Mridul K. Thomas
  • Home
  • Research
  • Publications
    • Errors
  • Data & Code
  • CV
  • Contact
  • Covid
ecology - phytoplankton - functional traits

09.07.2020

2/1/2021

0 Comments

 
Coronavirus news #7. As always, please point out my errors. I started making notes a couple of weeks ago before getting side-tracked. In the meantime, we've had 2 million new reported cases - about 20% of the total.

1) About 4000-5000 people are reported to die everyday because of this virus, or about 3% of all deaths. Globally, that has been stable for nearly 2 months. The true number is almost certainly higher, but it's unlikely to be by a whole lot (high uncertainty in India and some other places, though). Please remember that the people who don't die don't all recover completely. Many, many more are being struck with long-term damage to their health. That includes a substantial number that don't have any symptoms and may not even realise it now; many asymptomatic people show signs of lung damage in scans. So though it's normal to have one's standards for safety slip after all this time and with fewer immediate consequences in our daily lives, please be safe. Wear a mask.

2) Though the US mostly leads the news, Brazil and possibly Mexico are likely in worse shape, with India bidding to take this miserable crown. South Africa also seems fairly likely to join that group. Nobody really knows what's happening in Russia, because there's clear signs that some of the data being released are junk. Unlike the other countries, the US has actually been declining in deaths per day since the beginning of May. Brazil, Mexico, South Africa and India are still increasing, and the first two have substantially more daily deaths than the US. There has been a worrying drop-off in reporting (even by scientists) about the state of the virus in the developing world, with Brazil being a partial exception.


3) The US has seen explosive growth in cases in the last three weeks but deaths per day have actually declined since early May. There's multiple reasons for this. A greater proportion of new cases are in younger people (~10-100X less likely to die), we have developed better methods to manage severe illnesses, and perhaps most importantly, we are catching cases earlier. The last point implies that the lag between increases in cases and increases in deaths should be increasing. It used to be about 2 weeks; it may be 3 or 4 now. If this hypothesis is true, we should see increases in deaths across much of the US in another week or so. We can be fairly confident that the growth in cases is not simply because of more testing, because many of the outbreaks are also seeing an increase in the percentage of positive tests - a sign that an increasing number of cases is being missed.


4) Europe shouldn't feel too thrilled. After getting things under control, they've messed up the reopening and experienced about a month of increasing daily cases. Switzerland has finally mandated mask usage in public transportation. I think that ought to be the minimum policy change that we should aim for. There seems to be reasonable evidence that mandatory mask usage and closing some activities that involve crowds or closed spaces (e.g. bars) would be sufficient to kill the pandemic. Mask usage alone does a lot of the heavy lifting. Even tighter restrictions would do it faster but might be overkill if the situation is not yet out of control.


5) Antibodies are the focus of 2 important types of testing: (i) testing to figure out if someone has had the disease and will therefore be immune, and (ii) testing to figure out what proportion of the population has been infected by the disease ('seroprevalence studies'). This is how we get exposed/infected numbers like 5% of the Spanish population (it's similar in the US). But we now know that antibodies actually decay below our ability to detect them relatively fast, roughly on the scale of months. This does not mean that these people have lost immunity; they almost certainly retain the ability to produce more antibodies in the future if re-infected by the virus. Also, some indeterminate (but probably low) number of people apparently tend to not show much of an antibody response. However, most (if not all) people develop another type of immunity that is mediated by specialised T-cells. These provide long-term immunity and are generally ignored because they are much harder to test for. They are not quite as effective as antibodies but are likely important, though we don't know much about their role in COVID immunity yet. Bottom line: (a) we are still fairly confident that people develop immunity after infection, (b) the antibodies we test for decrease over time but we retain immunity, (c) this antibody decrease & T-cell response adds an unknown but probably small bias to our estimates of how much of the population has been exposed.

6) There appears to be some pre-existing immunity in people who were infected by SARS-1 many years ago and by an unknown coronavirus. There are 7 documented coronaviruses circulating in human populations, but there is likely an unknown one circulating in South-East Asia that has protected populations there through this 'cross-immunity'.

7) Most evidence points towards us being very far from any sort of herd immunity. If the governments of the US, India & Brazil give up on containing this, I would say we are heading towards millions of deaths with high probability (given their failure so far, we are headed there already with moderate probability). To speculate even further, tens of millions is not implausible in this scenario.


8) Last month, I wrote that perhaps 80% of new cases are caused by 10% of infected people (superspreaders). This has important implications for how to do contact tracing. The standard method is to find everyone an infected person was in contact with and notify/quarantine them. But since they were almost certainly infected by a superspreader, an improvement on this is to identify the superspreader who passed it to them and then trace everyone they were in contact with. This 'backward contact tracing' can be thought of as a second level that makes the whole procedure much more effective at stopping an outbreak. That said, contact tracing isn't feasible when you have a massive ongoing outbreak.


9) There is now plausible but inconclusive evidence that the virus has adapted during its spread. A mutation that increases the spread in limited lab experiments has also come to dominate in new cases around the world in the last 3-4 months. But it could have spread more just by chance, by being in the first cases that spread to new places. However, the mutation does not appear to do anything to the virus' lethality, so this evolution has not made things any better for us. And it does not change anything about how we should prepare for or treat it.

10) Despite all the press about treatments, no specific medication we have so far is close to being a cure. Hydroxychloroquine does not do much and has wasted time and money. Remdesivir and Dexamethasone have relatively weak effects or limited use-cases; they will save lives but are not cures. When used most effectively, dexamethosone appears to save about 25% of lives lost. This is progress, but we need a lot better. Both also have severe, dangerous side-effects that can kill. We are on the way to having monoclonal antibody treatments, probably within a few months. This would be a big improvement but is probably going to be expensive. The US appears to be doing its best to corner the market on these right now, though that investment should stimulate more development.

11) Children appear to be about 1/3 to 1/2 as likely to get infected. And they are much less likely to fall seriously ill or die once infected; they have ~100X lower probability of dying if infected, relative to people over 80. This has obvious implications for reopening schools and colleges. I think doing so is probably feasible if coupled with distancing, mandatory masking, and mass testing on a regular basis to catch outbreaks early. But the details matter a lot.


12) There's now even more evidence that surfaces are not that important for spreading the virus.


-----------------

In response to a question about whether we were headed in the right direction at this time:

Absolutely in the wrong direction in the US right now.

We still don't have enough tests or protective equipment, mask usage is not high enough, people in outbreak areas appear to still be frequenting indoor restaurants (less than normal, but still), new cases are going up rapidly, and the federal government & the CDC continue to be disastrous (more on that soon).

The main good news is that deaths per day are much lower than May, but as I said above, I expect that to increase soon. There's a phenomenal amount of scientific talent, of course, but it's not much good if the people in charge are incompetent, malicious or craven (or some combination of the above).


Other that that, the prospect of monoclonal antibodies, some advances on treatment, and some progress on the vaccine front are all encouraging. But none of them is close to deployment.
0 Comments

02.06.2020

2/1/2021

0 Comments

 
Coronavirus news #6. As always, please point out my errors. Not an expert in this specific domain, but I have relevant training and am trying to keep up to date. I encourage those of you with relevant expertise to also spread your understanding as best you can.

1) It's happened without much fanfare, but it seems as though contact with surfaces is no longer thought to be a very important method of spreading the virus. In March, there was a lot of advocacy around using gloves, washing hands, & avoiding face touching. As more evidence has poured in, the cautions and analyses have shifted towards focussing strongly on inhalation of viruses. I have yet to see clear research on the relative risks (please tell me if you know!), but I think this is a fair summary of epidemiologists' present beliefs. I'm not saying that viruses do not spread through contact with surfaces, so continue to take reasonable precautions. But I offer this to allay the fears of friends who go to incredible lengths to sanitise everything they buy or that enters their house. Letting new acquisitions sit in the sun for a week is overkill.

