Making decisions when conditions are uncertain
The world is currently experiencing the worst health crisis since the flu pandemic (H1N1) of 1918. To fight the COVID-19 pandemic, which is likely still in its early stages, governments must urgently implement draconian measures despite the incredibly uncertain conditions to prevent the collapse of health care systems. Policy-makers have therefore created advisory committees composed of experts with diverse backgrounds, who provide reasoned judgments about various control strategies and the strategies’ potential political, social, and economic consequences. The French government is receiving guidance from two groups: the French Scientific Council and the Council for Analysis, Research, and Expertise (CARE).
Conditions worldwide as well as control measures in specific countries are changing rapidly. To get an overview of daily shifts in the global situation, you can use several constantly updated online resources. The most useful may be the website run by the World Health Organisation or the interactive map put together by John Hopkins University. You can inform yourself about current circumstances in France here.
Countries are using different approaches to deal with the epidemic, and the measures taken may follow a certain order. Some governments are engaging in mitigation: they are trying to stagger the onset of infections by slowing but not interrupting the virus's spread. This is the policy adopted by the Netherlands and, at least initially, by the UK. Other governments are aiming for suppression by locking down cities containing several million inhabitants as soon as a few cases have been detected. This was the approach taken by China in the province of Hubei.
R0: the variable driving decision-making
R0 is the average number of individuals who are infected by someone carrying the virus
R0, or R-naught, is shorthand for the "basic reproduction number". For a given pathogen within a defined population, R0 is the average number of secondary infections produced by a single infection. In other words, it is the mean number of individuals that an infected individual will infect in turn. To properly assess disease risk, it is crucial to know the value of R0. If R0 is greater than 1, an epidemic will probably spread (at a level that generally correlates with the parameter's value). However, if R0 is less than 1, an epidemic is likely to peter out on its own.
From a mathematical perspective, R0 is the product of the contact rate (the number of times a person comes in contact with other individuals over the course of a day, for example), the probability that contact results in transmission, and the length of the infectious period, the period during which an infected person is capable of transmitting the pathogen. Consequently, policy-makers can use strategies that target these three parameters to bring R0 down below 1. For example, social distancing, school closures and population lockdowns are measures that serve to lower the contact rate. Engaging in protective behaviour and wearing a mask can help reduce the likelihood that contact will lead to transmission. Finally, treatment with therapeutic drugs can shorten the length of the infectious period.
During the first weeks of the epidemic in the city of Wuhan, R0 was probably slightly greater than 2. When the city was placed under lockdown and social distancing measures were implemented, mean daily contacts declined by a factor of 7 to 9, such that R0 dropped below 1 in Wuhan and Shanghai. In general, strategies for controlling infectious agents like SARS-CoV-2 are centred on reducing the value of R0.
When governments strive for mitigation, they are seeking to lower R0 to a value that is close to (but still above) 1: the virus continues to spread within the population but at a lower rate, taking some of the pressure off the health care system.
When governments aim for suppression, they are seeking to reduce R0 to a value below 1: on average, an infected individual will cause less than one secondary case, which interrupts transmission and leads to the virus's disappearance.
What would have happened in the absence of any government action?
Epidemiological models have been run using parameter estimates based on data from the epidemic's first weeks, when no measures had been taken to control COVID-19. They assumed no change in human behaviour (which is a rather unlikely scenario). The models predicted that, under these conditions, a large percentage of the world population would become infected. Although COVID-19 has a relatively low mortality rate (less than 1% [of those infected] based on current figures), the sheer number of infections would result in hospitals being overwhelmed by COVID-19 patients needing intensive care. It would thus become impossible to help other people afflicted with the disease or those with other illnesses who would normally be placed in intensive care units. In the USA, more than 4 million people per year are admitted into intensive care units, and nearly 500,000 of them die despite the high level of care received (mortality rate: 8–19%). Consequently, if hospitals became overburdened, it would have catastrophic consequences because the number of collateral deaths would increase dramatically, although estimating the exact figures is difficult.
Thus, in addition to encouraging people to engage in protective behaviour and wear masks, which are individual gestures that can effectively reduce the probability of transmission during contact, governments must choose between two broad-scale management strategies: mitigating or suppressing virus transmission.
Mitigating or suppressing virus transmission
Controlling the spread of the virus
In the mitigation strategy, the virus is allowed to spread within the population at a more controlled rate. Measures are put in place to lower R0 to a value close to but slightly above 1 while also shielding the population's most vulnerable groups. Mitigation has two objectives. The first is to slow the spread of the virus to delay and reduce the height of the epidemic's peak, which is often referred to as "flattening the curve". The aim is to stagger the onset of infections (i.e., prevent too many infected individuals from appearing at the same time), reducing the chances that the health care system will be overwhelmed. The second is create a large pool of people who have acquired immunity to the virus, an outcome that would help interrupt the virus's spread over the longer term. When R0 is 2.35, around 60% of the population would need to be immune to achieve this goal.
