Many of us have had our lives dramatically affected over the last few months by the COVID-19 pandemic. If you’re wondering when and how things are going to get back to “normal,” you’re not alone! We’ve been researching these questions and have put together for you the information that we’ve found. This post will cover the six major unknowns that impact how and when governments will get society back to “normal” and the four strategies that they might use to do so.
This is a time-sensitive article based on information compiled by our founder, Spencer Greenberg (a mathematician and an entrepreneur) as of April 27th, 2020. We’ve done our best to present an accurate picture of the COVID-19 pandemic as it stands now, but please note that we’re not experts in epidemiology or pandemic-response strategies. If any of the information below is inaccurate, please reach out to us and let us know (ideally providing a credible source) so that we can update our article.
If you’d prefer to see the content of this article in video format, check out this link to see Spencer talk you through the following information! You can also access the slides for the presentation here.
Six Big Unknowns
Before we get into the four strategies for getting our daily lives back to “normal,” it’s important to consider the features of the virus where major unknowns still exist. The choice of best strategy will ultimately depend on resolving these unknowns:
Infection Fatality Rate - what percentage of people who get COVID-19 die from it? This rate is going to vary by population (based on age and pre-existing health conditions), but for any given population, it is a measurable quantity. The most accurate estimates of the fatality rate seem now to be in the 0.3% to 1.3% range. Note that at least 0.14% of all people in New York City have already died from COVID-19 (12,287 confirmed covid deaths as of April 29th, out of a population of 8,400,000 people). Excess mortality (more people dying than we would typically expect who haven't tested positive for COVID-19) suggests that the number who have died from the virus in New York City is substantially higher. Since fewer than 50% of people in New York City appear to have the virus, this puts the infection fatality rate (in New York City) at an absolute minimum of about 0.3% (for populations similar in age and health comparable to people in New York City). To compare this to the regular flu, about 0.13% of people who have flu symptoms end up dying from the flu, but the estimated infection fatality rate for the flu appears to be only 0.04% (since not everyone who gets it is symptomatic). That means that COVID-19 appears to be at minimum 7x and perhaps as much as 32x deadlier than the flu, while also spreading through the population at a much, much faster rate than the flu normally does. There is some chance that there's something strange about what's happening in New York City that could make the infection fatality rate there higher than in other places (e.g., if especially vulnerable populations were being hit disproportionally hard), but it's hard to say at this point.
Control Openness - how much can we ease societal restrictions (getting economic activity close to “normal” and allowing individual freedoms) while still preventing the virus from exponentially spreading through the population?
Vaccine or Treatment Time - how long will it take to make a vaccine or a form of medical treatment that very significantly lowers the fatality rate of the virus? Some scientists estimate that it will be 12-24 months before a vaccine is ready, but there is also the question of whether it is even possible to make a vaccine works reliably for this virus (while we have developed vaccines for some viruses, others we have never developed a vaccine against, and still others require a new vaccine every year due to mutating strains).
Immunity Time - how long does an individual’s COVID-19 immunity last after they have been infected with the virus? It’s almost certainly the case that for a few weeks - and probably for a few months - someone who has had COVID-19 won’t be able to get re-infected, but will this immunity last for as long as a year?
Existing Prevalence - what percentage of people have been affected with COVID-19 in a given area? The prevalence of the virus will vary significantly by region, and change quickly as the virus spreads. It’s worth noting that the number of confirmed positive cases is much lower than the actual number of infected people, as most regions are far from able to test everyone who has symptoms (the actual number of infections could be 5x or even 50x the number of confirmed cases).
Complication Rate - what percentage of people who survive COVID-19 end up with serious long-term health complications? Many different health problems seem like they may be associated with severe cases of COVID-19, including lung damage from pneumonia, ARDS (Acute Respiratory Distress Syndrome), or sepsis, but the frequency and duration of these effects are unknown.
Regardless of which long-term strategy is decided upon, it is essential that society continues to take the following actions:
Keep the rate of virus growth low enough to prevent hospitals from being overwhelmed. As long as we think it really does help people to go to hospital, we want to keep hospitals at manageable capacity so that sick people can access them when they need to. Fortunately, hospital overruns have so far been less of a problem in the U.S. than original projections suggested.
Increase production and distribution of medical supplies. In many regions, hospitals still don’t have all the supplies that they need, which has led to hospital staff becoming ill and, in some cases, even infecting already vulnerable patients.
