buckling of health care systems by Friday the 10th of April. The US healthcare system is already buckling in local regions such as NYC and Louisiana and that will only expand into the next week. The Death Rates of both Canada and the US appear to have begun their “fish-hook reversals” to begin climbing.
(3/24 Update) The United States will surpass Chinese cases within three days on the current rate, and many of these cases are clustered in New York City which may have 60K cases before the end of the week. With cases in all 50 States there is still significant room for the United States to continue growing meaning it may exceed all other countries in confirmed cases in 2-3 weeks. Both the US and Canada Rate’s of Growth remain on average 30%, meaning their confirmed case count will double every ~2days. There may be a point where there aren’t enough tests, or testing has been suspended in hot-spots such as Los Angeles and New York City, where the case count begins to artificially flatten.
(3/21 Update) The United States effort to #FlattenTheCurve with the Fire Break model appears to have failed as the growth rate accelerated and cases continue to grow rapidly. Over the weekend expect an increasing number of States to independently adopt the Mobilization & Lock Down method of Europe.
(3/17 Update) The data continues to show exponential growth in the United States and Canada, with a doubling period of 2-3x days. This doesn’t reflect the mass changes that were put in place late last week, as confirmed cases reflect infections of a week or more ago. But with this growth rate US residents should prepare for seeing news of 10,000 confirmed cases later this week and over 20,000 confirmed cases by the end of the weekend or early next week. Although Canada is at a smaller total level, it’s growth rate is similar and will likely double twice again at least by next Monday. Given the doubling rate in both countries, and measures by Europe, residents should not be surprised to see National Guard units mobilized by the end of the week as a proactive measure.
(3/10 Update) We’ll start with North America. As of today, it looks like the United States could be heading to an outbreak behavior. Bear in mind however that this aggregates all the United States into one country and with the geographic expanse we could combining several containable contagions into one number that looks at outbreak. On the other hand, conflicting guidelines on who should be tested, and a lack of testing kits for the early part of the crisis may mean that across the board cases are being systemically under reported.
(3/14 Update) It looks like the United States was unable to contain the contagion and an outbreak is underway. This is a consequence largely of (in-hindsight) incorrect policy choices made on developing and deploying sufficient test kits and conflicting guidance. (We discuss this more fully on an InfoMullet FB Live AMA now shared on YouTube.) On a positive note however the private and community sector of the United States has caught up with the risk leading to a wave of voluntary cancellations of non-essential events and shifting workforces, where possible, to work remote. Many European countries including Italy and the BeNeLux failed to take this step early enough – Germany just yesterday began announcing closures. However, because of the lack of adequate testing there is a real risk that confirmed cases in the United States are being under-reported.
Northern Europe (3/28 Update)
(3/24 Update) Northern Europe is beginning to decline in daily rates of change but still remains in a 10-20% range where the case count doubles every 3-7days. The lowest growth rates are in the Nordic and BeNeLux regions which is important because these smaller areas already have disproportionately high cases. Germany on the other hand is on track to surpass China in 1.5-2weeks at it’s current rate even as it declines in daily growth given already 30K people are confirmed infected.
(3/17 Update) There is little apparent effect so far in Lock Down & Mobilization causing a decrease in growth in France. The United Kingdom, having adopted an Induced Herd Immunity approach continues to rise. Only the Nordic countries have appeared to lower their growth rate sufficient to begin flattening their total cases into the more familiar S-Curve. Unfortunately, given their small populations, even a 10% growth rate implies a doubling period of 7 days meaning the smaller Nordic countries may be facing 20,000 cases by next Saturday.
(3/17 Update) Northern Europe continues to experience high levels of growth. Trailing behind Southern Europe, and Italy, by a week or two, they should begin preparing now for using geography and distributed networks to #FlattenTheCurve even as they implement social distancing. The United Kingdom has implemented an interesting contrarian policy that advises against isolation, at this point, which we’ll explore in an upcoming post comparing response options. It sounds counter-intuitive, but has a basis in rationale – though whether it will work is unknown.
(3/14 Update) The concerns over Northern Europe have manifested as all tracked countries and regions are accelerating through the outbreak threshold on steep curves. The question now is how will local governments enhance containment and self-isolation procedures while also addressing the immediate healthcare needs. This will be crucial in smaller countries such as those making up the Nordic and BeNeLux regions where, despite relatively better healthcare infrastructure, they have out sized total cases relative to their smaller populations.
