Tough Road to Recoveries and Hard Balance of Risks

COVID-19: How Data Can Warn Us Early and Show Us the Way and Hope? (10)

April 26, 2020

It’s been almost three weeks since I first wrote here about the outlook and preparation for recovery on April 6 (https://medium.com/@powertocw/covid-19-how-data-can-warn-us-early-and-show-us-the-way-and-hope-06-85ed0ee75b9c). While the growth rates of this wave of pandemic have slowed significantly, the total numbers are still getting grimmer by the day. The global total confirmed cases will pass 3 million when I finish, the total deaths have already passed 206,000.

Around the world, some levels of normalcy are beginning to return in places where control and containment had been effective, enough times have gone by, or the worst seemed to have passed. Schools and universities in China are re-opening, after more than two and half months of delays. South Koreans are going out to the parks and eating in the restaurants. Several European countries, including France, Italy and Spain that are among the worst hit, are easing the restrictions and letting children out to play.

In contrast, there is the unique example of Sweden with a very different strategy of much relaxed control and no shutdowns. Although not publicly acknowledged, it is pointed out by many that this is effectively a strategy to achieve herd immunity. The calculation seems to be based on calculating and balancing risks in economy vs those from COVID-19, e.g. “Sweden’s coronavirus response protects its economy — Swedish official Ann Linde says government feels it’s unhealthy to force citizens to stay inside” (https://www.foxbusiness.com/economy/sweden-coronavirus-response-economy). We will come back to the risk calculations later.

There is also the cautionary tale of Hokkaido, “This Japanese Island Lifted Its Coronavirus Lockdown Too Soon and Became a Warning to the World” (https://time.com/5826918/hokkaido-coronavirus-lockdown/). The new sources are mostly from the other parts of Japan, showing the difficulties in re-opening societies when the control measures, including lockdowns, are not in-sync with each other, and levels of controls differ from region to region.

There have been a lot of controversies and debates in the US about the pathways and timetables of recovery. Most states now tentatively set dates for returning to work in mid- to late-May, nearly two months after declaring National Emergency, and over a month after most states ordered shutdowns of varying degrees.

My early April outline of the controlled recovery model contained the majority of the key assumptions and steps of the emerging proposed plans, with some refinement or revision necessary to reflect the changing reality. I will add some additional clarifications on the control dynamics, and the estimates and balances of the main risks in the general economies and the COVID-19 pandemic.

There is now a growing consensus that the recovery will need mass testing, contact tracing, and optimally, effective treatments with vaccines and antiviral medicines. But experiences so far show that we need more time than one month to put the first two measures in place.

I was too optimistic in assuming that the control and containment measures, including SIP, WFH, SAH, wearing face coverings and shutdowns, would be implemented quickly and uniformly given the gravity of the situations. But the situations have improved significantly as people began to realize the threats and learn the lessons. In the local supermarket today, I saw over 90% of the shoppers and clerks were wearing face masks, keeping distances, and wiping down carts and merchants. If such practices are kept for more than two to three weeks, most existing infection sources would be identified and isolated, and new sources eliminated or kept to a minimum. People will also form certain habits and norms that will help us speed up the recovery, and reduce the potential onset and impact of the following waves.

Now a few word on the control dynamics of the recovery strategy — what is an under damped system and why is it applicable here?

Figure 1. Under damped, critical and over damped dynamic systems.

When a system starts from a place, e.g. when knocked off from and returns to an old equilibrium, or approaches a new equilibrium under control, there are three classes of pathways depending on the strength of damping or resistance (Fig. 1). Under the over damped conditions, the system approaches the goal slowly and never overshoots. Under the critically damped condition, the system approaches the goal with the shortest time with no overshoot. Under the under damped conditions, the system overshoots, then undershoots the goal, and gradually reaches to the goal after multiple oscillations.

So why should we use the recovery strategy of under damped oscillations?

First, unlike in a model dynamic system, we do not really know what the new state will be. Returning to the old is not only impossible, but undesirable because the pandemic has exposed and attacked many vulnerabilities and blind spots of our previous models and systems. Under damped oscillations will give us more chances of finding and adapting to the new normal(s), instead of returning to the old, getting stuck in some unsustainable states, or inadvertently creating fragile new systems.

Second, while effective vaccines or treatments are still many months or maybe even years away, we cannot afford to wait to till the other measures, mass testing and contact tracing, are all in places either. We already know which demographic groups or persons with pre-existing health or medical conditions are mostly at risks, and the main transmission modes and pathways, we can apply selective control, containment and protection measures, and try to achieve some levels of collective immunity quickly with minimal and controllable risks.

