Why is Asia outperforming the West: Seeking Explanations for COVID containment

Why pandemics matter
I have always been interested in pandemics, although, only casually. Anything that kills lots of people is of interest to me and in a way touches my field. Pandemics as the great killers of yore, are also important to study if you have a Pinkerian view of human progress, which is to say, if you believe that our terrible world is getting better, albeit slowly. An epidemic is a horrifying natural experiment that allows us to learn a lot, especially the need to be humble.

Misjudging the epidemic
If you had asked me in 2019 whether pandemic preparedness is important, I would have definitely said yes. I am not insane like the yellow man. In fact, as far as public health is concerned - excluding aging research - I am usually in agreement with Bill Gates. Not so much because he is a genius techno-capitalist (he might well be), but rather because his money has bought him very good expertise at the Bill and Melinda Gates foundation. So when he proclaimed in 2015 that we are not ready for the next big pandemic in this now famous TED talk it was not a surprise to me and anyone in the know.

If you had asked me whether we should listen to experts, I would have said yes. We should listen to appropriate experts, though. Coronavirus researchers, statisticians, infectious disease modelers and the doctors and other architects of the SARS and MERS eradication efforts. Not every random infectious disease expert has a worthwhile opinion and similarly not every lay person is wrong about SARS-CoV2 .

Beyond that, however, my performance in predicting this pandemic and how it plays out has been terrible. The first step to learn from your mistakes is to admit failure. I was wrong. What is more, the whole West underestimated SARS-CoV2 and we need to stop making excuses (see here for an admission by a German epidemiologist that we were under-prepared). Although, as I will discuss below we had good reason to think this outbreak may not be "the big one"; even so, we should have been prepared. This editorial, published in NEJM on the 28th of February, penned by Fauci is also quite typical for our collective failure (1). He highlights that SARS-CoV2 may turn out only as dangerous as the flu and makes the now classic mistake of suggesting containment measures after community spread when it would be already too late: "Community spread in the United States could require a shift from containment to mitigation strategies such as social distancing in order to reduce transmission. Such strategies could include isolating ill persons (including voluntary isolation at home), school closures, and telecommuting where possible"

Of course by the end of February it seems likely the US had many thousands of undiagnosed infections already.

Ok, enough about Fauci's failures. Let me highlight four themes that clouded my judgement. This may be a good exercise, because I think these may also have mislead other people.


1/ As of February and March, nCoV R0 and fatality rates were similar or below the levels seen with the related SARS and MERS viruses, which compared with SARS-CoV2 were rather easily contained. If I had to speculate why this was misleading, I would pin it on asymptomatic and/or presymptomatic transmission. Cutting down R0 is harder for a disease with a sizable fraction of "asymptomatic" spread.

2/ This BMJ article published on the 19th of February, suggesting that mortality outside of Hubei was 0.05% and there were no "sustained chains of transmission ... outside Asia." Of course we know now that an incubation time of ~6 days, a protracted disease course that can last a month and reporting lags all reduce initial mortality (CFR) estimates and case numbers. We've seen this development play out in many countries that started with a low CFR of <1%, like South Korea, and end up with a CFR of >2% at the levels of China. (Of course regarding estimates of true fatality rates, like IFR, we have to consider different biases, especially, the large number of asymptomatic cases that push the mortality rate down.)

3/ My utilitarian views are consistent with the idea of trade-offs*. In a resource constrained world you cannot save every life. Period. However, the tricky part is smart resource allocation and there can be reasonable disagreement over the best strategy. Given my thinking, this Vox hit piece arguing against travel bans was appealing to my biases. I am not saying they don't have a point. In many cases, for a relatively harmless disease (e.g. regular influenza) or for the prevention of rare events (e.g. a terrorist attacks), we can assume that airport screenings are a waste of money, but perhaps not so much for SARS-CoV2. While, yes, airport screening, quarantine and travel bans are a "leaky" intervention, given the large number of asymptomatic and pre-symptomatic cases and already established chains of transmission, nonetheless they do have a role to play. Especially because we don't have many other effective measures.

*for example, I find the different pseudo-herd immunity approaches proposed by politicians in Sweden and the UK worth discussing and not per se beyond the pale. However, ultimately they have to be reject based on precautionary principles and promising evidence from countries that slowed or eliminated endemic transmission without such large blood sacrifices and the death of thousands.