Develop and test vaccines and other treatment options, like antivirals, as fast as possible and in a robust, rigorous way.
Increase the capacity to test for COVID-19 infections, as many people who are displaying symptoms of the virus, or who have been in contact with an infected person, are currently unable to get tested.
Increase rapid contact-tracing capabilities. When someone is infected, we want to be able to track and notify all the people who have spent time with that person and make sure they can isolate themselves to prevent further spread.
Right now, the big obstacle for most governments is getting R - the effective reproduction number - under control. R tells us, on average, how many other people a person with COVID-19 will infect. For example, if, on average, a person with COVID-19 infects two other people (before they are no longer infectious), then R = 2.
R is affected by a lot of different factors, including:
how inherently infectious the disease is.
the behaviors of individuals (e.g., whether people choose to stay at home).
how many people are immune (if people have already had the virus, they won’t be able to catch it from someone else).
early detection, isolation, and treatment of those infected.
the environment (e.g., temperature, humidity, sunlight).
R is not be confused with Ro (pronounced R naught), which is the basic reproduction number. Ro assumes a) no immunity in the population, and b) no behavior change or government action. Ro is probably between 2 and 4.5 for COVID-19, whereas Ro is about 1.3 for the regular flu, meaning that, with no intervention, COVID-19 spreads far faster than the flu. Ro reflects the spread of the virus in a hypothetical situation, whereas R reflects the spread of the virus in our current situation.
R > 1
When R is greater than one, this means that, on average, a person with the virus infects more than one other person, which causes the number of new people infected each week to grow exponentially. This is what we saw happen initially with COVID-19 in many countries.
R ≈ 1
When R is approximately one, this means that, on average, the number of new people infected each week remains relatively consistent; it is neither growing exponentially or shrinking exponentially and remains in a steady state.
R < 1
When R is less than one, this means that, on average, each person with the virus infects fewer than one person, and so the number of new people infected each week will fall in the population.
As well as keeping R below one, we might also need to control the variation of R in subpopulations to prevent explosive virus growth. For example, if a subpopulation has an effective reproduction number of three (i.e., R = 3), despite the average reproduction rate in the population being less than one (i.e., R < 1), the reproduction rate in this small subpopulation will grow exponentially and may rapidly impact the general population average.
Four Big Strategies
Based on our long-term aim of getting daily life back to “normal” while keeping R below one, there are four big strategies we can use against COVID-19:
STRATEGY 1: LOCKDOWN
LOCKDOWN involves strict controls on behavior, like stay-at-home orders and the closure of non-essential businesses, to get R < 1 and (eventually) greatly reduce the percentage of people with active infections. This is the stage that many regions (including much of the U.S.) are at now. Success in this strategy requires the vast majority of the population to abide by these controls. LOCKDOWN can play a very important role in reducing the number of active infections and preventing hospitals from becoming overwhelmed. However, LOCKDOWN comes with enormous costs to both individual freedom and economic security, preventing many people from earning a living. This makes it an unrealistic long-term strategy. It is, unfortunately, also unlikely that we could use LOCKDOWN to fully eradicate the virus.
Is the infection rate falling in the U.S.?
Data from people taking their temperature with smart thermometers in the U.S. is able to show us that the number of observed influenza-like illnesses has fallen below the expected number since people have been in lockdown. This data is imperfect because healthy people may also simply be taking their temperatures more, but it provides some evidence of the effectiveness of lockdown.
Data from Google searches also shows us that the number of people searching “I have a fever,” “I have a cough,” “I have a sore throat,” or “I have shortness of breath," is almost back down to normal after a spike in February and March. This data is also imperfect, because an increase in such Google searches doesn't necessarily correspond proportionally to an increase in illness, but it does provide some evidence that lockdown is working.
Data from worldometers shows us that the number of deaths in the U.S. (which is more reliable than the data on confirmed positive infections, due to most regions’ limited testing capacity) have stopped growing exponentially. On this logarithmic plot, a line means exponential growth, and the curve tilting to the right indicates that the exponential growth is slowing.
The three charts above all suggest that lockdown is working to effectively reduce the number of weekly active infections. Of course, we don't know for certain what these curves would look like if the U.S. was not using the LOCKDOWN strategy. Still, data in dense cities like New York City did demonstrate that the virus was spreading very rapidly at an exponential pace prior to lockdown.