The Netherlands has opted for a mitigation strategy, as stated by their prime minister on March 16, 2020. Various mitigation measures include isolating people who are sick (except those who are asymptomatic, of course), quarantining households where infections have occurred, closing schools and universities, and practicing social distancing exclusively around the most vulnerable members of the population.
Interrupting the virus’s spread
The suppression strategy attempts to directly stop the virus's spread within the population. It is a more forceful tactic than mitigation because it seeks to push R0 below 1. The goals are to rapidly flatten the epidemic's curve, prevent the health care system from being overwhelmed, and eradicate the virus. The main suppression measures are generalised social distancing and the lockdown of the entire population (i.e., not just the segment of the population that is symptomatic). The daily contact rate is thus greatly diminished, and, theoretically, the virus can no longer spread beyond individual households. The Chinese government rapidly applied this strategy in the main cities of the Hubei province. Several European countries did the same, but only after a rather lengthy delay, following the detection of a large number of cases. At present, at least 42 countries or regions around the world have placed their populations under lockdown. For example, Italy, France, Spain, and India all have lockdowns in place. Overall, 2.5 billion people are subject to shelter-in-place orders. Another way to suppress virus transmission is to conduct widespread testing in high-risk areas to identify infected individuals, rapidly trace their contacts, and immediately quarantine them, which was the method used in South Korea.
Virus suppression is a strategy that seeks to buy time to:
(1) develop therapeutic drugs
(2) develop and make publicly available a vaccine that can protect the most vulnerable members of the population or those who have not yet been infected (however, this process could take as long as 18 months)
(3) carry out broad-scale serological testing to estimate the percentage of the population with acquired immunity
In contrast to mitigation, suppression seeks to limit the number of people who become infected. As a result, suppression measures also limit the acquisition of immunity. There is therefore a high risk that the virus will begin to actively spread again once control measures are lifted unless a large percentage of the population is vaccinated. It is crucial to plan for actions that can be taken to prevent the virus's resurgence. The results of mathematical models suggest that periodically implementing social distancing after the lockdown could help control the epidemic in the longer term. China is slowly lifting the lockdown in the Hubei province. It is therefore essential to keep an eye on how events are progressing there over the next few weeks and to draw lessons from these observations.
Which strategy to choose?
Guidance from mathematical models
When managing a health crisis of this magnitude, it is extremely difficult to decide which strategy to adopt. Both have clear advantages and disadvantages. Although suppression seems to have been successful in China and South Korea, there have been substantial social and economic costs that will have long-term effects on the health and well-being of those countries’ populations. Furthermore, when suppression measures are not applied rapidly enough, health care systems may nonetheless become overburdened, as we observed in Italy. Mitigation also has its drawbacks. This strategy cannot fully protect those at greatest risk of falling ill with serious forms of the disease. Consequently, there is still a significant risk that the number of deaths will be extremely high and that health care systems will be inundated.
Mathematical models are important tools that are being increasingly used by policy-makers seeking guidance as they make this choice. The usefulness of these models stems from the fact that they can be employed to identify and rank transmission pathways, predict the number of cases that will be detected in the coming weeks, and compare the effectiveness of different control measures. However, to obtain trustworthy results, model parameter estimates must be calculated using reliable epidemiological data. Also, models must be able to take into account changes in how individuals are behaving. Both of these requirements may be a bit challenging in the midst of a health crisis, where decisions must be made very quickly.
Whether a government chooses to suppress virus transmission, and thus interrupt the virus's spread by imposing a lockdown on an entire country, or it chooses to mitigate virus transmission, by maintaining activity levels but taking the risk that hundreds of thousands will die and the country's health care system will collapse, there will be dramatic economic consequences. However, these consequences may differ in nature. Surprisingly, the models currently guiding decision-making do not take into account the economic impacts of control strategies or the increased social inequality that is resulting from lockdown, which could further destabilise already weakened economies. It is therefore crucial to foster collaborations between epidemiologists, economists, and sociologists so that the social and economic consequences of control strategies are incorporated into the next generation of mathematical models used to guide public health decisions.
Timothée Vergne is a veterinary epidemiologist and associate professor in veterinary public health at the National Veterinary School of Toulouse. His research focuses on the transmission of infectious animal diseases and methods for evaluating the effectiveness of control and monitoring strategies. He is a member of the EPIDEC team within the ENVT-INRAE Joint Research Unit for Pathogen-Host Interactions. Vergne extends his gratitude to Gaël Beaunée (Oniris, Oniris-INRAE Joint Research Unit for Biology, Epidemiology, and Risk Analysis) and Benjamin Roche (IRD, IRD-CNRS-UM Joint Research Unit for Infectious Diseases and Vectors: Ecology, Genetics, Evolution, and Control) for the conversations that helped shape this article.