April 6, 2020
I had a bit of a special event last week in that my group’s work was mentioned in a tour de force New Yorker article by the beautiful writer Siddhartha Mukherjee:
Sid mentioned the following article which shows that the amount of human herpes virus 6 shed by mothers and siblings in Ugandan households is predictive of transmission to infants within the next week:
For aficionados, this is an even cooler paper which shows that the amount of cytomegalovirus (CMV) being shed in an infant’s mouth is highly predictive of re-infection of the mom within the next week. We think that this re-infection of the mother may be the cause of mothers infecting their next child in utero or at birth. Congenital CMV remains a sizeable public health issue across the globe and this circuitous path to an infant being struck is remarkable:
In any event, this was really an unexpected surprise which brought me great joy for several reasons. First, the work was shared among two fantastic colleagues and friends, Bryan Mayer & Soren Gantt, both of whom are mentioned in the article. Bryan and Soren are unassuming, low key geniuses. Soren is always on the leading edge of the curve. He is full of ideas, many of which are so out of the box, that they only sink on second or third pass. Bryan is the ultimate data purist and skeptic. His BS filter is next level. Few people can logically and quietly destroy an analysis plan or a paper as effectively Bryan. He is one of the nicest people I know, but we are all a bit scared of him. This project bumbled along in our hands for a couple of years and was really enjoyable for its entirety.
Second, the work is pretty obscure. It is one of my group’s B side productions. We studied HHV-6 which is an important human pathogen in transplant patients, but not in Ugandan kids. To be honest, we did this paper more for fun than practical value. We just thought it was a cool concept. To have someone of Sid’s intellect notice it, and get excited about it, was gratifying.
Finally, the greater scope of the New Yorker piece involved viral load which is my group’s bread and butter. Here is my ode to viral load and how we model it:
On March 12, I wrote about R0 in a population and epidemics in general. I spoke of three things: sparks (single infected people) which can seed an epidemic in a local region; conditions on the ground which may allow a fire to take off (R0>1) or not (R0<1); and in the event of a fire, how quickly and effective the fire department is at moving Reff (real time measure of R0) from >1 to <1. I then gave a framework using these 3 pieces of the puzzle to contrast events in Wuhan, Taiwan, Vietnam, Korea, Italy, Iran and the US:
Similar principles apply when trying to understand whether an infection will take off in a person. Once again, the first event is a successful spark which in this case is defined by a single virus entering a cell in a person’s body and making copies of itself. Bryan’s paper was one of the first to show in humans, that exposure to a greater number of sparks predicts a higher likelihood of transmission. I guess it is sort of obvious that things would play out this way. The more subtle point for HHV6 and the other herpes viruses we studied in Uganda, is that the transmission process is really inefficient. These infants were literally bombarded with low amounts of herpes viruses on a daily basis. However, they only seemed to get infected when there was a high amount of virus around.
This is completely relevant for COVID which seems to stay in a person for a month and perhaps longer, albeit at low levels (at least based on the limited data available). People who were infected with COVID two weeks ago and feel totally well, may still be periodically shedding out of every orifice. If so, there is likely to be SARS CoV-2 in every room in the house. Just not enough for transmission to occur to other family members. Viral load matters.
Sid also talks about a point of intense interest at present which is whether more sparks, or a greater viral dose or inoculum, predisposes to more severe disease. We all get the sense that this is probably true. The stories of young health care workers becoming critically ill seem more than anecdotal. See the story of SARS CoV-1 below, where an inoculum – severity relation was pretty clearly established.
While this relationship seems obvious, mathematically it is challenging. Adding more sparks to the mix does not change R0 (see next paragraph). My group wonders if the concept is analogous to multiple sparks starting multiple fires in different parts of a dry Western state. In other words, viruses land in various sections of the lung creating lots of concurrent small infection sites which sum up to one large disaster. This work is in progress and is a touch more complicated than it sounds.
Once an infection fire starts, the key is then the R0. If you are a spiritual person, perhaps save a space in your prayers tonight for gratitude that R0 is less than one in your body for dozens of potentially fatal microbes. In our stem cell transplantation patients, we see uncontrollable lung infections from viruses that typically only cause the common cold, or molds which are omnipresent in dirt. With past exposure to these viruses and a normal immune system, or just having good neutrophils in your body in the case of molds, your R0 is less than one. This means that you are exposed to a number of these pathogens on a daily basis but clear them rapidly without feeling a thing, and that you are able to survive and thrive for this very reason. Do not let all of this talk of cytokine storm bias you. The human immune system does most of its work quietly and efficiently. You only notice it when it is not there.
