Updated Covid19 Tracking

Going over projections I made at this time last month, and at this time two months ago, it's clear that some variables turned out to be much better than projected, and some were much better.

Two months ago, it looked like the US death rate would be somewhere around 1-2%, and that was staggeringly lower than where we sits today at over 5% for the US. New York and New Jersey are now higher than 6%. I incorrectly assumed that the death rate would get better as we implemented new technologies and found existing medications (already approved by FDA for other uses) to be helpful in the fight against Covid19 such as Hydroxychloroquin.

When considering the US mortality rate of Covid19, the cumulative number doesn't tell you that much. The daily number will show more precise changes especially in the context of a possible rise or fall due to drugs like Hydroxychloroquin or government policy changes. The chart to the right shows Daily Dead compared to Daily New Positive Cases. Daily Dead clearly lags a few days behind Daily New Positives in the initial ramp up, as expected. The correlation is evident, as expected, but I think this chart shows a lot more than that. The spikes from mid to late April don't lag behind spikes in new infections, but correlate in real time. One possible explanation is that as people are coming into the hospitals, and overwhelming the systems and staff, there are fewer hands on deck to deal with critical patients. Monday's number was the first time we've had fewer than 1000 deaths in the US since April 1st, and given where we were 10-14 days ago, deaths shouldn't be leading the way down. There should be a lag as most people don't just die on the same day they are admitted. We hit our peak for new infections 10 days ago with over 40,000 new cases. Even 4-6 days ago we were still close to our peak with over 30,000 per day. So, the fact that we're dropping so quickly in daily deaths tells me that a change in treatment may be the cause. Indeed, the FDA issued a statement 10 days ago, "Hydroxychloroquine and chloroquine have not been shown to be safe and effective for treating or preventing COVID-19."

Another somewhat incorrect assumption I made was that 14 days after the infection diagnosis, a patient would be cleared, and no longer infectious. Unfortunately, some people (hopefully in extremely rare cases) are still testing positive up to 70 days later! In the second chart, on the right, I compare new cases to a calculated "recovered" group which is basically the new infections number on day X minus dead on day X +14. The problem is, 14 days was just a recommended estimate, and the WHO is investigating reports that patients could be testing positive a second time. In other words, they're concerned that a patient could recover after the 14 days, tests negative, then the antibodies built up in the immune system potentially aren't sufficient to fight off a newer strain of the virus. That makes modeling very difficult, and may be why the cumulative number of positive test results from 14 days ago, over 700,000, doesn't match up with the number of recovered cases being reported, 180,000. Who knows what that real number should be? Once we're "recovered", we can't just be thrown in a group assuming we're permanently recovered. That hasn't been proven correct, at least not yet, and may be proven incorrect at some point.

I also incorrectly assumed that testing would be dramatically higher than it is today, and I can't even begin to comprehend how we've absolutely failed in this regard. The chart to right shows new tests compared to new daily positive cases, and clearly shows a high correlation. The more we test, the higher the positive outcomes. Well, it's our job to expand testing so that correlation no longer holds. There shouldn't be excess positives that we're turning away from getting tested. We need a more proactive approach. The good news is that although the small increase in testing still shows a high correlation, see April 25th and May 1st the two highest new positive days matches with testing spikes, the trend for new positives is decreasing while the trend for new tests is increasing.

Despite the shortcomings of the "recovered" calculation, it is still the best rough estimate calculation I can come up with, and provides more detail as we look at individual states.

Here we see that New Positive Cases have peaked and are strongly trending downward in New York. New Positive Cases are not influenced by "recovered" in any way. These are the raw data numbers reported by the state, and collected by www.CovidTracking.com. As long as the Daily Recovered is higher than New Positives, we'll have a net decline in the number of Active Cases, which is a calculation I've made, and is also reduced by the number of dead.

New York is by far the highest number of Cumulative Positive Infections (318,953), followed by New Jersey (128,269), Massachusetts (69,087), and Illinois (63,840), so it's great to see New York's new positives drop to just 2534 for yesterday when they've had eight days above 10,000 new positives. New Jersey and Massachusetts have had much smaller new positives as well with 1525 and 1000 respectively for May4th.

Illinois, however, has not peeked. In fact, as we seemed to flatten off in new positives from April 16th to 22nd, we've since accelerated higher, and the new infections 2,341 on May 4th, is over twice the amount of calculated "recovered", 1103.

When considering what will happen when we loosen restrictions on social distancing, and expand on businesses opening back up, I'm not convinced we're past the worst of Covid19. Looking at the chart below, the level of active cases in the US has plateaued over the last 25 days and that's considering our very best efforts to isolate and snuff out this virus. Maybe the warmer temperatures will help as summer comes around the corner, but my money is on a renewed spike in cases and a bigger one than the first spike given that we're starting from a much worse position. We've had roughly 1.2M Americans test positive, and there are about another 330M that could potentially contract the virus.