STRATEGY 2: CONTROL
CONTROL involves easing current restrictions on society (getting economic activity back to normal and allowing individual freedoms) while keeping R ≤ 1. Success in this strategy requires the ability to track R very accurately, which means huge testing capacity, and is also likely to involve continued behavior change (like social distancing, people working from home when they can, and perhaps the use of masks). It's important to note that the control strategy does not bring life fully back to normal, but gets as close as possible while still controlling virus spread. This strategy would be used to wait things out until we have vaccines or more effective treatments, at which point we would likely switch to one of the other strategies.
Is CONTROL a good long-term strategy? This hinges on two of the six unknowns.
Control Openness - the more we can ease restrictions on businesses and individual freedoms while still keeping R ≤ 1, the more appealing CONTROL is as a long-term strategy, as it would allow us to go back to living a somewhat “normal” life. However, it might be true that it is very difficult to keep R ≤ 1 without many severe restrictions on peoples’ lives (and preventing the economy from recovering). We don’t yet know enough about how many restrictions we can lift and still keep R ≤ 1 and, if we risk lifting restrictions too quickly, we might have to rapidly return to lockdown to get the virus under control again. As we get better and smarter at Control Openness (e.g., by leveraging effective strategies and technologies to reduce the viruses spread), the CONTROL strategy becomes less costly.
Vaccine or Treatment Time - the sooner that we will have either a vaccine or effective treatments for COVID-19, the more appealing CONTROL is as a strategy. Conversely, if a feasible vaccine or effective treatment remains years away, then we would have to endure CONTROL for a long time, which imposes greater costs on society.
STRATEGY 3: COCOON
COCOON involves opening society to all low-risk individuals as fast as possible without overwhelming hospitals, while keeping people who are elderly, immunosuppressed, or otherwise high-risk “cocooned” safely in isolation. This strategy relies on the principle of “herd immunity,” which means allowing some people to get infected, recover, and then be naturally immune to further infection. The more people that have immunity, the fewer people there are to become infected, causing the R of the virus to fall naturally. If carefully applied, this strategy doesn’t lead to everyone getting the virus, but rather tops out at 40% to 80% of the population (this calculation depends on knowing the Ro, which there is still some uncertainty around).
Critical to the COCOON strategy is the idea that the risk of death from COVID-19 seems to vary tremendously based on age and pre-existing health conditions. For instance, early data suggested that people who are over 80 years old may be at more than 50x the risk of death than people who are younger than 30. Similarly, people with severe pre-existing health conditions seem to be at greater risk of experiencing extreme negative effects from the disease
Success in this strategy requires being able to “cocoon” high-risk individuals (and any low-risk individuals they live with) effectively. Many countries haven’t succeeded in this so far, as demonstrated by the high rates of infections in elderly care homes. Even if this strategy is implemented successfully, a drawback is that it requires severely limiting the freedom of high-risk individuals, some of whom might already be more at risk of suffering loneliness and depression. If you're interested in the effects of COVID-19 on mental health, you may want to see our blog post on this topic here.
Is COCOON a good long-term strategy? This hinges on five of the six unknowns.
Infection Fatality Rate - the lower the infection fatality rate is (specifically, the lower the fatality rate is in low-risk individuals), the more appealing COCOON is as a strategy, as it makes it more acceptable for individuals to risk infection.
Vaccine or Treatment Time - if it takes longer to find an effective treatment or vaccine for COVID-19, then the more appealing COCOON is as a strategy, as it may allow low-risk individuals to resume a largely “normal” lifestyle sooner.
Immunity Time - COCOON will be a totally useless strategy if “immune” individuals only remain immune for short amounts of time, an outcome that can’t yet be ruled out. If low-risk people were to get infected and then re-infected over and again, that would increase the fatality rate and further disrupt the economy.
Existing Prevalence - the larger the already infected population is, the less costly it is to use the COCOON strategy and reach “herd immunity,” as we’re already part of the way there
Complication Rate - if a large percentage of people who get the virus suffer long-term health complications, then the COCOON strategy is less appealing, as we’re exposing many more people to the risk of increased negative health effects and potentially even a reduced lifespan.