(3/10 Update) Outside of China, Europe is struggling the most to contain the contagion. In Northern Europe Germany is crossing the threshold risk with BeNeLux (Belgium, Netherlands, and Luxembourg) and Nordic countries (Iceland, Finland, Sweden, Norway, Denmark, and Greenland) not far behind entering the same risk area. This is worrying because BeNeLux and the Nordic countries are a fraction of the size of Germany, so comparatively they are getting hit worse. However, they may also be reporting more cases because they have more testing kits and are being more expansive in their testing protocols – so bear that in mind.
Southern Europe (3/24 Update)
(3/24 Update) Southern Europe is on track to surpass China’s total confirmed cases in three separate countries: Italy by the weekend, Spain and France sometime next week. Even as Italy’s rate of growth continues to decline it is above 10%, and both Spain and France have daily growth rates in the 20% range.
(3/17 Update) The COVID19 contagion continues to rage through southern Europe with over 84,000 confirmed cases between Italy, France, Spain and Switzerland. The growth rates of all countries are declining but even a growth rate of 10% doubles every 7 days, meaning Italy, France, Spain are likely to each break China’s record for total cases before the end of the month. Both Italy and Spain show increasing death rates while so far France and Switzerland’s remain low for now.
(3/17 Update) Southern Europe continues to be the region hardest hit outside of China by COVID-19. Authorities in areas outside of Italy are beginning to sound the alarm bell and implement isolation policies. In both France and Switzerland the military has been mobilized both to help police prohibitions on gatherings but also prepare to help in case the hospital system collapses as it has in Italy.
(3/10 Update) Southern Europe, especially Italy, are struggling to contain the contagion. Italy’s recent measures of mass quarantines are being widely reported and the chart shows why. Italy’s growth entered the risk threshold and has already cleared it and is still growing strong. France, Spain, and Switzerland are also now in, or entering, the risk zone, all with steep growth which implies they will enter an outbreak mode.
(3/14 Update) Southern Europe across the board has failed to contain the contagion and is now facing outbreaks similar in scope/severity to Italy’s. The question now is can the governments react sufficiently to #FlattenTheCurve through measures related to geographic dispersion and allocation since the time-based approach appears to have been lost.
Focus on Italy (3/24 Update)
Because of the crisis in Italy and their role as the canary in the coal mine for other countries we’ve also included three charts that isolate Italy’s data from other Southern European countries. Comparing it only to China, these are included below.
The growth in total cases and corresponding increase in Death Rate (1st & 3rd charts) speaks to the failure to #FlattenTheCurve. That the Death Rate is dynamically increasing over time also challenges that this can be attributed solely to a higher age demographic in Italy. As age distribution of populations would be considered a constant (unchanging) relative to the course of an outbreak it’s probably not a full answer as the death rate continues to climb higher (changing). Add to this the recent data from Iran with a death rate of 5%, in a population that skews much younger.
What’s noteworthy is that the Italy daily rate of change (2nd chart) continues to decline over time. Though this seems good at first, the problem is the large number of cases Italy has already accumulated. Even if they manage to slow the doubling period from x2-3 days to x5-6 days, they’re “doubling” a much higher infected. Even if the daily rate of change continues to decline through the end of March Italy could easily crest at between 30k-70k cases depending on how fast the rate of change drops.
(3/24 Update) Although the declining rate of growth is noteworthy in Italy, in the charts below it shows how difficult this situation still is. Italy’s cases are near China’s total, and set to surpass them over the weekend. Their death rate has reached 10%, and their rate of change, though declining, still remains significantly higher than China’s at a comparable period in the life cycle. This means Italy’s cases will continue to double in number approximately every 7days, leading to nearly 130K cases by this time next week absent a sharp drop in daily new cases.
East Europe (3/24 Update)
(3/24) Slovenia and Czechia are showing promising declines in growth rate as they cross the risk threshold. Unfortunately, Austria has already crossed the risk threshold and a high growth rate and Russia is approaching on a similar trajectory, indicating that both countries will experience significant outbreaks in the coming 2-3 weeks.
(3/20) Eastern Europe breaks through the risk threshold in multiple countries as Russia begins it’s climb. Daily new case rate growth is declining in Austria, Czechia, and Slovenia but not enough to avoid Italy or French style total loads. Meanwhile Russia’s rate of growth, which had been suspiciously flat, is now close to 30% meaning confirmed cases there will double every ~2 days.