2) Brazil and India have exponentially increasing case numbers despite lockdowns - in India's case, one of the most brutal in the world. The outbreaks are concentrated in some regions, so it is possible that they will get contained. But given exhaustion with the cruelly mismanaged lockdowns (at least in India), I'm doubtful they will be. And they have
20% of the world's population between them.


3) As a consequence, the situation right now is far worse than I think most people realise. Global daily deaths are slowly decreasing (currently ~3000 per day), driven by improvements in the situation in Europe and the US, which were hit worst. The news is moving on to the economy, tourism, protests, and so on as local situations seem to stabilise, and people get bored or desperate. But this apparent progress can be quickly reversed by just one or two new outbreaks that can happen anywhere, and these seed new outbreaks in turn. We do not have a single pandemic, but a patchwork. Regions/countries with outbreaks can be isolated, but the success rate needed to prevent them seeding new outbreaks is very high. Most countries have not displayed the competence needed. We are stuck in this situation till we have a vaccine.

4) As India's & Brazil's numbers rise, global daily deaths will start to increase again. We are now looking at possibly millions to even tens of millions of deaths. This probably seems unlikely to you and I hope it's totally wrong. But in March, the possibility of tens of thousands of deaths was seen by many as unwarranted pessimism, and hundreds of thousands as wild fantasy. And yet here we are. We ought to learn something from this about how unreliable gut impressions about worst-case scenarios and tail risks are. And perhaps eventually turn that insight towards another area where scientists have spent decades venting their spleens about tail risks: climate change.

5) A better argument for why millions of deaths is plausible: a few weeks ago, Spain and France did the biggest & best random testing programs run so far, helping to fill a massive gap in basic data. These found that the infection fatality rate is about 1%, just as epidemiologists have been saying since at least March. Remember that infection fatality rate is the % of people who ever got infected that die, which is the number we are most interested in. Most reports have been about the case fatality rate, which leaves out all the people who got infected but did not know about or report it; this varies a lot based on medical system and reporting practices. Also, about 5% of those countries had been infected (i.e. had antibodies) at the time. If the virus were to continue to spread, you would not get 20 times as many deaths, but you would get very large numbers that are in line with those that basic epidemiological models predicted months ago. Apply that logic and the basic numbers to other parts of the world and you gets millions to tens of millions.

6) A particularly worrisome pattern: there are many more young people infected and hospitalised in India and some other countries than was expected. There's multiple reasons why this may be, such as exposure patterns and data biases. There are more scary possibilities but right now no real evidence for them. But even so, outbreaks among young people are a massive threat to countries with much younger populations, where the hope was that the vulnerable elderly could be isolated while the young slowly return to work.

7) Since I'm thoroughly sick of the endless debate around Sweden and its tactics, allow me to point out stories from other countries that are not discussed much. Take Mongolia and Vietnam, which have had zero deaths between them with a combined population of 100 million. Zero. And they share a border with China! Senegal, though further away, has had about 40 deaths in a population of 31 million. Ghana: about 35 deaths among 16 million people. Ethiopia: 12 deaths out of 115 million. All these countries are relatively poor and lack the capacity of European and American healthcare systems, in terms of manpower, training, and technological sophistication. What they share is that they took action in January, while more developed countries did nothing. I'll add that they are also younger, but I think that is a less important factor. Ghana and Ethiopia are noteworthy, because I understand they implemented better contact tracing than most of Europe or the US. Ghana also tested large numbers of people by using pooled testing, an old method I discussed in the very first of these posts in March. Germany and India have also been using the same strategy, which should have been employed months ago. I was personally terrified about outbreaks in developing countries, and want to highlight that many of them have outperformed the wealthiest countries to an impressive degree, one that ought to shock us. They may yet suffer badly, but they have done well so far.

8) Reopening in many countries is politically inevitable now and justified in many cases. Did the lockdowns work? It's a question that's going to be contested very strongly because of the political and economic implications. But please keep in mind that this is not exactly a yes/no question. There was not a single type of 'lockdown'. Instead, there was a wide range of activity restrictions, ranging from stay-at-home with almost no exceptions (China, India, Italy, France) to limitations on gathering places such as restaurants & movie theatres but not much else (Switzerland) to essentially no legal limitations, just recommendations (Sweden, Ethiopia). There were successes in countries in all these categories. So a case could be made that the strictest stay-at-home orders were not needed, at least in areas without high population densities. But let's put that in context. Strict restrictions were entirely justifiable at the time given the limited information we had and the very real threat of having healthcare systems overwhelmed (as happened in northern Italy). Shutdowns were protection against very real tail risks. Additionally, even the places with no limitations did have big reductions in gatherings and foot traffic because most people are not idiots and generally took more care than usual (more on this in the next point). So the absence of stay-at-home orders did not mean that economies did well; businesses lost a lot of customers because people chose to stay away from each other.

9) It's increasingly clear that disease spread is largely driven by superspreaders. The numbers are far from precise, but something like 10% of infected people cause 80% of new infections, while perhaps 70% do not infect anybody. This is a more extreme 'clustering' than many other diseases. We don't know if there's anything particularly different about that 10% that makes this happen, in terms of physiology or immunology. But this represents an important opportunity: if those superspreading events can be contained, the epidemic may die out by itself. [The following is my speculation] I suspect this is why even weak reductions in population activity were extremely effective in suppressing the disease. And why many places have not seen surges in cases after reopening. When people took the logical step of avoiding the most crowded areas (stadiums, public transportation, bars, restaurants, etc), they effectively prevented a lot of potential superspreading events. I'd like to stress this point: it shows that even weak compliance with distancing recommendations/rules can be immensely beneficial for the whole population. This is worth keeping in mind as the lockdowns are lifted, new outbreaks happen, and plenty of people can't be arsed to obey the resulting lockdown reimposition.

10) As countries reopen, there is going to be intense debate and lobbying about what to allow. Given what we know, crowded places/events with a lot of people expelling a lot of air are probably the riskiest. An incomplete list would be: gyms, sports stadiums, concerts, university dorms & big classes, some places of worship (there seems to be a weirdly high number of reported outbreaks among choirs), nightclubs, some kinds of restaurants & bars. There's some evidence that younger kids don't get infected much and so reopening schools may be less of a concern (only moderate confidence in this claim right now). I expect places of worship will be opened sooner than ideal, so we ought to be talking to religious leaders about how best to manage the consequences of the outbreaks that will ensue.

11) As we reopen, there are going to be new outbreaks. In an even mildly competent world, we'd have built up the capacity to do lots of testing, tracing, and isolation of newly infected cases. I don't think we have done this to the degree needed in a lot of countries. Since we have not done this, a second, poorer option is cycles of shutdowns and reopenings as outbreaks are detected and controlled. Except plenty of societies are barely tolerating the first shutdown, so subsequent shutdowns seem almost guaranteed to be ignored. So what happens instead? I expect we'll see plenty of businesses change how they function for the near future till we have this under control. Some businesses may also start offering exceptions to these restrictions if customers can provide proof of immunity from antibody tests. This will essentially provide extra services to people who have immunity, creating a market for both fake test results and for people to voluntarily get infected. A while ago I mentioned that governments manage this by providing people with official 'immunity passports'. This is understandably opposed by civil liberties advocates, but we might end up with a messy private version of this instead.

12) I mentioned universities earlier and want to expand on this. A combination of factors are going to send a lot of universities - especially in the US - into financial freefall. The risk in crowding lots of people together into classes & dorms, the poorer educational experience of video conferencing, high tuition fees, immigration restrictions cutting off high-paying foreign students. and the inevitable recession-driven slashing of government funding (in the US) are all going to hit the bottom line. Also, universities have apparently been mismanaged to the extent that many do not have the financial reserves to weather these storms (if they can't even manage money, why on earth did we let the corporate world start running them?). The famously large endowments are not (according to university spokespeople) funds that can be used to manage the crisis, because money willed to the university comes with specific conditions attached. Using them to tide over general budget shortfalls apparently violates a bunch of legal agreements, ones which universities are less happy to break than say, union-negotiated job contracts and pension liabilities. The biggest universities will probably manage somehow, in part by taking in students who might otherwise have gone to lower-rung universities. At least some of those lower-rung universities may have to have to radically change or close down. Hopefully some good will come from the necessary experimentation. Teaching in the lockdown has kicked many of us out of our rut and may help improve what we do. With more support and practice, we could perhaps be teaching more students, better. But it seems likely to be a bleak few years to be an academic, especially an early-career one.

