Where are we in our COVID-19 management?

Updated: May 11

In response to a friend's concern that the lifting of lockdown restrictions is somehow encouraging and enabling reckless behavior, I put together the following post in which I try to assemble some of my thoughts of where we are and where we should be going. Prospectively apologizing for its length, here it is:


There are many ways to look at what is currently going on in society, and it brings out many fundamental questions about role of government, control, and freedom.  But in my thinking, I’m trying not to get caught up in all of that.  I think the hyper-partisanship in the USA and the world in general is an impediment to finding the best solutions, and think the politics gets in the way.  It really doesn’t help that this is a presidential election year with an extreme and polarizing incumbent who was running on a strong economy, and who runs his mouth all the time about whatever comes into his head.  But all that is a distraction -- the bottom line to me is to try to clear out all of that haze, forget about politics, ideology, and blame, and try to figure out what we know about this disease and what we should do about it.  I think that is the approach we all need to take.

Early on, many Americans took this lightly, like the SARS and MERS epidemics that never came to our shores or directly affected us.  Up until March, most Americans thought this was not a big deal, and even into March, many people thought this was “blown out of proportion”.   Only after the public was educated about the SIR and SEIR models and exponential growth of the pandemic, and the sight of ICUs in Italy being overrun, and not having enough ventilators and having to choose who would get the limited resources did our public get the seriousness of this.  The Imperial College report really woke us up. Motivated by the suppression strategy recommended in that report, one by one, states started lockdowns, gradually being adopted in almost every state in the USA.  We were told by epidemiologistis at this time that it was “too late” to control and contain. The social distancing was sold to society as a means to “flatten the curve” – accepting that until there was a vaccine or natural herd immunity the virus would keep spreading throughout the country, but hoping these policies would spread out the incidence so that the healthcare system was not overwhelmed, and could treat those who were seriously ill, and avoid the nightmares of rationed care and choosing who would live and die.

Well, fast forward 7 weeks, and how have the lockdowns worked?  We clearly have “flattened the curve” in the sense that the healthcare system has not been overwhelmed and the slope of incidence has leveled off.  The Rt rate of the virus has gone below 1 in many locations, leading to reduction in incidents. With the exception of a few hospitals in hotspots, notably in Queens, most hospitals are not even close to capacity, and we have not had the shortages that we worried about.  Even here in center city Philly our hosptials are at 50% capacity, and we have had a major outbreak in our urban communities (12% of black pregnant women in the city are seropositive in study just submitted to Nature). 


As I put in my op-ed, we have also learned a lot more about this disease in the interim: its biggest problem is that seemingly a high % of infected (maybe 50%) are asymptomatic, and many others (maybe 40%) only get mild symptoms, and it seems they are contagious before symptoms appear.  That allows fast spread.  Plus we know more about how it spreads.  Not just when someone sneezes or coughs directly on you, but it can stay alive on surfaces for some time, or can spread a certain distance in the air especially when propelled (by cough or sneeze, but also singing or breathing hard during a workout, e.g.).  Again, this is why the spread is so bad but we know more now.  Also, we see that only about 10% of infected need hospitalization, and only maybe 3-4% need ICU.  Thus, for most infected it can be managed without taxing the health care system too much.  In terms of ventilators, the initial estimates for how many needed were way, way off, by orders of magnitudes.  It turns out ventilators are not the front line tool that we thought they would be – we have found that putting people on ventilators too early just based on % oxygen levels even when the patient seems to not feel too bad is a bad idea, and should only be used as last resort.  We also found that a majority of ventilated patients die, so this is not the life saving device we maybe thought it would be. 

We know who is at risk of dying of this disease pretty clearly.  Older people, especially with certain co-morbidities, most notably those with hypertension and metabolic syndrome.  We have learned it is not primarily a lung disease, even though the lungs are one of the first primary targets once it gets into the body, but targets cells with ACE2.  We see that the disease course that is so devastating and dangerous involves clotting and hyperinflammatory response – not like typical pneumonias.  These provide clues about how to treat which I’ll mention in a moment.  But when we look at deaths, nationally 25% of deaths are at long term care facilities, and in many states (including PA and NJ) it is half or more of deaths.  Plus the other ones are almost exclusively those with co-morbidities.  Not 100% but pretty darn close.  Again, this gives us information that is useful in crafting strategies.  Also, lots of serology tests coming out that show that the number of true cases is about 10-15x the number of official cases, which is bad and good news.  Bad because there are more asymptomatic infected than we thought and thus harder to prevent spread.  But good that it is not as deadly as we initially thought – with a case death rate that seems to be between 1/200 and 1/400 or so.  Still bad but not the end of the world.  If we consider non-long-term care, that cuts the death rate by about ¼, and if we stratify, the death rate among those who do not have certain pre-existing conditions is much much lower.