(3/17 Update) We’ve added coverage Eastern Europe by charting Austria, Czechia, Slovenia, and Russia. Unfortunately Eastern Europe looks depressingly like Southern and Northern Europe a few weeks later. It’s unclear whether this is a factor of insufficient testing (there are far more actual cases than confirmed) or simply a time delay of the contagion reaching Eastern Europe later. With a rate-of-change (2nd chart) of ~20-30% Eastern Europe looks set to double their cases every 2-3 days. Note that Russia’s behavior looks suspiciously like suppressed information up to recently when it began a steep climb.
South Asia (3/24 Update)
(3/24) All three countries are finally reporting cases, though it still appears that India and Bangladesh are significantly under counting. Pakistan’s rate of change has dropped to 10% as it enters the risk threshold, if they can continue to drop it they may be able to contain and have a trajectory similar to South Korea, assuming their reporting is accurate.
(3/21 Update) Only Pakistan appears to be reporting accurate information at this point, and it’s breakout pattern is typical of many other countries. India has a suspiciously low rate-of-change in daily new cases, indicating a managed testing program or selective reporting. Typically all countries, including South Korea and Japan, have high rates of change early in the outbreak and only begin to bring them in control later. Finally Bangladesh appears to be hiding its confirmed cases entirely, still reporting only 20 cases. This may be on account of inadequate testing, classifying COVID19 cases as something else such as ‘atypical pneumonia’ or government suppression of data.
(3/17 Update) Pakistan begins growing at an exponential rate of cases in this update indicating a likely future outbreak unless significant change occurs. Bangladesh continues to report less than 10 cases which is doubtful number. India’s growth has slowed. It shows only an average 10% rate of growth (2nd chart) which means a doubling period of ~7 days, if these numbers are accurate.
(3/14 Update) South Asia’s numbers remain very low according to WHO Daily Situation Reports. However, local reports are that, in Bangladesh, COVID19 cases have been classified as “atypical pneumonia” which means the infection may be much larger than reported. This presents an urgent problem for the Bangladesh government as miss-classification may have meant high rates of exposure to their healthcare workers. Exacerbating this will be a difficulty in obtaining, distributing, and administering testing test kits to understand where the contagion is, and where it might be going.
(3/10 Update) There aren’t many confirmed cases in South Asia. Singapore demonstrates an excellent example of a contained outbreak. And India, Pakistan, and Bangladesh are all reporting only a handful of cases. This could be good news, that a combination of unfavorable environmental and demographic conditions mitigates the spread. But given that these countries also have massive populations, often living in close proximity, and often with inadequate access to healthcare we’d caution against too rosy an estimate. It may be that the contagions are just now starting, or that cases and fatalities are not being reported, there’s not enough testing available, governments are hiding numbers – or a combination of all of the above.
East Asia (3/24 Update)
(3/24) East Asia continues to demonstrate containment, with now South Korea and China both demonstrating a sustained upper formation of the “s-curve” which indicates containment. Although the scale does not show it well, there are still additional cases occurring in the country, largely now from foreign travelers returning. Japan continues to grow in cases, but only at a growth rate of~5%. This means it will take it two weeks to double which is flat enough, even as it crosses the risk threshold, to prevent an uncontained outbreak and manage the curve.
(3/21 Update) If there’s any good news going on it’s coming out of East Asia. China, if we believe their data, continues to hold the case rate fairly stable. There are additional cases being reported, mostly from travelers returning to the country. The same holds true for South Korea. Japan, although it is growing, is doing so at a slow rate. Though concerns about testing rigor have led to questions on whether Japan is testing enough.
(3/17 Update) East Asia continues to show containment in China and South Korea and a successful #FlattenTheCurve approach in Japan. This is based only on WHO Data we have. There have been allegations raised that China’s s-curve is “too perfect”. If they are suppressing confirmed cases, then they’re also suppressing deaths because there haven’t been an increase in those either. Likewise South Korea shows a similar s-curve of successful containment. Ultimately, epidemics are hard to conceal over time and word will get it out. But for now we’re relying on the data as reported by the WHO.