4/ Another issue is political, as we did not realize that China had become a "real country". Many people thought: this is just happening in Hubei, in a poor place far away. Getting back to the Nature news article with the R0 estimates, actually highlights that point. While still flawed, China's response to SARS-CoV2 was rapid and mature. We should have realized, if China cannot contain it easily, no one can, because they use the same technology as we do. Hubris, if you will, brought down the West. Hopefully we will learn our lesson.

Current status in Asia
Regarding predictions, if you can say I made any, this one is holding up rather well. In my last blog post not only did I lament the somewhat dishonest reporting on masks seen in the West but also our overall incompetence in handling the crisis. Has the situation changed? Not much. In some ways I am pleasantly surprised with the European success coming out of the famous lockdown initiated mid March. This is another thing I did not predict, since I expected a surge of cases in May. On the other hand, I did expect continued Asian superiority and this is still the case. Let us review old and new data.

In a regional break down Asia has among the fewest COVID cases. However, this is not a very informative way of looking at the issue. I am hard pressed to put countries like India, Bangladesh or Cambodia in the same group as Korea and Japan and then compare them with Europe or North America. These countries are simply too different. While I applaud the good numbers coming out of Africa and so far from India, a lot of this is due to 1) under reporting, 2) extremely young population demography, 3) climate 4) lack of travel and thus few seeding events by Chinese travelers and later on international arrivals. Data from South America supports this thesis. They looked like "winners" when in fact the outbreak was just delayed.

That is why in my last post I defined an elite group of Asian countries and regions consisting of Taiwan, Singapore, Hongkong, Japan, Korea and China. They have a couple of things in common that matter. They are all located in East and South East Asia (so we'll call that group the S/SEA or SEA cluster). They are the wealthiest in that region. These places were hit earlier than the West and have very high population density putting them at an inherent disadvantage compared to the West. Culturally, they all have a strong Chinese influence* and a shared Confucian/Buddhist heritage; with somewhat lower levels of individualism as compared to the West; which is to say that no matter how we slice it, these countries are more similar to each other than they are to the West, allowing us to define a rather tidy "cluster". Some of these characteristics may be a bit speculative, but they will do for an initial analysis.

*through sometimes not very peaceful exchanges between each other and a history of mutual wars and war crimes

Poverty was one of the biases I had in mind when I thought about fair data selection. Given that, my biggest regret about this choice of countries is the omission of Malaysia. At a 2017 GDP of 11500$ they are technically richer than China (at around 7500$) and probably capable of screening for COVID and reporting reliable data. For further analyses I will thus consider the initial SEAsian cluster with Malaysia.

The superiority of the Asian elite cluster became obvious early on, around March/April, because these countries were hit earlier than the world and yet showed better numbers than Western countries that were pre-peak (when the data would bias in their favor). The key prediction was that the difference between the elite cluster and the West would continue to grow for a while, but could eventually shrink if the world adopts methods from the elite cluster. One corollary is that the GDP and economic data would favor Asia further accelerating America's demise as the go-to superpower.


Figure: Cases and Deaths in Asia vs Europe a comparison of early April and late May data. The newest addition to the S/SEA elite cluster Malaysia is shown in red. We can see that while Singapore has failed to contain the spread of the disease, they continue to show superior mortality data in line with other Asian countries. Overall we can see that, as predicted, the situation in Europe deteriorated over time but not so much in Asia.

When we look at preliminary Q1 economic data we see how this has played out. For example, South Korea shows a 1.4% contraction even though they had an ongoing crisis for most of Q1. Contrast this with Germany at 2.2% and Italy at 4.8% who only started to deal with COVID in late March. The difference is astounding. Only China, the country earliest hit and with the tightest lockdown, foreshadows the economic toll we have yet to face in Europe with a contraction of about 10%. However, we can expect China's Q2 to be considerably better than Europe's offsetting the initial disadvantage.

What really is the recipe for success?
Perhaps it is easier to say what it is not. It is not abundant testing, although, it may be smart testing (test early, test & trace). Whether you look at the Our World in Data graphs or the (lower quality) worldometer dataset, we see that Europe has a testing advantage over Asia. Skeptics now might argue that the West has more cases than Asia due to this statistical anomaly, i.e. because they detect more cases. Without getting into details, this seems unlikely because Asia's lead in COVID mortality is even more crushing. Asia's low test positivity rate is also consistent with the conclusion that their case numbers really are low. More importantly, this goes a bit against the "test, test, test" mantra. Although, it is very possible that high rates of testing in Europe are good, necessary and able to partially compensate for some hard-to-quantify weakness of our system, by itself this is not enough. To rephrase, it seems to me that testing could compensate for inferior contact tracing and hygienic prevention measures in Europe (speculation).

