Wednesday , July 15 2020
Home / Naked Capitalism / Why Did Covid-19 Spiral Out of Control?

Why Did Covid-19 Spiral Out of Control?

Summary:
Yves here. It’s worth remembering that medical professionals and officials were successful in limiting the spread of the earlier SARS and MERS viruses. Why did Covid-19 instead become a pandemic? Ignacio picks apart some of the key lapses. I wonder how much of the disease spread was bad luck, with the infections taking hold in Hubei right before the Chinese New Year, putting the central government in the position of having to shut down holiday travel and partying (meaning spending) if it had acted promptly. Recall also the refusal to shut down a large official banquet in Wuhan during the window when it might have been possible to contain the disease. This Associated Press story supports Ignacio’s analysis: In the six days after top Chinese officials secretly determined they likely were

Topics:
Yves Smith considers the following as important: , , , , , ,

This could be interesting, too:

Lambert Strether writes 2:00PM Water Cooler 7/14/2020

Jerri-Lynn Scofield writes COVID-19 Mistakes to Avoid: Don’t Conflate Contact Tracing with Contact Tracing By App

Jerri-Lynn Scofield writes Links 7/14/2020

Jerri-Lynn Scofield writes Save the Tiger, Shift to other Big Cats: Illegal Wildlife Trafficking and the Rise in Zoonotic Diseases

Yves here. It’s worth remembering that medical professionals and officials were successful in limiting the spread of the earlier SARS and MERS viruses. Why did Covid-19 instead become a pandemic? Ignacio picks apart some of the key lapses.

I wonder how much of the disease spread was bad luck, with the infections taking hold in Hubei right before the Chinese New Year, putting the central government in the position of having to shut down holiday travel and partying (meaning spending) if it had acted promptly. Recall also the refusal to shut down a large official banquet in Wuhan during the window when it might have been possible to contain the disease.

This Associated Press story supports Ignacio’s analysis:

In the six days after top Chinese officials secretly determined they likely were facing a pandemic from a new coronavirus, the city of Wuhan at the epicenter of the disease hosted a mass banquet for tens of thousands of people; millions began traveling through for Lunar New Year celebrations.

President Xi Jinping warned the public on the seventh day, Jan. 20. But by that time, more than 3,000 people had been infected during almost a week of public silence, according to internal documents obtained by The Associated Press and expert estimates based on retrospective infection data.

Six days.

That delay from Jan. 14 to Jan. 20 was neither the first mistake made by Chinese officials at all levels in confronting the outbreak, nor the longest lag, as governments around the world have dragged their feet for weeks and even months in addressing the virus.

But the delay by the first country to face the new coronavirus came at a critical time — the beginning of the outbreak. China’s attempt to walk a line between alerting the public and avoiding panic set the stage for a pandemic that has infected more than 2 million people and taken more than 133,000 lives.

As Ignacio points out, there is plenty of blame to spread around. Other countries’ failure to stop or severely restrict travel from China (for instance, imposing strict quarantines on arriving passengers) was another big fail.

By Ignacio Moreno Echanove, an epidemiologist

These days, lots of analyses, that retrospectively try to explain Covid-19 outcomes in different places are, I believe, quite faulty. Articles assert that this country has done far better than that other country, frequently on the basis of a single snapshot of data.

My favourite expression that I borrowed from NC commenter David, is that we are immersed in tunnel vision, overwhelmed by the enormous amounts of info coming to our radar through all possible channels.  Much of the information and analyses are, again, flawed, ideologically driven, and in the worst cases totally disingenuous.

I thought it would be useful to perform some additional analysis. Instead of relying on snapshots, one should carefully analyse the cascades of events resulting in the pandemic with a hierarchy. I soon realised this was quite a hard job so I have modestly tried to scrutinize a few of these events.