13) I've argued previously that you should ignore the headline of the day, because most science papers offer provisional results, the media communicates science badly, and you will end up misinformed. I stand by that, but feel obligated to talk about one study in the Lancet that claimed that hydroxychloroquine does not help with COVID and actually kills people instead. For obvious political reasons, this story spread like wildfire. Well, there are massive problems with that study, and not of the usual scientists-disagree-about-how-to-interpret-everything kind. There's an extremely good chance that it is a hoax and the data completely made up. Here's one valuable critique you can read: http://freerangestats.info/.../implausible-health-data-firm . Note that this does not mean that hydroxychloroquine is actually good.

14) We're likely to see more protests of all kinds soon, and not just in the US. These protests are extremely likely to spread the virus. If you argue that the goals of a specific protest are more important than the possibility of spreading the virus, be prepared to listen when a similar argument is made by those protesting for goals you dislike. Dismissing their arguments out of hand will lead to charges of hypocrisy, and people who disagree with you politically will have less reason to listen to your advocacy on any topic. In any case, we should be preparing for outbreaks in all these cities where protests happen. The sad fact, though, is that contact tracing is going to be a much harder sell when protestors view it as a way for authorities to track down and suppress political dissent. Contact tracing that maintains privacy may be vital here.

15) Slow vaccine progress continues to be made on a variety of fronts. Two of the best initiatives on that front are those by the Gates Foundation and Operation Warp Speed by the US government, both of which are spending billions to build capacity for vaccines that we do not have yet. Most of that money will end up being spent on vaccines that do not work. But if one or two work, it could save millions of lives and trillions of dollars. This is the kind of targeted, wise initiative that we need more of, and not just to deal with this virus.

------------------

In response to 2 questions about my speculation in point #9 that weak measures may be sufficient to stop superspreading:

(i) about modelling rare events

A toy model should be relatively simple to implement. If you want to get fancier, here's some resources you might find interesting/useful:
https://www.nature.com/articles/nature04153 (old & general, and already supports the basic intuition)
https://covid.idmod.org/.../Stochasticity_heterogeneity... (very new, specific to COVID)
https://epidemicsonnetworks.readthedocs.io/.../index.html (Python module for modelling epidemics on networks)

(ii) explaining my intuition behind why weak measures can stop superspreading:

I'm going to invoke some realistic numbers for this based on what we now know about COVID:

Imagine 10 people are infected. 7 of them do not pass it on; actually, let's say they *cannot* pass it on, for argument's sake. Of the remaining 3 people, 2 pass the infection on to 1 other person each. Let's say this is to a family member or someone they are closely associated with. The disease spread by these 9 people may be unaffected by distancing & lockdowns, but it only results in 2 infections.

For the disease to continue to spread, the remaining 1 person has to infect more than 8 people, since the initial 10 has to infect more than 10 or it dies out. In ordinary times, this may happen when they encounter crowds at the office, or church, or a restaurant, or on the bus, or a concert, or so on.

When people are distancing and the most crowded areas are shut down or avoided, this one infective person encounters fewer people and at lower densities. This makes them much less likely to pass it on to more than 8 people. Even if people are not strictly following the rules or the lockdown is quite mild, they may infect only 4 or 5 more. The newly infected cohort is 7 people (2 + 5), which is lower than the previous cohort. Since the same logic applies to them, over time the number of infected dies down and goes to zero.

Basically, preventing a few people from each infecting 10 people is easier than preventing lots of people from each infecting 2 people.

------------------

In response to a question about a Medium article about Covid being a 'blood-vessel disease': 


It does seem to be a plausible mechanism, though the Medium title is hyperbole. There's been a couple of months of reporting on this, and the story does seem to be gaining in prominence (see this from yesterday, for example: www.sciencemag.org/news/2020/06/blood-vessel-attack-could-trigger-coronavirus-fatal-second-phase )

There's an odd angle to this story that I spent some time on. The Medium article cites only one scientific paper, so I took at look at it. Bizarrely, it shares a couple of authors with the hydroxychloroquine paper that appears to be a hoax (the one I wrote about above in point #13). That made me worry about how reliable this paper was, so I dug into that a bit. As best I can tell, the shared co-authors were not the main players in the (alleged but very likely) hoax. I'd guess they were innocent; their other co-authors on the dodgy paper were from a company that seems to have fabricated the data. It does call these doctors' judgement into question, though.

------------------

In response to a question about the role of population density and infrastructure:

I agree that density & infrastructure are likely to be very important, and probably a major reason why New York was hit so horribly. And yeah, that also implies huge differences between states, which is likely to make co-ordinating policy tougher. Low-density states are generally Republican and already relatively sceptical of scientists. If they are fortunate and escape big outbreaks, they are likely to become even less supportive of public health efforts, including vaccination when that becomes possible.
​
Europe basically suspended Schengen when it closed all borders, and each country chose its own path and felt the consequences of its own decisions. US states will be pursuing different strategies but won't be able to close borders, which makes the outcomes dependent partly on the policies of their neighbours. I expect this will cause a lot of friction.
0 Comments

29.05.2020

2/1/2021

0 Comments

 
In response to a question about the WHO covering for China:

I'm no fan of the Chinese government, and I think the WHO has taken some pretty lousy political decisions in the past few years (it screwed up on Ebola before this). I also agree that the WHO has deferred too much to China on this. But much of this popular story you refer to seems to me to be an effort by governments to shift blame and excuse their own negligence.

China contributes a relatively small portion of its funds. You know who contributes more? The US. And the UK. And Germany. And Japan. And the EU. And the Gates Foundation. Even goddamned Rotary International pays more than China.

Of course, maybe money is not the main factor. After all, the director-general was elected with support from China. So who runs the WHO? It has 194 member states and an executive board of 34 members that is chaired by a leader that rotates geographically. The chair probably doesn't have much power, but FWIW, it was led by Japan till a week ago - hardly a shill for China. Where's the US in all this? Oh, it's not bothered to nominate anyone to its executive board seat, which has been left vacant for years. This is consistent with the US government's choice to ignore or weaken pretty much every international organisation. If the US chooses to ignore these organisations, is it any wonder that other entities increase in influence, if only by virtue of being at the table?

The WHO made statements in January and February that were stupid and wrong, even given only the information available publicly. But you can call it a strategy: they were probably trying to coax China into giving their experts access to Wuhan to study what was going on. To some extent it worked, China did let them in. It was not worth the damage caused by the misinformation. But the WHO has no power to force any government to do anything, so they probably thought mollifying the Chinese leadership was the best option they had. Especially when the only powerful country that could back them in this fight, the US, was in absentia and busy praising the Chinese leadership for their efforts.

Most importantly, however, plenty of countries did get the obvious message back in January and February and took action. Vietnam has had 0 deaths and 300 cases. Mongolia has had 0 deaths and 200 cases. These are relatively weak, poor countries that share a border with China. They simply jumped into action immediately, based largely on information that you could read in any international newspaper at the time. Even Kerala had a plan in place in January.

So whatever influence China may have at the WHO, most of the lives lost have been because of the incompetence of authorities in the US and Europe. This was not inevitable. There was more than enough information available in January and February to take action, and for reasons that I do not understand, most governments and health authorities twiddled their thumbs.

