COVID-19 Projections For USA Are Likely Highly Underestimated

This is a brief glance to the new projections IHME provides currently into August, 2020. Current data shows that daily number of new cases is growing linearly (see Fig. 1)

Fig. 1. Total number of confirmed COVID-19 cases. Total USA cases grow linearly, but NY State cases are slowing down. The total number for the rest of the country is also growing linearly.

The alarming point is that, in a big picture, only New York state cases are slowing down, the number of cases in the rest of the country is still growing.

Looking at the daily stats (Fig. 2) one can see that total numbers in average stay the same for a month-long period since 4/7/2020. If we exclude the NY state cases, the number of cases in the rest of the country was quickly growing until the middle of April. Since then, it has slowed down the growth, but still there are no signs of decrease. Only NY state shows the expected slow down, but still

Fig. 2. Daily USA cases. NY state shows slow down, the rest of the country is growing.

The mortality rates show similar behavior (Fig 3, 4). The daily numbers were calculated from the totals.

Total (cumulative) USA mortality. NY state numbers slow down, but the rest of the country numbers grow linearly.
Total (cumulative) USA mortality. NY state numbers slow down, but the rest of the country numbers grow linearly.
Fig. 3. Total (cumulative) USA mortality. Cumulative curves are expected to show a tendency to approach a horizontal line, but only NY State probably fits that behavior. The rest of the country shows linear growth.
Daily COVID-19 USA mortality rates show no slowing down for a month now
Daily COVID-19 USA mortality rates show no slowing down for a month now
Fig. 4. Daily USA mortality. Totals are not changing from 4/7/2020. NY state has passed the peak, but still has about 400 cases daily in average. Note that fluctuations are caused by lags in the reports or other reasons, in one case by obvious mistake in the reported numbers on 4/19/2020, when the totals were less than on the previous day.

The revised on May the 4th IHME model has predicted 134,000 fatalities by August 4th, a significant increase due to many new factors taken into account, as well as the impact of the reopening the states.

The model expects that daily mortality numbers go down to about 100, but this is not a realistic expectation, as we see now. Current daily mortality is around 2000, and with reopening the states, we can only expect this number to grow.

We can make a simple estimate of a minimal number of fatalities during the 90-day period from May 4 till August 4. Suppose that reopening does not cause the increase in daily cases and fatalities. Multiplying 2000 daily fatalities by 90 days, we get 180,000 new fatalities during that period. Adding that to 69,000 fatalities on May 4th, we get the minimal number equal to 249,000 on August 4. That’s almost twice as the IHME prediction.

So, it seems that IHME model has some input parameters that do not describe correctly the current COVID-19 spread behavior, probably too optimistic estimates of current impact of the restrictions on mobility.

It’s worth noting that the original 134,000 fatalities estimate has been changed by IHME on May 10, now predicting 137,184 COVID-19 deaths, based on the projections that took into account the data from the beginning of the outbreak until May 7. This correction most probably was needed because the projected curves predict steep decrease up to 70 fatalities a day. I will not be surprised if the subsequent corrections to the model will increase the predicted results every time.

In conclusion, it seems that incorrect assumptions/input values somewhere deep in the IHME model significantly downplay the predicted death toll , even if the spread continues in a linear fashion. If the states’ lifted restrictions trigger exponential growth again, as it was in the beginning of the spread, the fatalities will start multiplying unpredictably.