STRATEGY 4: OPEN
OPEN involves opening all of society to everyone as fast as possible without letting hospitals get overwhelmed. With this strategy, it's likely that 40% to 80% of people will eventually get infected (and perhaps even more than that if the virus becomes permanent and seasonal like the flu is).
Is OPEN a good long-term strategy? This hinges on five of the six unknowns.
Infection Fatality Rate - the OPEN strategy is only viable if the fatality rate of COVID-19 is sufficiently low. Otherwise it would result in an unacceptable loss of life.
Vaccine or Treatment Time - if it takes a long time (or appears impossible) to find an effective treatment or vaccine for COVID-19, OPEN is a more appealing strategy, as it could allow us to return to “normal” life more quickly (albeit at the risk of many lives being lost).
Immunity Time - if prior COVID-19 infection doesn’t grant long-term immunity, OPEN would be a more dangerous strategy, as people would get infected over and over again, driving fatality rates up and further disrupting society.
Existing Prevalence - assuming that long-term immunity is a possibility, a greater number of pre-existing infections would make OPEN a more realistic strategy, as a large amount of the population would already be immune.
Complication Rate - if many people who are infected with COVID-19 end up with long-term health complications, the OPEN strategy becomes less appealing, as we would be exposing many people to the risks of these long-term health problems.
The six major unknowns, and the information we have about them, are going to shift over time. This means that the best societal strategy to use against the virus is also likely to change. It's therefore essential for society to continually track these unknowns in order to make the best decisions. For example:
Infection Fatality Rate
We may develop better strategies over time for allowing society to continue unrestricted while keeping R ≤ 1, perhaps using new technologies or behavioral strategies.
Changes to our environment, such as temperature, humidity, or sunlight, might change how easy it is to keep R ≤ 1 (for instance, some viruses speed up transmission during the cold winter months).
Vaccine or Treatment Time
As time progresses, we’ll have greater clarity on how far away a vaccine or effective treatment is. Note that Remdesivir has very recently shown promise as a treatment in a large randomized controlled trial, dropping recovery time from 15 days to 11 days (highly statistically significant) and perhaps dropping death rates from 11.6% to 8% (unfortunately this drop in deaths is not statistically significant, so is far from certain, and may simply be the result of noise). While certainly encouraging, these numbers are not the silver bullet we would hope for in an ideal treatment.
The more time that has elapsed since people were infected, the more we can learn about the duration of immunity. Furthermore, naturally occurring mutations in the virus could limit the duration of immunity.
To pick the best strategy, or best combinations of strategies, we must increase our understanding and ability to measure these six major unknowns. Right now, we have lots of low-quality evidence, but we desperately need more high-quality evidence. The information that we do have comes from studies and trials that are limited by funding, time-pressure, and the physical restrictions in place because of the virus. It is, however, possible to get greater clarity on these critical unknowns, with appropriate funding and carefully designed studies.
How can we pin down these six unknowns so that we choose the right strategy?
Infection Fatality Rate - we can accurately track all COVID-19 infections and outcomes in the entirety of an age-diverse population where nearly everyone was exposed, which would give us a better sense of the true fatality rate. Most existing studies have problems like high false-negative rates, are only testing for active infection, or conduct poor follow-ups that don’t track the outcome for everyone who was infected in a population.
Control Openness - we can learn from other regions about what worked and failed when they attempted to open society more, and by doing gradual experiments to lift restrictions while carefully monitoring R.
Vaccine or Treatment Time - we can closely monitor vaccine studies and early trials being conducted on the efficacy of vaccines (and other treatment options) to get a clearer sense of the feasibility of an effective vaccine or treatment.
Immunity Time - we can periodically retest (with accurate tests) a group of people who were previously infected to see if they have lost immunity.
Existing Prevalence - we can randomly sample people (e.g., by home address) for serological testing in each region. Most existing studies have limited accuracy due to not sampling in a sufficiently random way, or by using tests that have inadequate false-positive / false-negative rates.
Complication Rate - we can carefully track the health status of previously infected individuals over a long period of time to see what complications they suffer and how long these last.
We're going to be stuck in this difficult situation for a while. But if we gather high-quality evidence about the unknowns related to this virus, we'll be better placed to create the best outcomes for society. If you’re feeling a bit overwhelmed after reading this, you might want to check out our list of meaningful things to do during a pandemic, which has lots of positive suggestions for how to spend your time.
Edited by Holly Muir