From an epidemiological perspective, the data points to trace should be contagions, particularly early contagions in each region. Then, using estimates on the reproductive rate of SARS CoV 2 (R0), hospitalization rates, ICU entry rates, mortality (M) trace the progress of the epidemic.  And do it with real-time data!

Since this information is not available, the published data have to be cured and lots of assumptions have to be made in the process of tracing.

By now, everybody knows that the number of confirmed cases is not a metric in both the time dimension (because gaps between contagion events and test reporting) and in quantitative terms (because the rate of tested/infected is almost always and everywhere much closer to 0 than to 1). That ratio will vary with time and geography, so confirmed cases is an unreliable source for epidemic tracing. This is truer during the early days of an epidemic, when events are critical.

Tracing casualties does a better job in quantitative terms but it is true that this information is also unreliable for some reasons. There may be political reasons to hide numbers or at least not to push for accuracy, and Covid-19 associated casualties are reported in most cases only if there is a test result demonstrating SARS CoV 2 infection. A large disease outburst will result in diagnostic and clinical systems very much overwhelmed with serious under-reporting of cases.  Finally, to make estimates using deaths you will have to make assumptions on mortality rates.

Another data source is the recording of registered deaths in countries that report these in a daily basis (MoMo, Spain MoMo as an example). Of course, this is not all about Covid-19 but you can trace “excess deaths” compared with “normal” expected deaths and assuming the excess was driven mostly or totally by the epidemic, in a few instances you might come with better quantitative estimates. A mortality spike can be used to trace the progress and extent of an epidemic, identify when contagions peaked etc. I will try this and subject the results to critical examination.

If one is to use casualties or mortality to estimate “real-time” contagion events, we need to know first the approximate lag between contagion and death events.

In China it was reported that median time from contagion to symptom onset is about 5 days (large dispersion with a long tail, so better using median than average). The median lag between symptom onset and death was estimated to be about 19 days (again high dispersion and even longer tail) so I use 25 days as a proximate contagion-to-death lag.  Official data in Spain validate the lags observed in China almost exactly.

Then, if there is available good data on mortality rates (M) it is possible to estimate numbers of contagions  in a given day resulting in  deaths about 25 days later. In Spain the mortality rate of Covid-19 has been estimated at about 1.3% on the basis of serological surveys and casualties though I don’t know if they used reported casualties or if they estimated M using MoMo “excess deaths” data. If they used reported casualties, the, real value of M in Spain would be closer to 1.6% once unreported “excess deaths” are added.

I compiled data from MoMo for Spain and using the official mortality rate (1.3%) estimated the number of contagions occurring about 25 days earlier (“real contagion time”, red line in next graph). Then, and supposing Covid-19 R0 = 6 (it has been estimated at 5.6 in some outbreaks and) we can trace events backwards up to patient 1-10 (blue-sky line before the red line in next graph).

The graph compares these estimates with reported cases as per WHO reports (dark blue) which I also represent 12 days backwards when the contagion events leading to such confirmed cases could have occurred approximately (brown line). So, the brown line would represent a picture of possible “real-time” contagion events based solely on official confirmed cases. Bear in mind the numbers depicted in the graph are in logarithmic scale.

Why Did Covid-19 Spiral Out of Control?

This tells a story on how events possibly unfolded in Spain and something similar could be done for countries like Italy, France, Belgium, the Netherlands, or the UK.MoMo data allow to conclude that in Spain the peak of daily contagions occurred between the 2nd and the 6th of March and then declined nearly steadily until and after the emergency and lockdown were enforced by the 14th of March. The peak of contagions is estimated at about 110.000 per day using M = 1.3% but would be much lower, about 42.000 contagions/day, if M = 1.6%.  Before this peak, by February the 26th, the R0 was already dropping sharply as the slope of the red line suggests, and continued to drop into the shadowed lockdown period (remember, this is logarithmic).

So, something was changing spontaneously before the government response was enforced. In my opinion, what was changing was public awareness on the risks of Covid-19. After this, the curves in different countries could diverge in response to the milder or stricter rules imposed.