(3/14 Update) East Asia continues to demonstrate successful containment strategies. Both prior to outbreak (in Japan) and post-outbreak (in South Korea and China.) The contrast between post-outbreak strategies between South Korea and China are useful for western governments to study in-depth rather than assuming China’s was the best possible choice. Additionally, given Italy’s reported death rate, there is serious question over whether China under-reported it’s death rate. This will have not just response questions to the rest of the world, but create lingering grievances that could lead to instability on the long-run for China, especially in Hubei, where initial suppression of information, heavy-handed measures, and incompetence has already created significant unrest in the local population hardest hit by COVID19. Note that even as Japan appears to be passing through the risk threshold, it’s rate of growth is low, indicating an incremental accumulation of cases rather than an exponential growth. This is a good example of #FlattenTheCurve in action.
(3/10 Update) The first region struck by the contagion, East Asia provides an excellent case study in comparing the behavior modes across the three countries. South Korea had an outbreak but appears to be successfully containing it at a level far below China’s, without resorting to authoritarian measures. There’s a lot of good lessons to learn from this. Japan doesn’t appear to be threatened by a breakout. Even as its cases climb towards the risk threshold it’s at a fairly flat incline.
Oceania (3/24 Update)
(3/24) All countries tracked in Oceania are now entering the risk threshold, though at lower trajectories of growth than seen in European, North American, and Chinese modes. All are now reporting growth rates between 7-15%, which still means a high risk but is much better than the 20-40% typically seen during this phase of the lifecycle.
(3/21 Update) Oceania continues to track upward growth into the risk threshold. Malaysia, Australia and Singapore are all entering the risk threshold for breakout. Malaysia has adopted a Chinese style People’s War national response while Singapore continues to implement a Pandemic Panopticon. Worryingly is that Singapore’s growth rate is beginning to accelerate. All three countries have flatter than usual trajectories relative to similar countries as they enter the threshold. Unfortunately Indonesia is accelerating into the risk area and continues to have a very high rate of growth.
(3/17 Update) We add Oceania coverage by charting Australia, Indonesia, Malaysia and moving Singapore from South Asia to Oceania. Although Malaysia and Australia’s total cases chart appear below the risk threshold, both have accelerating rates of change in growth (meaning new daily cases are increasing rather than decreasing, see 2nd chart.) This is concerning as both countries approach the threshold of risk.
Middle East (3/24 Update)
(3/24) Iran, if it’s numbers are to be believed, is beginning to show the turn into an upper s-curve behavior mode as their growth rate drops to 5% daily. Likewise the GCC is reporting a very slow growth. Turkey and Israel, by contrast, continue to show far too high growth rates of ~20-30% as they enter the risk threshold, indicating likely outbreaks of significant scale in the next 2-3 weeks.
(3/21 Update) We add Israel and Turkey to the Middle East chart. Turkey, having only reported it’s first confirmed cases recently, is shooting upwards in a near vertical trajectory. This is steeper than even Iran’s growth angle, which we suspected of being because of suppressed information. Turkey’s rapid rise may be for similar shenanigans or because they are testing very heavily. Israel is also added as it accelerates into, and likely through, the risk threshold. Iraq and GCC still seem to living suspiciously blessed lives with slower rates of growth than virtually any other country in the world given the relative time of the outbreaks. Additionally the GCC isn’t reporting a single death to COVID-19, a statistical improbability. As they have over a 1,000 confirmed cases between the six countries making up the GCC, and the virus kills 1-4% on average, there should be at least 10-40 fatalities now, but none are reported.
(3/17 Update) Iran, if their data is to be believed, is beginning to show a slight flattening in case growth. Both Iran and the GCC are reporting daily rates-of-change around 10% growth, meaning the doubling has been slowed to seven days. This may help the GCC exit the threshold of risk without a full outbreak. For Iran the challenge is they are already at 15,000 confirmed cases, and their death rate has climbed to over 5%. Even if it only doubles every 7 days it’s doubling a much larger number, continuing to put strain on the healthcare system.
(3/14 Update) Iran continues to show the consequence of governments failing to adequately address early risks through suppression of information. The GCC is also spiking into the risk threshold for outbreak. It’s growth rate is very choppy, indicating potentially batched reporting that may not accurately reflect the nature of growth. This may be a problem of all three countries depicted: either inability to accurately report daily numbers, test sufficiently, or government efforts to massage information to manage appearances.
(3/10 Update) The most hard-hit country in the Middle East is Iran. Interestingly it’s behavior curve begins sooner in the outbreak cycle, and is steeper, than China’s. This may be because of local factors but could also represent the Iranian regime suppressing information until it could not be hidden anymore. The GCC stands for the Gulf Cooperation Council which includes Saudi Arabia, Kuwait, Qatar, Oman, Bahrain and the United Arab Emirates.