For COVID-19, the R0 for a vast majority, and perhaps all, humans in November 2019 was >1. To the best of our knowledge, humans had little predisposing exposure and therefore immunity to SARS CoV-2. Much like the virus can spread exponentially from person to person at Mardi Gras, it can also do the same within a person’s airways and lung during an initial unchecked period of infection. We are all dry kindling.
From existing data, it looks as if SARS CoV-2 viral load peaks in the nasal passage at 2-6 days after infection. This is when Reffective=1 and occurs because the fire department has arrived. From initial data, it appears that COVID clearance is protracted over many weeks. There are also periodic flare ups with brief higher viral loads, though it is unclear whether this is true biology or issues related to accurate sampling of the virus. Hopefully, early antiviral therapy can accelerate elimination of infection.
The point of tracking viral load so closely and using mathematical models to capture the dynamics of the system is that it allows one to probe the timing and intensity of the immune response against a virus, even without taking a single measurement of the immune cells and antibodies mediating that response. This information can be crucial to the design of clinical trials and for understanding why the immune response is successful under certain circumstances and fails under others. Some models account for space as well and try to recreate the 3D architecture of how an infection spreads in and is eliminated from tissue
The magic is that no two human viruses have quite the same pattern of viral load over time. COVID has a somewhat sharp initial peak, then a plateau, followed by a steep drop off. Influenza looks like this as well, but the timeframe is condensed into less than a week. HIV levels in blood are just a flat line in blood during chronic infection. However, treatment induces a succession of phases, where the virus disappears rapidly, then moderately, then slowly and finally not at all. Each slope represents the death of a different type of cell. On causal glance blood levels of hepatitis B and hepatitis C during treatment do something similar but the slopes and the biology are actually entirely different. Genital herpes looks like the silhouette of a jagged mountain range with multiple, unpredictable sharp peaks and drops: each peak represents a time when R0 is changing rapidly from >1 to <1. EBV, the virus that causes mono has a distinctive Table Mountain like morphology. Oral CMV in infants rises gently in a steady and predictable fashion over many months and then decreases at a pace so slow that it is barely perceptible.
The immune forces that govern these contrasting shapes for different viruses are mysterious. Scientists are slowly developing tools to measure the local immune response with the same frequency and accuracy as viral loads. We are entering an exciting period where math models can be used to probe the complex interplay between virus and host in enormous detail. It is a fun time in our field and I hope we can leverage some of this technology towards advancing human health.
The role of math modeling. From the Atlantic, Zeynep Tufekci eloquently described what I tried to articulate in last week’s blog. Models are useful, and necessary, under our current circumstance. However, they are not precise prediction tools. Rather, they help inform policies to the extent possible given the available data.
Data visualization. A couple of friends / colleagues sent me these 2 tools this week and I think they are spectacularly useful. The Children’s platform allows a comparative state by state glimpse of total and per capita infection. A quick glimpse at Washington’s per capita case and death rates show that we are crossing other states meaning that our social distancing efforts are having an earlier effect. The per capital glimpse shows that many states are quite similar but not all. New York truly is getting slammed and is by far the highest per capita. Oregon appeared to start flattening the curve before things really got started. Good job neighbors but our food is still better, and our volcanoes are bigger.
The second domo.com website is a tour de force for data nerds. Kudos to the developer. I particularly like the per capita by county breakdown: it is an unmistakable trend that rich ski resort towns are disproportionately getting hit.
SARS CoV1. My team is trying to look for a way to link viral load exposure to viral load in the person to disease severity for SARS CoV2. We may have an opportunity to do this in assisted living facilities outside of Seattle. In thinking about this problem, I went to the SARS CoV1 literature and found these two papers which I would say are 2 of the coolest epidemiologic papers I have ever read. I would go as far as to call these magnificent works of art.