Conclusion 1Most lockdowns were late and reactive measures that didn’t avoid the worst but only accelerated partial clearance of the epidemic. Alarm felt by the population at large was probably the first force to reduce R0 by ways of self-isolation of symptomatic individuals in countries like Italy, Spain, France the UK and in the State of NY. The “let’s flatten the curve” message was issued very late in such places when the curve had already been flattening spontaneously for some time.

The distance between the brown line (confirmed cases, moved back 12 days) and the red line (estimated contagions), is more or less equal to three orders of magnitude in early days and then drops. This shows that the detection systems in place were absolutely not up the challenge.

Conclusion 2: Lack of preparedness was severe. One cannot get ready for a challenge like this in a couple of weeks or even in a month. You need to have equipped and trained teams for sampling and testing with validated methods all around the country, fever check-points etc.

Conclusion 3: This severe lack of preparedness was probably common to most European countries, but some coped much better than others with Covid-19 suggesting that in the absence of proper preparedness risk awareness, should have done the trick.. Being late in the pandemic probably helped a lot to increase awareness at all levels. Probably, one can speculate, public awareness was inversely related with the R0.

Last but not least, the sky-blue line representing estimated daily contagions backwards assuming R0 = 6 suggests that the SARS CoV entries that resulted in the outburst in Spain could have occurred by the end of January. We should give that estimate ample margin and guesstimate that those entries possibly took place somewhere between Jan 15th and Feb 5th.

Conclusion 4: flights from/to China should have been grounded globally by Jan 15th at the latest. Scientists had long ago estimated that by mid-January SARS CoV 2 was already in Italy, unnoticed but starting the cascades of events that led to the first large European outburst.

Next graph shows a time course of excess mortality “Z score” composed with maps from 2020 week 6 to week 15 according to the MoMo web in Europe. Not all countries report, only those in grey-blue.

Why Did Covid-19 Spiral Out of Control?

The table below the maps is mine. It summarizes the number of weeks under “extremely high”, “very high” and “high” mortality rates, plus the max “Z score” value and the first week when excess mortality is detected by this metric.

In between those countries showing clearly positive Z scores, Switzerland did the best while the UK did the worst relegating Spain to an honourable 2nd worst position. Scotland probably benefited from suffering the latest outburst with about three-week gap to become aware of what was coming. Sweden had two weeks to improve awareness and this is probably the reason they could avoid extremely high Z scores compared to other countries that showed earlier unnoticed outbursts. You can see that countries with a more ‘liberal’ approach (UK, Netherlands and Sweden) showed excess deaths lingering up to week 15 (and possibly later). When this graph was done, not all data for weeks 14-15 was gathered and the definitive results could vary with current data.

If one is to select a country that did well, Norway, Finland, Austria, Hungary, Greece, Portugal, Denmark or Ireland would be much better examples than Sweden. Switzerland, that was earlier than Sweden in the epidemic did better by all those metrics suggesting much better reaction by both, the leadership, and the population.

In those maps above the absent elephant in the room is Germany represented only by the states of Hesse and Berlin which didn’t show significant mortality spikes by the Z value.

Germany, as WHO accounts show, was amongst the first countries in Europe to report confirmed cases in significant numbers. Anecdotally, the first positive case reported in Spain was a German citizen and I find this very telling. My opinion is that German citizens were more aware than Spanish on Covid-19 risks and more prone to report their symptoms, isolate themselves etc. Besides, Angela Merkel was famously said to trust her actions on German institutions experienced in epidemiology and this country was the first to confirm home contagions and to report untraceable community transmission. This suggests both better preparedness and awareness than the other large hyperconnected countries in Europe.