In terms of treatments, we still have no definitive answers but I see the medical research community thinking more precisely now as some things come more into focus – in the sense that it is being recognized there are various stages of this disease and different stages require different courses of treatment.   The Remdesivir positive results from the this past week are promising, and interestingly Remdesivir based on its course of action seems like something that could be given early in the hospitalization, before symptoms are too advanced.  Other treatments that target anti-inflammatory and anti-immune response to control the cytokine storm seem to be better given later, when a patient is progressing towards cytokine storm induced ARS that seems to be one of the big killers.  Some currently tested treatments with promising results seem well suited for this stage.  The clotting problem caused by the disease is related to some of the same inflammatory cytokines, and so this also raises the potential of other mechanisms of treatment with anti-coagulants.  The medical community is learning some more about treatments and lots of studies are ongoing so we should continue to learn what works and what doesn’t and when it works as these results roll in.  I have seen enough promise in a number of currently studied treatment that I am hopeful once we figure out when/how/for whom to use them they can be effective weapons against the disease and especially managing the small % that become severe.

Getting back to strategies.  the lockdowns were NEVER intended to be long term solutions until a vaccine or universal testing or some other long term endpoint is reached.  At least that was not the discussion in most of society -- they were marketed as a short to medium term solution to flatten the curve. However, the "suppression" strategy in the Imperial College report suggests they be kept in effect until the curve is not just flattened by snuffed out. However, it seems to me that they CANNOT continue indefinitely.  There is a bunch of collateral damage they are doing in society.  Unemployment.  Bankruptcies especially in small businesses.  Neglect of other health conditions.  Mental health problems from social isolation.  At some point supply chains will break down if this keeps up and that is probably already happening in other countries.   These things are causing great harm to society themselves.  The key question is “are they necessary?”  I strongly believe that total lockdowns, while the right approach initially based on what we knew then, are clearly overkill as a mitigation and social distancing strategy.  Since the viral spread ruled out control and contain, I think our strategy needs to focus on mitigation and management.   We have to learn to live with the virus in society until a vaccine is found (if indeed it is found – it is not 100% a single vaccine can control this). 


Going back to why I think lockdowns are overkill – from what we’ve learned, if basic social distancing is practiced, avoidance of large gatherings, disinfecting of often-touched surfaces in public buildings, and public mask wearing is practiced I propose that we can retain a high % of the mitigation benefit of lockdowns while letting people get back to living their lives – and mitigating the collateral damage caused by the lockdowns.  Look at Sweden – they never locked down, and have higher death rate and higher incidence than their neighbors, but for them a huge % of deaths are from long term care facilities as well, and it is not the disaster some expected.  Their economy is taking a hit but far less than its neighbors.  Based on what is said by those promoting lockdown as the best strategy, Sweden should be experiencing uncontrolled exponential growth, but it is not.  We should continue to see what happens there and learn.  They have become guinea pigs so I guess we should just learn from what happens from their strategy.

More about strategies:  Has any country in the world really found an effective strategy?  I wish we had a president who could effectively lead and project empathy and confidence, and that sure would help, but I am not sure what any leader could have done in this situation that would have produced significantly better outcomes.  Countries in Asia have seen this type of thing before and had systems in place to test, contain, and contact trace to control the spread of the virus.  China did it by heavy handed approach of total lockdown, with forced isolation/quarantine/testing of those with virus and their families, using vacant hotels to pull them away.  South Korea had some testing capabilities partially developed for MERS and manufacturing capabilities that they quickly turned towards testing, and were able to quickly implement testing and tracing, and seemingly had already developed contact tracing apps from previous outbreaks, and quickly implemented these to control and contain the spread. 

Outside of these two countries no other country has been able to control and contain this.   Many have adopted some variant of the same lockdown policies as in the USA.  We see that the incidence and deaths are much lower than the static homogeneous-rate epidemiological models predicted, and many people take this as evidence of the effectiveness of the lockdown strategy, and some may think that relaxing this strategy will result in the unconstrainted exponential growth.  We as statisticians know the fallacy of that thinking.  The causal effect against which we can compare is “what would have happened if we didn’t lockdown”, which could have some variants “what if we did nothing as a society?” or “what if we used milder or suggestive social distancing procedures?”.  Because almost every country followed the same lockdown strategy we have little to no information on this counterfactual.  Sweden is the closest we have to that and their results are quite encouraging, suggesting perhaps the additional benefit from more strict lockdowns, while nonzero, may not be that great, and perhaps not worth the collateral damage it incurs.  In the USA, states have a great deal of autonomy in certain matters, which is why states have different responses to this.  This may be a good thing since it is clear that different states, and different counties within states, have different degrees of risk of spread and growth, and one size does not fit all.  But regardless, the states have followed almost the same strategy to date.  That is starting to change now.