Third, there are other grave risks, including increased mortalities from deeply damaged and contracting economies. We cannot afford to form tunnel visions or more blind spots by focusing only on one major risk. We should use data to calculate such risks and choose acceptable balances. This is the one aspect we will focus on now.

The relationships between various mortalities and states of economic development have been studied and disseminated. From one published study (M Harvey Brenner, “Commentary: Economic growth is the basis of mortality rate decline in the 20th century — experience of the United States 1901–2000”, International Journal of Epidemiology 2005; 34:1214–1221), we estimate that the relationship between age-adjusted death rates and GDP per capita can be approximately described as a power law with an exponent of -2/3 (Fig. 2), i.e. for every 10% drop in GDP per capita, the death rate increases 7.3%; for a 5% drop, a 3.5% increase.

Figure 2. Age-adjusted death rates vs real GDP per capita in the US from 1900 to 2000 (natural logarithms).

Figure 3. History of age-adjusted mortality rates in the US, 1900–2016.

The Conference Board forecast on April 9 that the US economy in 2020 will contract between 3.6% and 7.4%, with a likely scenario of 6.5%. This corresponds to approximately 4.6% increase of death rate. Based on the current mortality of about 9 in 1,000, or 0.9% (Fig. 3), a 4.6% increase corresponds to 41 more deaths in 100,000. This is to be compared to the current total deaths from COVID-19 in the US, about 55,415 out of a population of 328.2 million, or 17 in 100,000.

This shows that prolonged shutdowns and delayed recoveries can potentially have more dire consequences not just in economic damages, but also in more human sufferings and lives lost.

Let’s take another look at Sweden (https://ourworldindata.org/country/sweden). The various death rates from diseases and accidents add up to about 0.84%, 8.4 in 1,000. The four leading causes are alcohol and drug use disorders, cardiovascular disease, cancers and smoking. The current total deaths from COVID-19 are 2,194, with a population of 10.1 million, or 22 in 100,000. Compared to the overall death rate of 0.84%, or 840 in 100,000, the death rate from COVID-19 is still very low. If the less strict measures taken by the Swedish government would lead to higher shares of collective immunity, and not lead to overwhelmed medical systems and drastically increased death rates later, while the economy stays relatively stable and healthy, then this strategy may work out better for Sweden. This provides a valuable reference case for the world.

With a very high standard of living, low population density and exceptionally high percentage of single-person households, however, Sweden is an exception rather than the rule. Whether its strategy will work out better in the end is also uncertain.

For the majority of the world, the dependence of mortality rates on GDP per capita dictates that deeply damaged economies will lead to more deaths. Fig. 4 is an example such dependence (in a log-log graph, the slope, or steepness, of a straight line or a band covering the majority of data points gives the exponent of the power law relationship. In this case, it is approximately -1.4/2, or -0.7, close to the US exponent estimated above).

Figure 4. Child mortality rates vs GDP per capita, 2017.

IMF projected in mid-April that the global economy could contract 3%. With the world death rate at about 0.76%, or 7.6 in 1,000, a 3% drop in GDP per capita could potential result in a 2.05% increase in death rate, or additional 15.6 in 100,000. For a world of 7.8 billion people, the increased deaths could reach over 1.2 million from 3% economic contraction! The current worldwide total deaths from COVID-19 is about 207,000.

It becomes clear through data-based risk calculation and comparison that we cannot let the world’s economy contract too much while we fight the pandemic. Our collective earlier reactions and ongoing responses in the face of rapidly growing threats with unknown limits are mostly correct and generally effective. Now that we are learning more about the risks and ways to fight the pandemic, and have managed to flatten the curves, we need to update and revise our strategies and plans. In particular, we need to think very hard and make difficult decisions to start the recovery processes before more damages to the economies are wrought.

All the above risk calculations, especially with regard to the COVID-19, are based on the extremely difficult but absolutely necessary policies, actions and sacrifices of the governments, institutions, companies and people have carried out and shouldered so far. When we work toward recovery, we all need to be super vigilant and do our utmost to prevent more outbreaks or rebound, and protect those at risk. Recovery is by no means returning to business as usual in the past.

The road to recovery is tough, and balancing the risks is hard, but we cannot afford to wait it out. We have to face the challenges together and do the right thing now.