The papers describe outbreaks of SARS in the enormous Amoy Gardens apartment complexes in Hong Kong (picture included next to today’s link) In the EID paper, they link proximity of infected people to the “index case” or first infected person. Closer proximity resulted in higher viral load and a much higher death rate. This is all attributed to virus laden particles from the first guy’s toilet, the southwest wind, overcrowding, unconnected pipes (yuck) and maybe some rats.
In the NEJM paper, fluid dynamics principles are used to predict the flow of the “virus laden plume.” Oh boy. Makes me want to live precisely in the middle of nowhere.
I can’t help but wonder about New York. I have heard some very bright people casually surmise that New York is worse than everywhere else in the country because of the crowding. That may be true but is not a sufficient explanation. Taipei is crowded. Hong Kong is crowded. Every secondary city in China outside of Wuhan is jam packed. Ha Noi is CROWDED. Many of these cities have subways. Yet, none had an explosion like New York.
Within the US, the devastation in New Orleans makes sense: Mardi Gras. However, there is a story about the tragedy in New York that has yet to be told. I can’t help but wonder if a set of airborne transmission events over the same few days increased the early number of cases and catapulted New York to rapid exponential growth more quickly.
The last paragraph is pure speculation on my part. I am probably wrong. I just hope somebody figures it out. There is so much to be learned for the next time (there will be a next time) that this happens.
On the long haul. I found this perspective from Marc Lipsitch and Yonatan Grad to be helpful. I think it is worthwhile to remember and to remind family / kids, that we are just getting started. This will be a long battle. We need to start thinking about how we will handle the 2ndwave.
On a professional level, I am trying to help manage the first wave as best I can. I am the furthest thing from a hero. Like all of my colleagues, I am seeing patients as before. I occasionally cover the COVID pager at UW. If local ICUs and ERs here get overrun, I will gladly sign up to do shifts to relieve beleaguered and sick colleagues.
However, as far as my team’s research goes, we are focused entirely on preparing for the 2nd wave. We want to learn about test and treat now, to perfect it for the fall.
SARS CoV2 epi in Europe. This state-of-the-art report from Imperial College estimates changes in R effective (real time R0) in ~12 countries as a result of different scaled interventions. It is interesting to compare R effective in Sweden where they are implementing less drastic measures to Italy and Spain, where they slammed on the breaks. Be careful as the shaded green represent the considerable uncertainty with these estimates. Check out Table 2 which shows estimates of lives saved with interventions in different countries: 38,000 lives saved so far in Italy which is more than 2.5 times the actual number of deaths. Imagine if Italy had done nothing! Table 1 shows estimates for % infected (again with considerable uncertainty which is due to the fact that we stilldo not know % asymptomatic infections). Spain is at 15% (95% CI 3.7%-41%). Germany and Denmark are much, much lower. This level of infection in Spain has me wondering about early herd immunity https://www.imperial.ac.uk/media/imperial-college/medicine/mrc-gida/2020-03-30-COVID19-Report-13.pdf ------------------
Exponential growth of bullshit. The current version of the republican party is so disappointing for so many reasons: the cruelty, the racism, the ham-handedness in managing all things, the nepotism, the laziness, the disregard for science, and the fealty to one of the world’s scummiest humans (also a father to some truly wretched people). If I were a fiscal conservative (I am not but was once grateful for this balancing force in western politics), I would be equally appalled at the complete abandonment of conservative principles.
The current crop of conservative media and politicians are more than happy to tout ideas that are easily proven wrong, to then be proven wrong, and to gleefully move on to the next barrage of dishonest and ignorant missives.
The most obvious, stupid talking point from the right / Fox in mid-March was that COVID was overblown. Imagine being so wrong about this, telling millions of people not to worry and as a result being indirectly responsible for so much suffering and dying. And to then feel no remorse. To not apologize to the families of the dead. To not apologize to the health care workers. What horrible people.
I selected the article below based on the graphic which shows an exponential curve of a bullshit idea that propagated through right wing media and into the minds of millions of followers. It is a death spiral.
This article describes the 70 days where our federal government failed us. Thousands of dead people, democrats and republicans alike, because of this. All in plain view for the rest of the world to see.
Cats can get infected. At least in this small study, they do not appear to get sick. I am a cat person, so this is just another little detail about this virus that pisses me off. This data may or may not matter in terms of viral persistence in cities and households, but definitely needs to be followed pretty closely.