So how do we look at some states relaxing lockdowns now, and others digging in?  Do we look at the states lifting some restrictions as “endorsing irresponsible behavior”?  Do we look at the states that won’t relax restrictions as “authoritarian states”?  Well, I don’t think either extreme is right.  States have different degrees of risk, and can effectively manage and mitigate the disease with different degrees of severity. Obviously a place like Manhattan with enormous population density and dependence on crowded public transportation is in a different setting than a small town in Montana where everyone lives far apart, drives everywhere, and even when at the store together are not close to many other people.  We have learned a lot about how this thing spreads, and can put that knowledge to use to construct effective mitigation strategies that may be far far less than total lockdowns in many areas, and may have to resemble more a total lockdown for some time more in more densely populated areas.  One benefit of the lack of centralized federal control is that we are about to get our counterfactuals.  Some states may lift restrictions too much by most of our opinions and seem reckless.  They will choose for their people to be “guinea pigs”.  I don’t like that.  But it will teach us a lot.  What if Georgia does not go back to exponential growth but shows a small uptick that still stays manageable, and their healthcare system is not overwhelmed?   The next few weeks will start to tell us a lot, but like we always do we will overreact to immediate data that comes in, when if we use our brains we realize that any changes made now will not affect official incidence numbers for a couple of weeks, and won’t affect deaths for a month or so.  So remember that when people start hyperreacting to every data point that comes in.  We have one counterfactual in Sweden and are about to get some others in US states.  The question is whether we as a society will be able to set aside our political baggage, unconscious preconceived notions, and a fear that can drive to irrationality enough to carefully evaluate what the data tell us.

BTW, now that we have better data we can model some of these things.  Static epi models are great educational tools to help the public understand exponential growth but don’t reflect the complex heterogeneous realities well.  That’s why I am enthusiastic about some modeling going on here at Penn-CHOP modeling the Rt rate county-by-county and covariate adjusted for density, termperature, and social distancing variables so we can understand how these factors affect the spread rate, and can use them to construct projections that can inform strategies for each locality – and we have to think county level at least as state is too heterogeneous.  One such effort has shown that temperature does have a significant effect – not enough to control the virus but enough to decrease the Rt at higher temperatures, and obviously density is a huge effect and when taken into account the spread in rural areas is so much slower and shouldn’t be treated the same way.  The social distancing variables are also extremely informative, but as these are modeled more precisely, the dial can be turned to assess how relaxing these policies will likely affect future incidence, and the model also accounts for the 90% or so of cases that don’t count in the official tallies because of asymptomic or mild disease and/or inadequate testing, which also have a significant impact on projections and planning.  I think these models are going to be very useful.  This one model developed at CHOP and whose statistical model building was led by my junior colleague in Penn Biostat, was presented to Deborah Birx last Friday and she showed a great deal of excitement over it, saying she didn’t see any other such modeling out there, and proposed for it to be presented before the 50 governors weekly meeting, and talked about bringing it before the president and vice president.  The model is not perfect but tries to glean insights from the information out there, and I think can be used as a tool to craft more targeted and sustainable mitigation strategies.

By the way, in parallel with the construction of these targeted mitigation strategies, we should be ramping up our testing and contact tracing capabilities – not so much for benefit now, but for the next wave presumably in the fall.  It appears likely that in many places incidence will subside to some degree in the summer, setting us up for potential “contain and control” strategies if incidence is low enough when a potential fall surge starts.  We have a little time to ramp this up, and this in my opinion is where the federal government’s planning should be focused (i.e. besides shoring up testing and PPE for long term care facilities that make up a large % of total deaths – we are literally spending trillions fighting this thing and we can’t effectively improve the situation in the known, fixed locations where the pandemic is killing the most?  Shameful and unnecessary).

This is a really hard problem and I am not surprised the USA and other countries have not found the best answer yet.  There are ditches on both sides of the road – too strict a response will irreparably damage other elements of society and too lax a strategy will lead to fast growth and needless death.  I think now is the time to be smarter.  This is not nuclear fallout, where if people step outside their homes they will die, and we need to stop treating it as such.  Lockdowns can’t continue for much longer, but we need to take what we have learned and put it to work in crafting more targeted mitigation and management strategies that are sustainable.   We can do it.  Some good thoughts are out there in articles here and there. It is scary but we have to face the fact that this virus is going to be with us for a while and we have to learn to live with it/manage it and think carefully about the key steps we can take to mitigate and manage it without destroying many other elements of society in the process.




 

©2020 by Covid Data Science. Proudly created with Wix.com