Again, digging in speculative territory it could be the case that political stability, or a government that had been standing for long could possibly help for better reaction. Macron hadn’t been that long the President of France (less than 3 years), neither Conti in Italy (nearly 2 stormy years), not to mention the inexperienced Johnson and Sanchez heavily immersed in difficult internal and international affairs. Mr. Rutte in the Netherlands had been governing nearly for 10 years but possibly had not as good reaction as Merkel. It might be the case that German preparedness can be traced back as long as to 1967 when a (then) rare haemorrhagic disease (Marburg Virus, Ebola-like) erupted in Hesse infecting 31 and killing 7. Who knows?

Back to the initial outbreak in Hubei province: My initial thinking about Hubei was that given it all started in Wuhan at an explosive rate, one could expect initial stupor resulting in not the best communication of risks. Two critical data items should have been be obtained ASAP: identification of the causative agent of the disease, and determination of possible human-to-human spread.

The identification and release of SARS CoV 2 genome sequence was achieved by January the 3rd, 5 days after the alarm sounded, fairly enough fast to start tracing the new disease soon.

By January the 7ththe new pathogen was confirmed as the causative agent of the new disease. Regarding human-to-human spread, critical for switching the Precautionary Principle on, it was not officially confirmed until the 20th of January,  way too late and when it had already been reported in some Chinese provinces,  Thailand, S. Korea and Japan and probably had already spread to many other countries.

Only ten days later, the US-CDC was confirming person-to-person transmission within the US. This, was way too late, as we can retrospectively say, to prevent global spread. The contrast between the relative promptness with the identification of the pathogen and the sluggish reporting of the other crucial epidemiological fact was stark.

Was this crucial info withheld until the international spread of the disease helped, for instance, to rule out a blanket travel ban from/to China?

Or, is it true that thorough confirmation was needed before ‘unnecessarily frightening’ the rest of the world?

The fact is that reading both the Ch-CDC report (Feb 2020) and a study published by March 26th, it can be concluded that human-to-human spread was strongly suggested by early data. Figure 3 from the study linked above shows that 5 out of 12 cases detected in December and grouped in 4 different clusters had NO exposure to the wet market in Wuhan, but exposure to other cases.

By the time the genome sequence was published, it should have also been announced that person-to-person transmission was, to say the least, very likely.

In fairness, though it might be true that Chinese bureaucracy delayed the release of crucial information, it should be admitted that an earlier notice might have met with deaf ears in many places around the world. I haven’t found any link showing foreign authorities forcefully asking for evidence or likelihood on human-to-human transmission, or lack thereof, before January the 15th. If anyone knows better, particularly Japanese, Vietnamese, S. Korean or German readers, please comment. This is not to say there wasn’t a clamour for better and timely release of information by Chinese authorities  from concerned scientists but not only scientists and through social media. As an example, this report titled “Pressure builds on China to share info on new coronavirus” and published on January10th was possibly issued hen to try to prompt a better response.

At this point, it would make sense to try to trace Hubei’s outbreak to its origin in order to check early responses and detection methods but there is no adequate data. Important questions about the origin of the disease are still unanswered and have been discussed in a previous post.

Yet we can try to trace the international spread of the virus and check downstream responses for mistakes.

First, we already have identified here one important point that was shared worldwide: most authorities in relevant positions, including the WHO, showed a lack of proper awareness of pandemic risks. Everyone was more worried about immediate effects of bold measures (like closing airways, railways and roads to human traffic from/to China) than the unknown risks of a pandemic.

As an example the head of the WHO was still ostensibly failing in his risk assessment by February 5th. In some sense he was right: By this date, SARS CoV 2 had already spread all around the world making a blanket travel ban unnecessary. By January 31st many countries started grounding flights to China but as we know this was already a late reaction that ended with worse consequences resulting in the grounding almost all fleets between 13-31st  of March.

The E-CDC Technical Advising Council enclosed for a two-day meeting (Feb 18-19th) near Stockholm, about two weeks before the peak contagion period in places like Lombardy and when so far only 45 cases had been confirmed in Europe. The minutes of the meeting have been recently commented upon at El Pais (in Spanish). They reveal too much casualness about Covid-19. By this time ECDC protocols for testing were still restricted to Wuhan-related patients. This allowed for SARS CoV 2 having a wild undetected run that led to the, so far, worst outcomes we have seen with the exception of the outburst of NY a week later.

