Hacker News Reader

Comment by CommieBobDole | Hacker News Reader

The Georgia death rate graph that you link to is sort of an odd case; there's apparently a variable lag between the occurrence and reporting of a death, sometimes up to a month - When a death is reported, they place it on its occurrence date on the graph, not on the reported date.

As a result, it will appear from this graph that deaths are always declining whether they are or not - the more recent deaths have not yet been reported, so they're not anywhere on the graph. They do the same thing with their daily cases map, but they're a little more transparent there about noting that numbers for the past few weeks are basically meaningless.

Reported deaths in GA have been fairly flat, but hospitalizations are rising slowly:

https://www.ajc.com/news/coronavirus-georgia-covid-dashboard...

View on HN

Another thing to keep in mind about hospitalizations is that many people have avoided going to hospitals during the pandemic for non-covid related visits for fear of contracting covid at the hospital and/or adding to hospitals caseload. It could also be the case that people are re-evaluating their risk profiles and no longer taking the same precautions with respect to hospital visits. I'm not sure what is really happening but wanted to emphasize how tricky it can be to interpret covid related metrics.

>As a result, it will appear from this graph that deaths are always declining whether they are or not

Most deaths are reported relatively quickly. If there were, as an example, a 2x increase in deaths it would show on the graph within a few days. It would show as something like a 1.8x increase before trickling up to the full 2x over the following weeks.

Death lags new case by as much as a month (due to time from infection to death). So we won't really know if there is an increase an death due to the new cases for a number of weeks after we see a rise in cases.

The death rate absolutely does respond more quickly than a month, and in a time series with this distribution the tail of the data is not the most important factor.

We should be seeing increased deaths from this in places that have reopened weeks ago.

The problem is, we accelerated deaths drastically due to insane public policies like housing the sick in close quarters with the elderly. In engineering, overengineering like this wastes time. In public health, it kills people.

My expectation is that we'll see an increase in infection rate, a decrease in new deaths per day, and we'll have spikes associated with pockets of the elderly who were previously not impacted being exposed.

Yeah, but the problem is the combination of these points:

* A certain percentage of infections lead to death

* Death usually only occurs one or two weeks in

* Pandemics follow a logistic curve in their spread (and this curve has an exponential part)

* Most infectious transmissions happen before one gets symptoms (otherwise staying at home once you cough would solve everything)

This is why a sudden rise in deaths would indicate that a few weeks ago something went terribly wrong. This also means that a decline in deaths does not mean you are doing just fine right now — if anything, it means you did fine 1-2 weeks ago.

And probably more like 2 to 3 weeks ago.

Reply on news.ycombinator.com