I traveled again this week, brave soul that I am, went to a business meeting. Given the current ventilation specs an airplane may be the safest place to be these days; they practically suck the air right out of your lungs. In any event, gave me a chance to ruminate on the epidemic, what we might have learned, what we are seeing now in the case surges in Europe and North America (why not Asia–did they have a much bigger epidemic last spring than anyone realized? Are they ignoring it now? Is there something meteorologically different there? It isn’t masks, etc., look at the pictures coming out of Chinese night clubs.)
Another big case dump today in Minnesota, but again, look at the case table by specimen date, pretty interesting pattern to follow. And more deaths to help us climb up to Sweden’s rate. Some European countries are already seeming to plateau or slow in their case growth rate. Everywhere the rate of deaths is much lower than in the spring, so people wonder why. I will go back to what I have been saying since the spring. This epidemic is front-loaded. Meaning that if you imagine the virus as an agent sampling the population with the purpose of killing those it samples, it wasn’t doing a random sample. It was preferentially finding and infecting the most vulnerable and it was killing many of them. It still is more likely to cause serious disease when it reaches the vulnerable, but frankly there are less of them. We have killed 2% of our LTC population in Minnesota. So while hospitalization rates are somewhat similar, death rates are lower. Better treatment accounts for some of this as well.
You may recall that in the spring and into the summer, the CDC, and the Minnesota Department of Health, were estimating that only around 10% of infections were being detected. This was based on antibody prevalence surveys and the knowledge that most infections were asymptomatic or mild, so most people didn’t even think they were infected with anything and most didn’t seek medical care. At that point there was very limited testing. The infamous Minnesota Model (still waiting for that update) made a similar assumption. If that were true throughout the epidemic, we would have 1,650,000 Minnesotans who had been infected, or about 30%. In anyone’s epidemic model, and especially in ones that take heterogeneity in contacts, infectiousness and susceptibility into account, that is enough to dramatically slow transmission. If the virus walks into a bar with ten people, three aren’t targets, and more realistically, only 5 out of 10 people are actually in the bar.
I don’t believe we are detecting only 10% of cases under the current testing regime. I have begun looking at some somewhat complicated math to equalize testing and detection rates, by assuming that as testing ramped up, detection improved. Someone may be able to figure out some better formulas than I have, and I will show my work shortly as part of my exercise in normalizing cases to testing. But I would guess, and it is a guess, that now we are probably catching at least 25% of infections. That assumes that people with symptoms are getting tested for sure (not a rigorous assumption) and that contact tracing, testing before health care appointments and mandated work testing are picking up a lot of asymptomatic or presymptomatic cases. God knows we are testing enough. If that is true we have had about 650,000 infections in Minnesota or around 10% of the population. Still enough to substantially slow transmission in a heterogenous population, largely because it is a very good assumption that those with the most contacts are the most likely to get infected first, so you are removing a large number of total contacts from the susceptible pool.
So again, the current features of the epidemic, with summary thoughts:
Testing–way too much, likely lots of false and low positives. Ineffective to guide contact tracing efforts.
Cases–a clear swell in cases, driven a little by testing, but more by meteorological factors–less sunlight, lower temperature, lower humidity–which apparently are more hospitable to CV-19; by human physiology–lower vitamin D levels, perhaps other biological changes; and by human behavioral changes–we are indoors more and we are at home more, both of which are associated with more transmission.
Hospitalizations–also clearly up, length of stay appears to be down, almost certainly driven by remdesivir administration and shorter observation stays to monitor oxygen levels. What do total hospital days look like? Rates per cases do not appear up.
Deaths–also up, rates per cases definitely not up. Still occurring largely among LTC residents, although a lower percent than in the Spring, partly due to so many already having been killed and partly to avoidance of being in an LTC facility. Occurring at as high or higher rate among the elderly as in the Spring. Community dwelling population rates of serious disease and death are low.
I will go out on the limb again today, which I really try to avoid doing, because there is so much uncertainty about this virus, and say that I believe we will see the case peak in this swell in a week or two, then with the usual pattern, cases will fall off first, followed shortly by hospitalizations, with deaths tailing off last. I could be very wrong, perhaps we will see some astounding surge all winter long, but I don’t think so. Look at NYC and New York as a whole. They have much lower current cases number than we do on a much higher population base, but they also had antibody prevalence rates as high as 25% or more. Current transmission rates appear strongly related to depth of epidemic in the Spring.