The WHO is a convenient scapegoat for their failures to save the lives of hundreds of thousands (likely over a million before this is over) and the livelihoods of hundreds of millions.
0 Comments

05.05.2020

2/1/2021

0 Comments

 
Coronavirus news #5. Developments have been slow, so this sort of aggregates the last ~3 weeks. I don't think any of it is out of date, but it's possible. Please point out my errors. And let me know if there's a question you want investigated. Several weeks without human contact and I now have no idea what is common knowledge and what isn't.

1) After the shock of exponential growth, we seem to have been in a bit of a holding pattern for the past few weeks. Reported deaths per day have even started to decrease in most places. But the decrease is slow, and we're still losing 5000-6000 people per day, not to mention identifying ~70,000 new cases. That's about 4% of the number of people who normally die everyday worldwide. This is not as bad a catastrophe as we feared it might develop into, but it is a lot of people. Especially since this is the death rate after locking down and taking a torch to global economies, an emergency measure that is unlikely to be extended much longer.

2) The good news is that several countries have avoided the explosive increases we were seeing in March, and some have actually steadily reduced the total number of cases. Sweden got perhaps too much attention, but it is interesting that the mild action they went with was sufficient to at least stabilise the rate of increase (the active cases increase steadily there). Switzerland has done better, with a month of steadily decreasing total cases. And many developing countries have done much better than avoiding the worst-case scenarios. India slowly increases in cases (which is worrying) but it's far, far better now than scenarios that seemed possible to me; I feared bodies piling in the streets and social breakdown. Other countries such as the US have done considerably worse in containing the problem, despite having many advantages.


3) This brings us to the inevitable question: why? Why did some countries have massive outbreaks and others that seemed equally or even less prepared escape them? A month ago (seems like years) I wrote that there were several theories, ranging from the plausible to the wild: local temperature & humidity, degree of contact between generations, very small numbers of 'superspreaders', history of BCG vaccination, smoking preventing infection, and more. Amazingly, none of them has really accumulated much supporting evidence, and several have been weakened by new data. What we know well is pretty much the same as a month and a half ago: it's bad for old people. And having other health problems (respiratory, heart, obesity) probably matters; it seemed very plausible a month ago and I haven't really seen much more about this that changes my mind one way or the other.


4) One important detail is that a huge number of deaths seem to have happened in old-age/retirement/nursing homes - possibly as much as half the deaths in Europe. These types of homes are rare in developing countries, where populations are younger and the elderly typically live with their families. So the concentration of old people, perhaps along with poor health protocols for their caregivers, is likely to have been a massive contributing factor. Other major outbreaks, especially in the US, seem to have happened in prisons and meatpacking facilities.


5) In many countries, we are still not testing enough, and the testing we are doing is not designed well enough to help us understand the broader picture. So months into this pandemic, we still don't know with much precision how many cases there are out there, or how deadly the virus is. As a result of this poor testing, the stabilisation at 70,000 new cases per day may partly reflect the fact that testing has stagnated.


6) Testing was only one part of the solution that epidemiologists have been banging on about for months. The others are tracing (finding people who came into contact with infected people) and isolation (separating infected and at-risk patients from the rest of the population, sometimes including their families). As I've said, testing is in bad shape. And best I can tell, tracing and isolation are virtually non-existent in much of the developed world. Lockdowns will help temporarily but cannot be sustained indefinitely. I don't understand the lack of tracing and isolation efforts. This is yet another colossal institutional failure.


7) Forget contact tracing apps for now. Most people are aware that the basic versions that many governments want give away far too much privacy and are ripe for abuse. There are solutions to this, and Apple & Google have done really great work in pushing for privacy preservation despite pressure from governments (Europe fought against privacy!). But there are huge problems with even the versions that protect privacy. For one thing, protecting privacy means that the system is extremely vulnerable to attack by bad actors - trolls, rival companies and countries, and so on. It requires an improbably large fraction of the population to use the apps for them to work. And because of the way they are designed, there are simply too many false positives and negatives; people will soon ignore the warnings as a consequence.


8) So what would tracing and isolation look like? Massive amounts of human effort. It can be done! Forget South Korea, it does not even require much sophistication. The first state in India to get hit, Kerala, has had teams of people track down every person to come into contact with an infected individual and isolated them. They had doctors and counsellors help every affected person. They provided food and shelter for people in need during the lockdown, included hundreds of thousands (millions?) of migrant workers. Kerala has a population of 35 million people, comparable to most European countries and bigger than most states in the US. It's much poorer than either. It's had 4 deaths and presently 37 active cases. And it is not unique. A smaller state, Goa, apparently had teams visit every single household to check on them. It's a major tourist destination, and it has zero active cases! The US presently has about 7000 people to do contact tracing, for a population of 330 million. That's probably 2 orders of magnitude lower that what is needed. I don't know the situation in Europe and I would love to read about it if any of you have links you can point me to. But I suspect it's not good. I really don't understand why this is not being done, especially when there are large numbers of young people without jobs who would gladly help in this effort and are much less susceptible to infection themselves.


9) Shutdowns have been met with resistance in several countries. To be fair, some of them seem to have stabilised the situation enough that a lockdown may at present be causing more suffering than not. So we should expect some reopening soon. When this happens, we should expect a second wave in most countries (except for those like the US that are still working on their first). Without the tracing and isolation infrastructure, this might spiral out of control once again, and require subsequent lockdowns. I fail to see how any of this is supposed to save economies, especially since people voluntarily stay home even before lockdowns are declared. Stopping the damn pandemic is really the only path back to normalcy.


10) This pandemic has brought out some of the worst tendencies of the press, so I'd very strongly urge you not to pay attention to the story of the day. With rare exception, journalists don't have the training to understand and criticise science, so we have a lot of glowing reports about the latest drug study (Remdesivir! Hydroxychloroquine!) or estimate of disease incidence. Most of these studies are not good enough to pay attention to. Almost no study you will hear about is done well enough and at a scale that will substantially shift the present consensus. At best they will prove to be the basis for further, better studies. If you're going to try and track this sudden profusion of studies, you will exhaust yourself, give yourself false hope, and likely end up misinformed. The normal progress of science involves most studies having substantial weaknesses, and there is plenty of disagreement and argument around them. You don't hear about it because it doesn't make its way into the press ordinarily. Right now, much of the press is reporting the studies based on the authors' interpretations without offering you the criticism from other specialists. Or when they quote someone, they don't understand the topic well enough to show you who is more believable - the author or the critic. If you want to follow one science writer, read Ed Yong in The Atlantic.


11) Aside from the press, many health agencies and governments have also done a pretty awful job of communication. They have stressed certainties ("Masks don't work!") when evidence was weak or absent. When subsequent evidence has proven them wrong, they have earned mistrust and enabled the conspiracy theorists. If you want good information, go straight to the sources: epidemiologists are laying it all out for anyone to see on twitter.


12) Speaking of uncertainty: the question of whether people can be reinfected by the virus keeps popping up. We don't know for sure but there's good reason to expect immunity. It's worth keeping in mind two things when you see reports that someone has caught the disease a second time: (i) There are a lot of false negative and false positive tests. When someone is thought to have been reinfected, it's almost certainly because the test that declared them recovered was a false negative. (ii) Immunity is not perfect. If your probability of catching the disease goes from say 30% to 0.03% (warning: made-up numbers), I would call it strong immunity but you will still have a lot of people catching the disease a second time.


13) You've seen or are going to see big fights over how many people died because of the virus, because of the implied culpability associated with higher numbers. Already, there's plenty of people who are suspicious that deaths are attributed to COVID without testing the corpse. But why would we waste our limited tests on a dead person? And people die for a combination of reasons and it's frequently impossible to say that the precise reason was previous lung problems or heart disease and not COVID; in reality, the combination of factors led to the death. Also, plenty of people are dying without having been checked, so frequently we have no idea. How do we solve this problem? The most reasonable solution is to compare how many people are dying right now with the 'normal' number of people that die at this time of year. We usually define 'normal' by taking the average of many years because some years are better and some are worse. This is not perfect: there are fewer traffic accidents now, less pollution, less flu transmission because of lockdowns, and other factors. But it's the best simple answer. There are clever ways to adjust for the other factors, but they are complicated and the conspiracy theorists will accuse the scientists doing those adjustments of cooking the books. Just be aware that there's a simple and pretty reliable answer that you can calculate yourself.