In defense of the ECDC it has to be said it is just a consulting institution without real responsibility. According to the minutes only the Danish representative prompted to search beyond Wuhan links and extend to any pneumonia. The Finish representative answered it was impossible, “everyone will be demanding tests and we don’t have enough”. This is telling; most representatives were focused strictly on clinical outcomes and forgetting about epidemiology. Only the German representative announced they had already delivered test kits to 20 hospitals and were well ahead of the rest on tests already done. Everyone was reporting difficulties purchasing clinical equipment and only Ireland stated they had already declared HC emergency and purchased in advance.

Other Asian outbreaks: tracing the epidemic in Southwestern and Pacific Asian countries would indeed be of high interest, particularly given the relatively mild outcomes in these countries that suggest more or less appropriate responses. A lot has been said about S. Korean response, less about Vietnam, Singapore, Taiwan or Japan. Unfortunately, studying separately each case would have required a lot of effort.

Though I somehow feel here out of my depth I think it can be safely concluded that there was at least one common feature to these countries: awareness on the risks of local outbreaks. Both, at the level of public authorities and citizens, awareness was widespread and this helped to keep the R0 low in these countries without the need of painful extended lockdowns. Besides, as in the case of South Korea, there was preparedness in the form of pre-existing alarm and early detection systems, as well as protective equipment etc. Their awareness was a fully internal thing that did not transpire into other foreign countries tenaciously focused on ignoring the risks. I don’t know whether they tried to give timely advice or how forcefully but this says something about the international state of affairs: Mutual trust and confidence are very much lacking.

Awareness of the population at large: Once the leadership had failed in so many places it is worth looking if the public response made any difference. As a proxy I used Google Trends and entered “Covid” as the searching term. This could somehow indicate how the population at large was aware of the risks and searching internet for information. The results are fairly consistent with previous observations.

Why Did Covid-19 Spiral Out of Control?

The three graphs in the left show the results obtained for a few Asian (upper graph) European (middle) and US states (lower) using Germany (the yellow lines) for comparison in the three graphs.

As expected, Japan, Taiwan or South Korea were faster to worry about Covid-19, suggesting these were more ready to increase social distancing, self-isolate if showing symptoms and accept alarm bells. Germans showed more prone to get informed than the rest of Western countries, though behind the Asian countries. It is surprising to find some initial interest in Italy in line with Germany that somehow faded later until it was too late.

The dates in red (table right to the graphs) show late reactions in countries/states that suffered the largest outbreaks probably guided by the realisation that something was wrong and when contagion peaks had already been passed. Ireland showed a jump in awareness in advance of hard-hit EU countries and I wonder what prompted this.

Regarding the US, Washington State was the canary in the mine that nobody dared to care about probably because the outburst was limited to a few counties there. Bad luck in some nursing homes!  A first large wave of reactions can be noticed in the US –represented by WA, CA, TX, IA, FL and NY in the graph– driven probably by news from the other side of the pond and a second wave almost certainly related with the outburst in NY that probably helped most of the rest of the US to keep the R0 below max levels seen there.  Public awareness possibly helped to keep many countries/states relatively safe as have been shown here using different approaches. Note that this metric is not about absolute search numbers but relative to a peak that has the value of 100.

One can conclude that Google Trends is a good tracker for “interest” but not necessarily suggest awareness. It was the job of the leadership and their councils to inform us about the risks of Covid-19 but most failed miserably.

As a concluding remark, I believe that future outbursts won’t be as explosive as the worst we have already seen given we have been primed to have good reaction and better detection systems are in place. I think that those tough lockdowns can and should be easily avoided.

Leave a Reply

Your email address will not be published. Required fields are marked *