Truthfully, our two cats have been saviors during this trying time. If the feds come to get them, no way we give up the goods. We’ll hide them in the attic like contraband.
Masks. This article is deservedly getting mentioned broadly and being touted aggressively as evidence that masks prevent infection. It is an extraordinary piece of work. I think it provides fairly powerful evidence that masks lower the amount of influenza shedding in droplets, as well wimpy coronavirus shedding in both droplets and aersosol (I am not a religious adherent to p<0.05) but no benefit for rhinovirus, the most common virus associated with the common cold. There are so many cool details in the paper: the use of a bioaerosol collecting device to separate big droplets from small aerosols, the standardized experimental technique, the attempt to confirm PCR findings with viral culture, in certain cases the use of certain study participants as their own controls, and the compelling results. Great work by the authors! I love that for influenza (extended figure 7), very tight correlations between levels in swabs, droplets and aerosols are shown, particularly if you use your imagination and remove the negative samples from the aerosols and droplets. The same is maybeevident for coronaviruses (not SARS but the wimpy ones, extended figure 6). This really does suggest to me that the level of virus we detect on a swab is predictive of a person’s infectiousness. Does the study provide enough evidence to unequivocally prove that widespread use of masks by the general public would lower transmission rates? No. In epidemiology, proving causality is a daunting task. I would say the study shows high biologic plausibility that masks have the potential to lower transmission. The problem is that the study is the furthest thing from a real world setting where people fiddle with their masks, put them on improperly and have exposures lasting more than 30 minutes. Do I think the study provides enough evidence to wear masks in a time of international crisis? Yes.Particularly when considered along with the meta-analysis I quoted last week which looked at more real-world situations and given that there is likely little harm in wearing masks provided they are not being hoarded from local health care providers, then I am very supportive. https://www.nature.com/articles/s41591-020-0843-2#Fig2 https://www.nature.com/articles/s41591-020-0843-2/figures/8 https://www.nature.com/articles/s41591-020-0843-2/figures/7 ------------------ India. Arundhati Roy’s piece is painful to read. It describes themes of poverty, classism, religious hatred and far right fanaticism, all colliding with an indifferent viral particle which follows much more simple rules. I am nothing close to an India expert and cannot vouch for the accuracy of all of the reporting. However, the themes are relevant for other hard-hit countries, like Brazil or the US, where many of the same forces are at work, albeit with different hues and nuances. However, the extremes in India seem more extreme, the sadness even more profound. https://www.ft.com/content/10d8f5e8-74eb-11ea-95fe-fcd274e920ca?fbclid=IwAR0P2TckfN-KJwFsi_kE2CHtRRxcf0itPUhLBQw7JYJGrJLlDUelVmrsO2c
Native Americans. Chihana and I spent one of the best months of our life in Tuba City Arizona during medical residency (Thank you Jon Murrow for this tip man years ago.). I did a clinical rotation at the local hospital. It was simple living. I used to walk home to eat lunch. Most evenings, we would drive to a new canyon to take in the otherworldly beauty of the desert southwest. Clinically, I went on home visits to see elderly folk who were hours and hours away from Flagstaff, but still received state of the art care. We also have a soft spot in our heart for Lummi Island, Washington mentioned in the article. Both places are special. Both places are clearly vulnerable. I can only pray they are spared. https://www.washingtonpost.com/climate-environment/2020/04/04/native-american-coronavirus/ -----------------------
Heroism. I could list dozens of articles here. I am grateful to the journalists who are sharing these tales. This is a bizarre tragedy, observed passively from the peace of our own homes. The true story is in the hospital wards and the emergency rooms across the globe.
Please also remember that in the absence of social distancing, things would be 3 times worse.
The call out to nurses is mandatory. Read it twice. It is painfully obvious that nurses spend much, much more time by the bedside than physicians. When this is over, the death toll among nurses will be higher. We should all be so grateful.
Mask creativity. Check out this video from a great friend from med school who is now in Portland, Oregon. It is just one of hundreds of great stories of local ingenuity and teamwork getting us through this crisis. Our federal government has failed us but our best citizens persever.
On grief. An extraordinary, touching piece of writing from Wynton Marsalis to his father who died from COVID this week.