14) We're now seeing a profusion of epidemiological models coming out. Keep in mind that for the most part they are not designed to tell you exactly what is going to happen. Instead, they tell you what is likely to happen if societies take different actions, so that we can decide about how best to respond. Any change we make (lockdown, reopening, voluntary distancing) changes the situation and makes the model predictions inaccurate. Criticising modellers for being alarmist when we have taken the actions they recommended to prevent their predictions coming true is both ridiculous and exactly what I said would happen in the first of these posts. Please fight these lousy critiques.


15) I've talked about different types of models and expressed my preference for mechanistic epidemiological models (specifically SEIR models) over phenomenological/statistical models (like the IHME model). I used to think that the basic versions I was taught had unjustifiable assumptions that make them ill-suited to making COVID predictions. But I'm now leaning towards the view that many of the complexities they ignore (variation in population density, number of people contacted, susceptibility, mobility, etc.) are not such a big problem, relatively speaking. The bigger issue is that all the models use parameters that are hard to measure directly, so instead we have to infer them from the very bad data we have on reported cases and deaths. None of the models is going to provide a precise picture of the future because of all these uncertainties; even the very best is likely to be wrong about the time of peak, total number of deaths, and more. But some are still better than others and if you're interested in predictions, I'd tentatively recommend https://covid19-projections.com/, https://github.com/ryansmcgee/seirsplus, https://mr-sir.herokuapp.com/main, https://epiforecasts.io/covid/posts/global/, https://covid19-scenarios.org/ as good resources. Keep in mind that ultimately, they all rely on data for their inputs. To the extent that the data has problems (and it does), this makes the model predictions have problems too.


16) We seem kind of stuck till we get a vaccine. The most informed voices are talking about a 2-year timeline for this; the fastest previous vaccine developed took 5 years. I have no expertise in this but if I had to bet, I think it will be done faster. That's because this situation is unprecedented and there are incredibly strong incentives at play. By dusting off every possible angle of attack and technology, skipping a lot of regulatory and testing hurdles, essentially throwing a million darts at the board, I imagine we will achieve this faster. You should not trust me on this, I cannot really justify this opinion.


17) That said, developing a vaccine is very different from deploying it at the scale of billions of people. So far, much of public health infrastructure has failed quite miserably. Much will depend on whether this shock enables the bureaucratic problems to be fixed in the next few months. But institutional knowledge and expertise is not easy to accumulate rapidly.


18) We don't know yet, but it seems likely that people who have recovered will suffer long-term health consequences. The toll from this pandemic is even higher than the body count.

0 Comments

15.04.2020

2/1/2021

0 Comments

 
Coronavirus news #4. An update to the last post about the IHME model. Bottom line: there's still good reason to be sceptical of its predictions.

1) I had expressed scepticism about a graph on the IHME website. It confused some epidemiologists too, but one of them finally figured out what was going on. It was driven by a combination of the nature of the model (atypical and not ideal, but not wrong), poor visualisation choices combined with a lack of explanation (uncertainty in the 'fitted' part of the graph was hidden without explanation), and bad luck (the date of max uncertainty was not simply the next day; it had remained stable from the previous week). Too long to explain here, but you can read more here if you're interested: https://twitter.com/CT_Bergstrom/status/1249135767613554690. I partially withdraw this particular criticism; it's definitely bad but not as egregious as it seemed. The other points still stand, and unfortunately some more have cropped up:

2) People with actual expertise in this figured out that the model has a strange quirk: it assumes that once the peak is reached, the epidemic will decline at the same speed at which it rose (once social distancing is in place). There is absolutely no reason to expect this to be true, and the patterns in China and a few other countries do not match this; the decline is generally quite a bit slower. This is not important for the model's main purpose (predicting peak hospital usage) but does matter because people & governments are making plans based on these predictions of numbers of deaths. A more thorough explanation of this is here: https://twitter.com/CT_Bergstrom/status/1250304069119275009

​3) A group of statisticians and epidemiologists tested how well the model predicted deaths on the following day and it did badly. The model was 'fitted' with data till March 29 and made predictions for deaths in all US states on March 30. Every prediction had a '95% prediction interval'; these represent a range of possibilities for the number of expected deaths. If the model worked well, 95% of the predictions should fall within those intervals. In reality, less than 30% did. In other words, the model was very overconfident in its expected range of possibilities - and this is just for predictions on the next day. The model has since been updated and this group will test the new predictions soon. Details in the full report here.

----------------

In response to a question about herd immunity:

Herd immunity is great, the question is how we get there. There are two paths: either we vaccinate most people or most people get infected. Most people getting infected likely means that many millions will die, especially if it happens quickly enough for medical facilities to get swamped. A vaccine will apparently take 1-2 years to develop. Lockdowns for that kind of time frame are basically impossible because people will revolt. If it were possible, it would still lead to a lot of deaths (I have no idea how many) through the damage to incomes and economies & the consequent drop in medical care and degradation of supply chains.

So we're buying ourselves time to prepare right now with shorter lockdowns, working to understand the virus and disease better, develop treatments and testing capabilities. We are probably going to reopen before the public health experts and doctors think is ideal because the public will demand it. I expert partial reopenings in many countries in 2-4 weeks. Switzerland and a few countries actually have a declining total number of infections, so a slight reduction of the lockdown is justifiable. How we manage the reopening will be crucial: Have we put in place the infrastructure to do testing and tracing on a large scale (necessary to avoid large loss of life, despite the privacy concerns)? How small do we start? How do we manage the inevitable jockeying between different entities that want their industry/business/hobby to be allowed? What is the sequence of reopening steps to follow? What planning do we put in place to trigger a second lockdown if the inevitable second wave of cases starts to threaten again?

----------------

In response to a question about reinfection:

Reinfection seems to be way down the list of most experts' concerns right now, and with good reason. Partly there's not enough data, though it has been looked at recently on a small scale. The general consensus seems to be that it would be extremely unusual for this virus to NOT lead to at least some post-infection immunity, because that's what happens with most comparable viruses. So it's a reasonable (but definitely important) assumption till we know more. The reports of re-infections have so far been rare and in some cases don't seem very reliable because of issues with testing.

But if we make the pessimistic assumption that immunity declines with time till it is eventually useless, whether herd immunity takes hold will I think depend on the rates of disease spread and the rate at which immunity declines. And it will be temporary anyway, since this scenario assumes everyone will lose their immunity.

Basically, most realistic paths end with us having herd immunity through vaccination.

If you want to read more about immunity to viruses, this seems like a good place to start (kind of technical, though):

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3899649/


----------------

In response to a question about the Swedish strategy and the modelling it was based on: 

This led me down a rabbit hole! First, I discovered that Worldometer, my usual source of data on deaths and cases, shows a bizarre weekly pattern in its data for Sweden, with very few deaths on weekends. And it's not a reporting problem, since there's no spike on Mondays. Digging into it further, the original data source does NOT show this weekly pattern, so I don't know what is going on [EDIT: this was because there were reporting lags and the original source would update its numbers when the data came in. But Worldometer would take only the latest day's number and ignore updates to previous days].

But back to the question: I'm not familiar with the Swedish modelling work because the only link I found explaining it was a while paper in Swedish. But I don't think that anyone is really trying to achieve herd immunity in the way that this is understood. Even the weird initial UK plan was not exactly about trying to achieve herd immunity; it was in some sense a side effect. Like with Sweden now, the plan to not lock down was based on the assumption that people will not tolerate the strict and long lockdown that would be needed to completely stop the disease spread. Or that the economic and consequently human costs of such a long lockdown would be unbearable. So given that a long, total lockdown is not possible, how best do we manage the situation? A short, strict lockdown would suppress the disease for now but lead to the same spike once the lockdown is lifted.