Figure: Testing rates in Asia are low. Image generated at ourworldindata.org. I only added Germany and Italy as an example so the graph doesn't get too crowded.

It is not lockdowns and mobility reductions. If we look at a point estimate from Google we can see that if anything Asia managed the crisis with less restrictions than the West (-33% vs -42%; Retail index). Of course S. Korea here is the poster child but even when we look at Japan who now has the tightest "lockdown" and state of emergency since the crisis began, their mobility reduction is not even close to the 60 to 80% decreases we have seen in New Zealand or Italy. To be fair, and without getting into details, there are several issues with the google data that preclude us from making strong statements (cross-country validity, badly chosen baseline, point estimates don't reflect the average). Doing a rigorous analysis of this is on my data to do list.


Figure. This figure shows the severity of lockdown and distancing measures as judged from google mobility data on the 27th of May (higher values represent a more severe restriction) and number of tests performed (per 1000 citizens). Unfortunately, we have no data from mainland China regarding number of tests performed and severity of distancing. However, we know that the lockdown was eased in China long ago, so just for fun I have imputed an estimated value for mainland China (in green) taking the mobility index of the special administrative region Hong Kong and adding a safety margin of 2 (in favour of higher mobility restrictions in the mainland and thus in favour of the West). Malaysia is again in red.

One theory might be that Asia has an inherent advantage that we just don't understand. Maybe regional genetics, climate or something else. However, this seems rather unlikely as Singapore and Hubei demonstrate that an explosion of cases is possible in Asia. Singapore is also an interesting case that deserves a closer look. Originally hailed as a success, cases eventually exploded in Singapore. This was against my predictions and I expected to see a sharp and brief peak like in South Korea. The opposite happened and cases are still growing in Singapore. Although, interestingly there are almost no deaths reported. The data is so good that I have a hard time believing it until it is validated through excess mortality (like we started doing for Japan, see below)! Taking the mortality data at face value Singapore remains one of the elite countries. It seems they effectively protected the elderly and the virus spreads among young (migrant worker?) communities.

The peculiar case of Japan
Not long ago I compared Japan's COVID response with a bumblebee and I still like this metaphor.
Commentators, understandably puzzled, write that "Japan appears to be doing everything wrong" and yet is surprisingly successful even though "it has world’s fastest-aging society" and a climate friendly to the virus (the foreign policy article is quite good, read it).

That is to say, many people don't know how Japan can stay afloat, but on deeper inspection Japan is doing well and there are good reasons for it. I've already talked a bit about Japan in my last blog post. Apart from early recommendations for (modest) social distancing, school closures, limits on large gatherings, Japan has also employed thorough contact tracing and their numbers were basically indistinguishable from Singapore (at first glance anyway); Singapore, that was hailed as one of the best countries for (analog) contact tracing! Nevertheless, there is still a lot I don't understand about Japan's apparent success. I recently watched a video discussion, contrasting the German and Japanese approach, and I can't say it clarified much. One interesting take home point: the speaker Norio Okada thinks that Japan was not as well prepared as S. Korea and both speakers agreed that their countries ignored the worst case scenarios during preparations and simulations. I concur.

Many articles highlight just how little we know.
Theories for the country’s relatively low mortality rate run the gamut from cultural attributes — widespread mask wearing, a practice of regular hand washing, a near absence of physical greetings like hugs and handshakes — to just plain luck.
A combination of many other factors, including government measures and changes in behavior among a public that feels strong pressure to follow the rules, could also be at work.

Unfortunately, I have yet to see an article which considers that Japan forms part of an Asian cluster of elite responders and answers might be found in the similarities between these countries. Let's start with one controversy.

Regarding isolation at homes vs at facilities, there rages a heated debate on twitter and elsewhere  (see for example this article by Vox) which seems nonsensical to me. Successful countries have employed both approaches. Both are good as long as they are strict and people comply. The issue of infection to household members can be addressed in efficient ways, so that strict at home quarantine would be virtually indistinguishable from out of home quarantine, except much cheaper and more pleasant for all mild cases (and there are quite a lot of mild cases, e.g. Oyungerel et al. 2020). In countries with sustained endemic transmission out of home quarantine seems like an insane economic drag. However, effective quarantine at home may require the use of app based surveillance which is sadly unacceptable in Europe.