South America (3/24 Update)
(3/24) Brazil, Argentina and Ecuador are coming through the risk threshold band at very high rates of growth, indicating significant outbreaks in the next 2-3 weeks. This is especially concerning in Brazil with it’s large, dense, metropolitan populations that replicate conditions in New York City but without the same healthcare infrastructure.
(3/20 Update) The three South American countries we’re tracking of Brazil, Ecuador, and Argentina continue to accelerate up, and likely through, the outbreak risk threshold. All three countries are reporting growth rates of more than 30%, meaning the cases double at least every 2 days.
(3/17 Update) Brazil, Ecuador and Argentina all continue to grow in cases as they are approaching the risk threshold.
(3/14 Update) We’re starting to see increases of cases in Brazil. Indicating that the risk of contagion and possible out break needs to be taken seriously. Assuming there isn’t under-reporting, and the WHO numbers are accurate, the Brazilian government should be taking the lessons of other countries in Europe and East Asia to heart and acting now to seek to contain the outbreak.
(3/10 Update) We’ve added South America as a placeholder – but there’s not much to report yet as far as WHO Data is concerned. Brazil, Ecuador and Argentina have less than 50 cases between them and other countries are all in single digits.
Death Rates over Time (3/24 Update)
Below is our first pass on a dynamic death rate for countries that have recorded significant fatalities. The value is the percentage of infected that ended up dying to the virus. To reduce noisy values at the start of an outbreak, we only recorded values after a country had passed 20 fatalities. It does show interesting variation however between countries. Specific to policy makers and those seeking to advocate better responses from public and private sector actors.
(3/24 Update) Accelerating death rates in Italy, Iran, France, and Spain are now demonstrating this isn’t just an “Italian” problem. Likewise the United States declining death rate has flattened and may be entering a “fish-hook” reversal, similar to Spain’s before growing.
(3/21 Update) Iran, Italy, France, and Spain are all demonstrating an accelerating death rate indicative of failing to #FlattenTheCurve. The US, 7-10 days behind Europe, continues to have a declining death rate though that will likely show the “reversal-hook” by the next update on Tuesday.
(3/17 Update) Both Italy, and now Iran’s growth in the death rate are examples example of the failure to #FlattenTheCurve. Although other countries appear to be faring better with declining death rates, those have to be taken in context that they have not experienced similar levels of strain on their health care as they are 1-2 weeks behind Italy and Iran in terms of case load.
Even More Context in the Way Back: Determining the Threshold of Risk
In the first article we presented two hypothetical behaviors of daily new cases, shown below.
In the outbreak mode, in red marked #1, a contagion of daily new cases grows exponentially at a high rate over time, but then collapses just as quickly. For reasons discussed in our first article, the contagion burns itself out. In a containment mode, marked with #3 , daily new cases grow more slowly and diminish more gradually. The area under either curve represents the total cases, which when plotted would appear as an S-Curve.
We’ve now seen China go through this entire cycle as the first nation to deal with COVID-19 and can plot their actual data to recreate the hypothetical charts above.
One piece of key data we were missing until last week was where to locate a threshold of risk to understand whether a contagion would manifest in containment or outbreak mode. Researchers at the University of Washington & the Fred Hutchinson Research center however compared Seattle, WA to Wuhan (based on WHO Data) and found:
“We know that Wuhan went from an index case in ~Nov-Dec 2019 to several thousand cases by mid-Jan 2020, thus going from initial seeding event to widespread local transmission in the span of ~9-10 weeks. We now believe that the Seattle area seeding event was ~Jan 15 and we’re now ~7 weeks later. I expect Seattle now to look like Wuhan around ~1 Jan, when they were reporting the first clusters of patients with unexplained viral pneumonia. We are currently estimating ~600 infections in Seattle, this matches my phylodynamic estimate of the number of infections in Wuhan on Jan 1. Three weeks later, Wuhan had thousands of infections and was put on large-scale lock-down.”(2)
Note they are describing estimates of actual cases, not confirmed. Using their timeline however we took the total confirmed cases reported in China by WHO along their timeline and located the band centered at 800 confirmed cases in the top chart and 40% growth rate in the bottom chart. The band extends up and down to represent both uncertainty and differences between locations. It’s a notional indication at best, but looking through the charts it does seem that those countries which cross the band at a steep line end up in an outbreak pattern, while those who avoid it or enter it on a flatter growth pattern have contained the contagion.