The Swedes argue that allowing a slow spread right now through weak measures is the best option available. Assuming they have no shortage of medical supplies (the best argument for a lockdown is preventing medical facilities from being overwhelmed), their plan may be wrong but it's not totally ridiculous. Sweden has a low density of population, so the risks are probably lower than most countries (Stockholm and perhaps some other cities will be more at risk). Switzerland has enforced a relatively weak lockdown (though still much stronger than Sweden) but had a lot of success containing the spread, so this is one point in favour of weak measures. And Sweden retains the ability to enforce a lockdown when things get worse (which they will) and have the population being willing to go along with it.


In the meantime, Michigan just had masses descend on the state capital to block roads (and hospitals!) to protest the short lockdown they have already experienced. What happens when things get worse and stronger measures are needed? I don't envy politicians and administrators having to take these decisions.


Ultimately, short-term lockdown followed by intense testing and tracing still seems the best strategy. But there's so much unexplained variation between countries that it's hard to claim with any confidence that the Swedish approach will fail.


0 Comments

12.04.2020

2/1/2021

0 Comments

 
Picture
Coronavirus news #3. Just one point today, but it's an important (and loooong) one. Summary at bottom if you want to skip. As usual, I welcome pushback.

COVID-19 models have been in the news lately and there is a brewing fight over whether to believe models - and consequently, scientists. This fight is going to have enormous political and social consequences for billions. But there is a lot of confusion about models in general. This is an area I know a bit about, so I'm going to try to briefly explain how they work. And while I will continue to stick up for epidemiological modelling efforts, I'm now worried about the hugely influential IHME model and will explain why.

Unfortunately, scientists use the term 'model' to mean a huge range of things. We confuse each other when we don't specify what we mean further; I can't tell you how common it is in my field to stop someone and ask "What kind of model are you talking about?". Here are 3 types of model that are relevant to this discussion (note: this is just a simple shorthand I came up with for this post, it's not some actual classification):

1) I plot some data on a graph and use some algorithm to draw a straight or curved line through the data. The line - or its mathematical description - is a model. If I want to predict/forecast what will happen outside the range of the data (e.g. in the future, if time is on your X axis), I just extend the line on the graph.

2) I expect to see a particular shape (e.g. a 'U' shape) either because I've seen data on a similar topic before or it seems logical. When I get new data, I plot it and then draw that same 'U' line over the new data. It can be stretched or squashed to 'fit' the new data but it will still be a U. The U line is a model. If I want to predict/forecast outside the range of the data, I can again extend the 'fitted' line on the graph.

3) I write a mathematical description of how a system works (a 'system' can be essentially anything that changes). This takes the form of a set of equations that capture the logic of the system. These equations jointly form a model. You can 'fit' these equations to data to understand a specific example of the system, or you can keep it abstract to understand how they work in general. If I want to predict/forecast what will happen in a specific example, I can no longer just extend a line. Instead, I must solve the equations to understand what will happen.

You can see that there is increasing complexity/sophistication in these three types of models. They all have their place, and I use all three in my work. In some situations a complex model will give you the same answer as a simple one, but the complex model will involve a lot more work and error-checking, so we go with a type 1 or type 2. But these can fail badly in some situations, especially when you are using them to predict/forecast outside the range of the data. For some common shapes, they will predict infinity or negative infinity if you extend your predictions far enough, which makes no sense. Type 3 is much more robust to this because it is built on and constrained by the logic of the system. However, they can also predict garbage if your equations are wrong i.e. you have not captured the logic correctly. A lot of time is spent thinking about ways to diagnose the model to make sure that the equations are correct (and consequently, the predictions reliable). In general, type 3 is more reliable to predict the future, and much of our world is based on science and technology built on this type of model.

Now about COVID-19.

All three types of models are currently being used, with little discussion about the assumptions, strengths and weaknesses. There are many sophisticated models out there (see a list at https://docs.google.com/.../1hUZlVDPfa5C8KgURoP.../edit...), though they are hobbled by poor data to feed into them. But at present, the US political and healthcare systems are making plans that are heavily reliant on the IHME model. You can read the model description here (https://www.medrxiv.org/.../2020.03.27.20043752v1.full.pdf). It is mostly type 2, and in my opinion (and more importantly, that of several epidemiologists) they make some large and worrisome assumptions.

The big one is that the type 2 'shape' that they rely on is entirely based on official data from Wuhan. Why is this a problem?(i) We don't fully know how reliable that data is. Data from Italy and Spain are thought to be off by a factor of 2, for reference, and I expect most people have even less faith in China's official numbers. [EDIT: Laura points out that the model has been recently updated to include data from Spain and Italy. That data is biased too so it does not fix the problem, but it is probably improves things.] (ii) It seems to assume that lockdown measures in the US will have the same effect that they did in China. But China implemented a lockdown that is far more strict that anything envisaged by Europe or North America; I doubt that most countries would tolerate anything comparable. Both of these issues point towards the direction of the model predictions being far too optimistic. As a consequence, their predictions of when the epidemic will peak and when it will end do not seem reliable.

All models make assumptions and have weaknesses, so how do we diagnose whether a model's predictions are good? That's a really complex topic! Epidemic models are not like weather models. Nothing we do affects the weather in the short term, so we can just evaluate how the weather played out after a prediction is made. But we change our behaviour and policies based on epidemic models, which makes important model assumptions wrong; our actions therefore make the predictions less likely to come true. So when the predictions fail, it could be either because of the change we made, or because the model was not good, and it's hard to distinguish those.

This post is long enough, so let's take a look at one graph that indicates that the IHME predictions are a bit questionable. The attached graph from the IHME website shows the number of deaths per day till April 10th and their forecasts thereafter. The shaded area indicates the model uncertainty; roughly speaking, the range of possible values. You will see that the uncertainty is highest immediately after the data stops and then decreases steadily afterwards. This is rather strange; do we really know much less about how many people will die tomorrow than about how many people will die 2 weeks from now? Our uncertainty should generally increase as we look into the future (at least for some time - if you look far enough ahead, uncertainty will decline because the pandemic will end). To me, this suggests that something may be wrong under the hood; the 'shape' is wrong in an important way. The alternative is that this is a case of poor data visualisation.

Bottom line: if you're in the US, be at least mildly sceptical about the official predictions about number of deaths at this point, since many seem to be based off a model whose assumptions are hard to justify. The predictions about the date of the peak seem even less reliable. The model projections could still end up being correct, but I'm not very confident in it and would prefer to rely on a model that includes mechanisms (a type 3 model, of which there are several out there) rather than questionable statistical relationships.

I would like to reiterate that we absolutely need to rely on models to make these kinds of predictions, and epidemiology is a sophisticated science - we are not stumbling in the dark here. But we can and should evaluate individual models critically, especially when they are guiding important policies.


0 Comments

05.04.2020

2/1/2021

0 Comments

 
In an earlier post, I mentioned that COVID deaths have been severely underreported in Europe. Comparing recent deaths in Italy with those in the same months in previous years suggests that the true number killed is about double the reported number. This is because the conditions required to be recorded as a COVID death are stringent; if you die at home, it does not count as being caused by the coronavirus. This data bias is not specific to Italy, and it matters a lot. The official numbers are used for calibrating models that predict deaths and epidemic progression in other countries. If we calibrate models based on death rates that are too low, we risk underestimating the challenge other countries face in the coming weeks and months.

Note that this is just one of many data biases, and some might be in the direction of overestimation. But underestimation is the major threat.

More at https://public.tableau.com/profile/isaia.invernizzi...
Picture
0 Comments

04.04.2020

2/1/2021

0 Comments

 
Underappreciated coronavirus news #2. Same caveats as last time - not an expert, and there is substantial uncertainty so these claims are not made with high confidence. I welcome pushback so I (and everyone reading this) can both learn and spread more accurate information.