At home quarantine is also much better suited for Europe, in some ways, than for Asia since single person households account for 40% of all households in Germany (1 in 5 people live in this way) and perhaps half that in China. Crowded housing and communal spaces also might be less common here.

Obviously household members can be protected by removal from the household, daily testing or en bloc quarantine. I don't see any magic here, except it is again important to increase compliance. As stated here, Germany uses a moderately strict at home quarantine in many cases. They don't even advise that other household members have to move out and I don't doubt that this contributes to Germany's inferior performance. Nonetheless, among the incompetent Western European countries, Germany still fares the best.

While it may be technically true that Japan sometimes uses out of home isolation(**), this is meaningless because they don't test enough people and just advise sick people to stay at home unless they are exceedingly sick. Korea also quarantines some patients at home. In contrast, it is true that China, Hong Kong, Singapore and AFAIK Malaysia quarantine cases out of home. The case of Taiwan (*) is a bit confusing, but I think I misread their policy and they also do out of home quarantine.

Japanese mortality data
Some critics raised the point that Japan is lying about their numbers, because at the beginning of the epidemic they still hoped to host the Olympic games this year. Surely, the testing data is consistent with this notion. Japan probably under-reports cases. However, we can say with some confidence that there was no excess mortality in March and no obvious cover up like in Russia. April and May data will be more interesting since most official deaths occurred in these months.


Figure. Excess mortality in Japan, comparison with figures from past years. Data for 2020 and 2019 is considered preliminary and will undergo revision. From what I've seen this is not a cause for concern because the revision doesn't change the numbers much and we can see that the number of deaths is comparable between March 2019 and 2020. Data acquisition: MI.


References
Fauci, Anthony S., H. Clifford Lane, and Robert R. Redfield. "Covid-19—navigating the uncharted." (2020): 1268-1269.

Vetter, Pauline, Isabella Eckerle, and Laurent Kaiser. "Covid-19: a puzzle with many missing pieces." (2020).

Byambasuren, Oyungerel, et al. "Estimating the extent of true asymptomatic COVID-19 and its potential for community transmission: systematic review and meta-analysis." Available at SSRN 3586675 (2020).

Notes
The assertion that Japanese greeting culture is one recipe for success during a pandemic, is not implausible. See below article for bacterial transfer, but no reason this wouldn't translate to viral droplets and aerosols that end up deposited on the hands:

Miela S and Whitworth DE. The fist bump: A more hygienic alternative to the handshake. American Journal of Infection Control, Volume 42, Issue 8 (August 2014)
"Nearly twice as many bacteria were transferred during a handshake ..compared with a high five, whereas the fist bump consistently gave the lowest transmission (Fig 1A)"

More on Japan's excess mortality
When preparing the data I was not aware that others were also looking at excess mortality except for the Economist dataset. So unfortunate, I wish I could have been the first to report the news when I had it around end of May. Oh well, never mind, when you write slowly you risk getting scooped. Here are the articles:

Published June 11, Tokyo shows considerable excess mortality in April: around 1000

Published May 25, Tokyo excess mortality through Jan to April:

Published April 14: no excess deaths in Tokyo yet

No increase in Japan wide deaths in the first quarter, MAY 28, 2020:

Comments regarding out of home quarantine
*e.g. The Taiwan situation is very confusing since the JAMA article about Taiwan's response does not mention out of home quarantine at all and basically ignores what happens to confirmed cases. It only mentions at home quarantine for travelers, even those showing symptoms. Like a million times. From this I concluded that Taiwan must be using at home quarantine for confirmed cases but this seems to be incorrect. Another source also says that "individuals who tested positive for the virus were quarantined either at home or at hotels" so there is a bit confusions regarding that policy.
This article by Taiwanese researchers, government policy I could find online and statements of friends, however, suggest that it is mostly out of home quarantine.

**although, there are nation wide regulations in Japan it is not that clear cut: "A national law on infectious diseases mandated that anyone who tested positive, even those who were asymptomatic, had to be placed in one of the country’s few isolation wards, creating a strong disincentive for doctors to test patients with milder symptoms"
My friend checked the official guidelines and they leave it up to local authorities how to handle cases, suggesting treatment can happen at home or out of home facilities. I am confused about the different reading and implication, so that's worth following up.

Malaysia: https://mpaeds.my/wp-content/uploads/2020/01/Guidelines-2019-Novel-Coronavirus-2019-nCoV-Management-In-Malaysia-No.-22020.pdf

Kommentare