Exponential growth is the world of epidemiologists. And financial analysts too.
This writer, a retired analyst and not an epidemiologist, has been tallying the daily stream of case numbers in both the US and Japan. Here are some observations.
- During January and February cases grew faster in Japan then in the US.
- That was when many cases flew below the radar in the US only to surface in March.
- March was a game changer: cases decelerated in Japan while they rocketed in the US.
- The US was invaded from multiple places while Japan enjoyed the island effect.
- From March to mid-April, the trend reversed again; numbers grew faster in Japan.
- And on April 29, after 22 days of semi-lockdown in Japan, the rates are the same.
Metering a rocket
So how do you examine a daily stream of somewhat random numbers increasing exponentially?
You take the new case count and calculate the most recent seven day average. You do that every day and it is called taking a seven-day-moving average (7DMA). It is also called smoothing the numbers or canceling the weekend effect.
The 7DMA number of US cases increased from 3,152 on Mar 17 to 24,141 on Mar 24 based on Johns Hopkins figures. That is up a diabolical 666% in seven days.
How about if we took the compound daily growth rate [ (X/Y)^(1/7)-1 ] where X is today’s number and Y is a week earlier. Now the number is a manageable 33.8% CDGR. The chart below shows the CDGR for Japan in red, based on health ministry statistics, and the US in blue.
Until March both the number of confirmed cases and the rate of growth were higher in Japan than the US. Then what could account for the sharp acceleration in the US in the first two weeks of March?
A huge inflow of cases from overseas? This was a period between the restriction of travelers from China, which went into effect on Feb 1, and the discovery of the first case in New York City, on Mar 3. Subsequent reports suggest that the stream of travelers from China continued.
Or could the acceleration be explained by a number of cases that flew below the radar, coming to surface on the West Coast. Recent autopsy studies in California and other western states suggest that there were more clusters on the west coast than previously reported.
Counting stealth clusters
Then how many cases escaped detection in January and February?
If we assume that the rates of growth reported in Japan are a “fair approximation” of how the rates of growth should have behaved on the US West Coast, we can work backwards from March numbers and estimate the hidden cases of February.
This exercise suggests that there may have been 380 to 450 cases on the West Coast rather than the 65 cases reported on February 28. That is twice the 210 confirmed in Japan on that date.
The chart below assumes these new estimates for the US. We are not claiming that this chart shows the “real picture.” Rather we are using these charts to raise questions to better understand this pandemic.
What accounts for the huge acceleration in US cases during the ten days between March 12 and March 22? The first case in New York City was reported on March 3. There may have been more but a scattering of cases cannot explain such an explosion in cases.
Stampede before lockdown
Restrictions on travel to the US from the 26 Schengen countries in Europe, which normally adhere to a single visa arrangement and keep their mutual borders open, went into effect on March 14, and travel to the US from the UK and Ireland was restricted from Mar 17. Flight data suggest that one million people traveled from Europe to the US during the 45 days before Mar 17.
The US may have “closed the door” on China on Feb 1, but left the door to Europe wide open until Mar 17. Most of the airport screening during that period was focused on flights from the East.
Tens of thousands of people packed the flights before March 14 and March 17 to beat the announced lockdown. Could this March explosion have been prevented if no travel restriction was announced?
Island effect
Meanwhile Japan enjoyed a deceleration in its compound daily growth rate from 15% in late February to 5% in late March.
South Korea, a virtual island nation, achieved similar deceleration through massive testing, Taiwan, New Zealand and Japan achieved it with meticulous airport screening and with tracing and isolation, although Japan’s isolation effort was later proven lax.
On March 7, four days after the first case in New York City, city health officials proposed a “sentinel surveillance team” to trace infections, but that was turned down by the mayor. On March 23, at the peak of the growth curve, the mayor finally agreed.
The sentinel surveillance team has been an integral part of Japan’s infectious disease countermeasure since 1948. Taiwan, Korea and Hong Kong have similar teams, many of them veterans of the SARS and MERS epidemics.
Ten days in March made all the difference for the US East Coast.
Never be complacent
Japan’s CDGR began to blink red in late March. It rose above 10% even while the US CDGR was slowing thanks to shutdowns and social distancing.
Japan’s positivity rate – the number of positive cases divided by the number of tests – also began to rise. The rate remained steady between 5% and 6% during March but began to rise toward 9% by April 22.
What was happening? Japan had been too lax about the catastrophic contagiousness of the virus. Isolation measures were inadequate compared with measures taken in Korea or Taiwan. Number of cases untraceable to known clusters was increasing and community spread was widening.
Prime Minister Abe had no choice but to declare a partial lockdown on April 7. Results are beginning to show and the compound daily growth rates for both Japan and the US are now 3.9%.