1) I claimed in the last post that testing people for antibodies to check if they are immune will be very important in getting societies and economies back on track. However, I neglected one extremely important detail. Every test generates 'false positives' and 'false negatives'. Even values that can sound very good for the whole population (5% for false positive and false negative rate, for example) can provide very bad information for an individual. Suppose you get a positive test result; does that mean you have the antibodies? Well, it depends a lot on how common the antibodies are in the population tested. If the antibodies are present in only 1% of the population (roughly what we may expect for many countries now), then getting a positive result means only about a ~16% probability that you actually have the antibodies. This is a simple, profound, and deeply confusing result that is at the core of teaching the usefulness of Bayesian statistical thinking. So this strategy to open up societies slowly is going to be harder than I thought. If you're interested in the maths (it has major implications for disease testing too), you can read about it at https://betterexplained.com/.../an-intuitive-and-short.../

2) The viruses are spread through the air. This may seem somewhat obvious, but it was contradicted by many authorities based on a difference between the plain-English meaning of that sentence and a more arcane technical definition that hinged on differences between droplets that were bigger and smaller than 5 microns. Things have now been somewhat resolved by evidence that the viruses are spread through droplets smaller than 5 microns too, so the plain-English and technical definitions are in agreement. But this is bad news for us, because it means that the virus spreads even more easily than thought.


3) Related to the previous point: jogging, biking and other physical activity is great in general but probably not ideal right now if it brings you near other people. Especially because exercising makes you and your fellow joggers breathe more heavily, spreading the virus even further. How much could this kind of thing matter? We don't really know. But in one horrifying story, a choir of >100 completely asymptomatic people met to practice while maintaining distance from each other. Nearly half of them later fell ill with the virus. A somewhat extreme example because singing involves intentionally expelling air, but physical activity does have a similar effect.


4) Say you have to go out (for e.g. to get groceries) and you therefore cannot totally avoid people. Some reasonable advice I encountered: imagine that they are smoking/vaping and you are trying to avoid the plume from them. Think about how far smoke can travel too, and how it is affected by air movement.


5) A consensus seems to be building that the proportion of infected people who never show symptoms (i.e. asymptomatic) is less than 50%, possibly much less. This is lower than the estimates I wrote about last time. It is also bad news, because it means that there are many more people to infect and the death rate is higher than we hoped. So remember that even if you and a friend have no symptoms, you are at risk and putting them at risk if you are physically close to them.


6) Even if they do develop symptoms, people pass on viruses before the symptoms appear (roughly for a few days). If you've been exposed to someone who developed symptoms days later, you ought to be extra careful to not expose new people.


7) The number of viruses you get hit with is likely to be important. Which also seems somewhat obvious, but it has important implications. Even if you cannot eliminate all risk, anything that reduces the number of viruses you get exposed to helps. Which is one reason why imperfect barriers such as homemade masks might be useful for personal protection (NOTE: the main benefit of masks is that it prevents you from spreading the virus if you are carrying it). And yet another reason why it is imperative that medical personnel get the very best protection, since they are exposed both frequently and to very high numbers of the virus.


8) There's a lot of concern about the evolution of the virus among members of the public. Some people are concerned that it could become more deadly, and some with a bit of training in evolution think it will become less dangerous. We have no evidence for either of these ideas, and good reason to think that they are both wrong. (i) There's plenty of work been done to sequence versions of the virus isolated from people around the world, so we have lots of information about how it has mutated and spread. To our knowledge, none of these mutations has had any effect on how dangerous the virus is or its spread. Mutations happen randomly and many of them do essentially nothing. (ii) Some pathogens do in fact evolve to become less dangerous to their host, because if they kill their host too quickly, they are less successful at spreading. However, this virus spreads very successfully even before symptoms appear, kills a small percentage of hosts, takes a while to kill, and some people never develop symptoms at all. So it is less susceptible to that kind of evolutionary dynamic. Also, some diseases such as smallpox existed with humans for thousands of years and remained lethal throughout; so it's not wise to expect a few months of evolution to save us.


9) Related to this, when you hear discussion about new 'strains' of the virus, it is not necessarily cause for alarm. What constitutes a strain is a rather technical topic; just because a new strain is found does not mean that there are new or greater risks. Listen to whatever the virologists say about it.


10) Everyone is focussed on their own countries and this is natural. But if rich countries fight to retain protective equipment, it is going to be catastrophic for poor ones. Poor countries already have orders of magnitude fewer doctors and medical equipment. If these few doctors get only minimal supplies of protective equipment, it may lead to deaths on the scale of tens of millions (frankly, my mind can't comprehend losses on that scale so I can't even be confident it will not be worse). This needs a global response; I fear we are not up to it and the poor will suffer in a way we have not seen since...I don't know when really. Somebody more historically-minded can educate me.


11) India is one such poor country, and there's a special place in hell for members of this government, who took a system already completely unprepared and hammered it further by (i) choosing this moment to award a monopoly on protective equipment to one company that promptly raised all prices, and (ii) refusing for months to allow private companies to manufacture protective equipment that the hospitals desperately needed to stock up on. India is on a particularly bad path right now, with tens of millions walking home en masse without food or resources, the government hell-bent on ignoring evidence, suppressing a media that is mostly pliant anyway, and blaming outbreaks on individuals and Muslims. Meanwhile, they have no strategy in place beyond one terribly-planned lockdown, banging some pots and pans together, and recommending cow urine as a cure (to be perfectly fair, which it grits my teeth to do, the cow urine recommendation is not official and comes from a subset of the ruling party).


12) The limited reporting I've seen from developing countries is scary. The lack of cases seems far more likely to be from extremely bad testing than from success in containing the virus or luck in other ways (viral susceptibility to environmental conditions, for example). The wealthier countries are ramping up testing but I don't know to what extent this is true in the poorer ones. This is probably going to get a lot worse. We need international news agencies and organisations to shed more light on what is going on in these countries.

0 Comments

02.04.2020

2/1/2021

0 Comments

 
Exponentials numb us. Each new number seems to lose all meaning. For reference, 2 weeks ago we were at ~200,000 reported cases and ~7000 recorded deaths in total. Tomorrow we will cross 1 million cases and 50,000 deaths. New York alone has more reported cases than all of China (whatever you think of the accuracy of China's numbers, would you have expected this?). All these numbers are underestimates even now.

Huge uncertainties remain and most projections for how this will play out are really, really hard to constrain (as in, come up with trustworthy upper & lower limits). The uncertainty seems largely to be because the data going into the models has lots of problems (governments are not prioritising this). The models themselves have weaknesses, but my sense is that this is a smaller issue right now. So please be sceptical of articles written by non-experts who seem to think that their machine learning, economics, or tech background means that they have special insight into the mathematics of disease. Nothing wrong with those fields, but I've communicated with some of these people and understand the models they are touting; most are staggeringly inferior in this situation to even the simpler epidemiological models being used by the experts.

A lot of the uncertainty appears to be in constraining the upper limits. Just 2-3 weeks ago it seemed like it might be possible - if we were very lucky about several things, and competent - to limit deaths to the low tens of thousands. Now, hundreds of thousands is being relatively optimistic. It's very hard to grapple with the possibility of millions to tens of millions of deaths, but that's within the range of possibilities. Countries like India have not yet got going; much will depend on how successful they are in avoiding an outbreak.

The massive potential losses mean that we have to err on the side of being careful for longer, though the economic destruction will be huge and the decision of when to ease up on lockdowns will be a very hard and acrimonious one.
Right now, doctors everywhere, including friends of mine, are making their own protective equipment to use when treating patients, because our societies are so unprepared. So please, however stir-crazy you might be: stay home. Getting fresh air is great, but for now it is a luxury we cannot afford in many places.

