A commenter asks where I think we will be one month from now. I think we will be in the dark, or at least I will be.
Recall that I prefer to track death rates. What I finally settled on as an indicator was the 7-day average death rate, using this data. This average held at around 1500 through the dark days of April and the first third of May, and then it finally began to trend lower after that, down to less than 600 in the third week in June. Then on one day, June 25, there were 2500 deaths, sending the seven-day average to over 800.
It is not that 2500 people died of the virus on the 25th. But that was the day that New Jersey included over 2000 previously unreported deaths from April and May. I am afraid that we have not seen the last of this “death harvesting.” I came across an article the other day, which I forgot to bookmark, that said that analysts are suspicious about the high rate of deaths reportedly caused by Alzheimer’s this year. So perhaps 10,000 more deaths will be re-stated as Covid deaths.
The data I wish I had is data on serious cases. I would define a serious case as a case that deprives the victim of normal activities for more than 30 days. Death obviously counts. Quibble with my definition all you want, but my point is that in order to calibrate my fear of the virus, I would like to know the prevalence of serious cases.
Anecdotal information: both of my direct reports had it within the past two weeks One in his 20s and another in his 50s with a compromised immune system. Both had extremely mild cases and did not require hospitalization. All of the other cases in my company have also been mild.
If this anecdotal information scales (which I have no clue about one way or the other) it indicates that either the virus has 1) become less lethal or 2) those most vulnerable made up a disproportionate share of the early cases.
I’m optimistic about the next 2-4 weeks, but I’ve got nothing to base this on other than these anecdotes. I’m flying blind again.
On the downside, I’m hearing anecdotally (but from good sources within Big Pharma) that the antibodies are only effective for 3-12 months. Hopefully this is proven incorrect or that we have a good vaccine soon (which may need to be administered every 6 months or so).
More anecdotal information — AKA data points: I’m aware of three mild cases in a high-rise living community. (Intense precautions applied by building management and complemented by most residents in the wearing of masks and gloves and social distancing. No deaths.) All three recovered, but have lost their senses of taste and smell. Separately, I know a person who lost her senses of taste and smell nearly two decades ago due to a stroke. She lives well and still is a great cook, but she can’t enjoy the aroma of her cooking nor the taste of food she prepares nor that prepared by anyone else.
Is “hospitalizations” a good proxy for “serious cases”?
I’ve heard from some people that if they are young and feel sick with COVID symptoms they may go into the hospital, but don’t stick around that long.
I’ve been looking at the data on the Worldometer website. They have 128800 total deaths to date vs. 119000 deaths to date on the COVID Tracking website. Not sure why there is such a difference.
Looking at the daily death (DD) data, I find that DD has been exponentially declining since about April 15. Doing a least squares fit I found the equation for the decline. I integrated this out to infinity and added in the total number of deaths up to April 15. Based on this I estimate that the total number of deaths to end up about 160000 for the USA (if this trend continues)
Over the last 10 days or so there has been a sharp increase in the number if daily cases. Though to date the DD continues to follow the exponential decay. Not sure how to explain this though it may be that we need only wait another week or so and then we will see an increase in DD.
The death rate is very sensitive to age/health.
If the group of infected for the June wave skews young & healthy (which is what I’m hearing and which seems consistent with the demographics of the recent protests*), then the death rate should be a great deal lower for this wave even if the overall rate of infection is somewhat higher.
*Some people have claimed that the protests weren’t a meaningful contributor to the recent spike in cases. I refuse to believe that large groups of people in close quarters yelling and screaming for an extended period isn’t a serious risk factor for COVID transmission. The timing of the change in case trend seems very consistent with the protests.
Inside vs outside makes all the difference
It makes a good deal of difference, but it’s not invincibility.
We have been far too cautious about things like opening up playgrounds and other outdoor activities and far to cavalier about things like mass protests.
The lower number is confirmed deaths, the higher number is confirmed deaths along with likely deaths.
Thanks
I keep a general eye on US and world numbers, but I have been most closely following my own metro area data, along with the data from the 3 other large metro areas in my state. I have spent enough time with these datasets that I have a feel for how they are presented, what’s missing, handling of mass testing, periodicity effects, etc. These patterns are so variable, even between the large cities in my state, that know I do not have the time to competently interpret datasets outside my immediate focus areas. Given the liberties I see taken daily in interpretations of the datasets with which I am familiar, I can only lament.
Not only was the regulatory state unprepared to control a widespread virus, but it was unprepared to count it.
Counting is hard. Much tougher than most people realize.
Just looking at Maryland state reported data, the intensive care hospitalization figures seem like they would be useful for calibrating fear levels. It’s bad enough that there are still 160 people in intensive care today. That is enough information for me to calibrate that minimizing public contact is probably worth the opportunity cost.
https://thedailyrecord.com/maryland-covid-19-chart/
How many of my activities need to be eliminated until risk is observably small. Adding one more activity noticeably raises my risk from covid.
You can back out the death harvesting by simply subtracting it from the new deaths totals, but keeping in the cumulative deaths. It keeps the harvesting from mucking up the concurrent 7 day average of new daily deaths, which is what you want to look at to determine trends.
I agree wholeheartedly with the hole in the data. There is a huge gulf between cases and deaths. Dividing non lethal cases between two categories or among three would be extremely helpful in risk assessment. Even more helpful would be by age bracket.
The data is dirty, so you can make any prediction you want, then then explain your missed calls away by blaming or reclassifying the data.
Case numbers are garbage, because the false positive and negatives are really high when asymptomatic people get tested, and because no one seems to care about correcting for increased testing. Not sure why we can’t get a graph of cases normalized for testing anywhere in the mainstream media. Seems simple.
Hospitalization numbers are garbage, because we don’t know how to classify what counts as a covid hospitalization. Some states are now counting anyone who is admitted for any reason and who happens to have covid as a covid hospitalization, and they change classification methods on a dime, so you can’t compare to past data. Filtering for the “serious” hospitalizations is a fool’s errand, when we can’t even agree on what counts as a Covid hospitalization.
Death numbers seem like the best data, but they’re messy too. When I was an intern, it was my job as the lowest man on the totem pole to fill out the death note and certificate. This was the lowest item on my to-do list, and I only got around to it when the coordinators would page me on my way out after being on call to get it done please. What kind of care goes into figuring out the cause of death in a 90-year-old who died from being 90. Was it the stroke? Was the stroke caused or exacerbated by covid? Was covid just another item on the list, along with renal failure, diabetes, etc?
Excess death calculations are garbage, because people aren’t just dying of covid compared to prior years. People are offing themselves or dying because of delayed care for their cancers or other chronic issues.
There seems to be a group of people with Asperger’s who honestly believe that if you look at enough garbage from enough different places, the sheer volume of garbage will let you average out the garbage and yield golden data. Golden data that will allow them to explain the past and make predictions about the future. These data prophets would be cute if so many innumerates weren’t relying on them to help guide policy.
Here’s the fact. X% of the population will need to get the virus before we get through this. Y% of those will die. Most of the dead will be old people. X and Y are unknowable. We can’t even estimate them accurately. X is anywhere between 20% – 60%. Y is anywhere between 0.01% to 0.1%, depending on deaths are being counted.
Your last paragraphe on X and Y seems reasonable to me (even it it sad for Y, their friends and family).
But is it not something that could have been said back in January? Has there been anything new under the sun and how is it different that any other pandemic of the past (or the future)?
By the way, how is it there is still seemingly absolutely no knowledge on how this virus spreads after 6 months with various countries adopting completely different approaches with seemingly random results?