[Update much later: the argument at the end against getting fresh air was wrong. Outdoor activity is very safe as long as not surrounded by crowds]

0 Comments

31.03.2020

2/1/2021

0 Comments

 
A few bits of underappreciated coronavirus news, from a semi-informed scientist without expertise on the topic but some relevant training. I'm trying to be more vocal about this (and encourage you to as well) because the completely misinformed are not holding their tongues, so refusing to speak without perfect expertise is leaving some public spaces to the least qualified. And yes, it continues to be important to wash your hands and stay home, and it will for a while.

1) Masks work. Even home-made ones provide some protection. Home-made ones are better for you to use because it frees up the good stuff for medical personnel, who are putting themselves in danger for lack of basic equipment. I have no idea why the WHO and other organisations insisted for so long that masks did no good. When this is over, these and other awful bureaucratic bungles will have cost thousands of lives at a bare minimum. [Edited to add: don't take more risks because you are wearing a mask or it could be worse than doing nothing. And learn how to use, take off & dispose of them]

2) While there's been lots of doubts raised about China's numbers, deaths in France & Italy are substantially underreported. If you didn't die in a hospital, it wasn't COVID according to official counts. There's also credible reporting that the situation in Japan is far worse than documented. Data biases such as these are a huge problem as we try to predict what's going to happen.

3) A large but uncertain proportion of cases are asymptomatic. I don't think there's a consensus yet, but I've encountered estimates ranging from 8% to nearly 90% (don't assume the average of those two is the best estimate). So even if you and the people you know don't feel ill, y'all could carry the virus. Just stay home. [Edit: newer evidence points towards less than 50% being asymptomatic]

4) Relatedly, it's possible that the number of infected people is a lot higher than we thought, because the other numbers that we have (R0 & date of start, for example) don't entirely make sense together. If so, it would imply that the death rate is lower than we thought. We don't have enough random testing to figure this out yet.

5) Differences in death rates between countries continues to be hard to explain. For e.g. Germany's extremely low death rates are still not well-understood. They have been testing a lot, but it's still a bit surprising. Lots of ideas are being explored, from demographics, social practices, smoking, and weather, to weirder ones like whether the country mandated BCG vaccination (having a bacterial vaccine protect against a new virus would be weird, but it's apparently a serious idea). So far it's all quite speculative.

6) We've ramped up testing in the last 2 weeks but are still testing far less than we should. And yet we've wasted more than a month not trying pooled testing i.e. putting samples from many people and testing them all together. If it comes out positive, at least one of the people from the set has it. All can be quarantined till more information is available. This simple idea might have saved a lot of lives.

7) Other old or outdated ideas that may become valuable now: using plasma from recovered patients to help treat patients (carries antibodies), and variolation (an ancient form of vaccination).

8) Contact tracing continues to be a really, really important follow-up. Despite hundreds of billions of dollars being deployed towards helping people and companies worldwide, little of this money is being directed towards ramping up testing and tracing, which is bizarre. We may need to do this kind of tracing for months to years.

9) Once you've recovered, most experts seem to think you'll be immune, at least for a while. These recovered people are going to be really, really important in getting a country back on its feet. When a country is shut down, it's not just money that is lost - lives are too. Especially in poorer countries i.e. where most of the people are. Germany is planning on making best use of recovered people by issuing them documents that could perhaps enable fewer restrictions on travel and work. Other countries will hopefully do this too.

10) Many countries are understandably banning export of important medical equipment and medicines. But everyone is going to suffer as a result. As medical supply chains get decimated, every country loses access to important components from these bans.

11) In the coming months, if millions do not die, you're going to hear a lot about how we overreacted, and the epidemiological modelling community in particular is going to come under attack. Prepare to fight these smears. This is an incredibly difficult job being done under a huge cloud of uncertainty, and counterfactuals are hard. If they do their job well and their advice is followed, the direst predictions will not come to pass. Proving them wrong by taking strong action is success and we should celebrate it. This does not mean these models are immune to criticism - there has already been strong, fast and justifiable criticism of some that have been published.

12) There's a huge number of volunteer efforts offline and online, and you don't need any particular expertise to be useful. If you feel up to it, you can join the fray. Ping me if you want suggestions.
Take care of yourselves, your productivity is probably not as important as it seems right now. We're in this for the long haul.

IMPORTANT: if you think any of these are wrong or questionable, please let me know! This is a fast-moving topic and I'd like to be accurate.


---------------------

In response to a question about the effectiveness of masks: 

I agree that wearing a mask will help a lot less and might even be worse than nothing if it means that you (1) touch your face more, or (2) do riskier activities that you would without a mask. I think the solution to this is education, not dissuading people from using masks, just as we don't dissuade people from using seat belts because they might drive more riskily.

My take-aways from the general reading listed below are: (1) masks are considerably better than nothing; home-made masks are far from perfect but provide some protection. (2) they probably reduce viral load in transmission, and it seems like that may be important, (3) they are even more important in protecting the public from the infected (whether symptomatic or not) than in protecting the individual wearer, (4) despite these limitations, they can be really important because they can help reduce the rapid growth/spread. In other words, if you care about the population level, it helps to reduce average individual risk by even a little.

Regarding reading, this is the most balanced (non-peer reviewed) piece I've read, by a doctor who summarises the literature. He's a reliable writer.
https://slatestarcodex.com/.../face-masks-much-more-than.../

To summarise a couple of papers (some of which are mentioned in the slatestarcodex piece):
This big review finds mixed evidence but if you peer closely, it's generally in the direction of helping (reviewed well in the piece above) and their conclusions seem too conservative given the present situation:
https://sci-hub.tw/10.1136/bmj.h694

This paper found that masks - including cloth ones - reduced aerosol exposure.
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0002618

This one found that surgical masks provide some imperfect protection against aerosolized flu virus (viral load reduced by a factor of 6 on average):
https://sci-hub.tw/.../article/abs/pii/S0195670113000698

This one shows that normal cloth masks can filter out ~50%-70% of a small virus:
https://www.researchgate.net/.../258525804_Testing_the...

This comparison study based on people who did and did not get SARS has problems but suggests that masks helped to protect:
https://wwwnc.cdc.gov/eid/article/10/2/03-0730_article

Here's a compilation of more papers and a summary of the arguments, though I can't vouch for how reliable/unbiased the author is:
https://docs.google.com/.../1HLrm0pqBN.../edit...

Two additional points. The fact that all the organisations that do not recommend masks in general do recommend them for infected people and for caregivers is clear evidence that they too believe that masks work. And the few countries that have normalised mass use of masks are doing relatively well. The comparison with East Asian countries is challenging because there are many differences between those countries and Europe/the US. But the Czech republic has also done pretty well in recent weeks after mandating mask use (note: I have not investigated this particular point carefully, so my confidence in that story is not very high yet).

---------------------
​
In response to a comment about masks mentioning this paper: 

There's two comparisons to think about here. The paper you link to compares cloth masks with medical masks and finds "the results caution against the use of cloth masks".


The result makes complete sense to me, but the conclusion is too broad. Because right now the comparison I and many other people are interested in is not cloth masks vs. medical masks, but cloth masks vs. no masks. And I contend that the evidence supports cloth masks if the alternative is no mask at all. I've posted a few relevant links in my response to the previous question. Additionally, if the public uses cloth masks, they both reduce their risks and make more of the good masks available to medical personnel, which is very, very important.

Of course, people should definitely not take additional risks because they have protection. Everyone should still be staying home. 
0 Comments
<<Previous
Forward>>

    Why I did this

    I am not an expert on Covid, viruses, or vaccines, but I am a scientist with relevant training. I believe we have a responsibility to clearly communicate science to the public, especially in emergencies. So I started to write summaries of Covid developments on facebook in March 2020 to help friends and family understand the situation as it unfolded. This is an archive of those posts (created much later).

    I also tracked the spread of alarming Covid variants for a few months at http://covidvarianttracking.com/ and mapped the consequences of faster variant spreading at https://tabsoft.co/2YwHCmZ.
    ​

Powered by Create your own unique website with customizable templates.