HARRISBURG, Pa. (WHTM) The fall resurgence of COVID-19 was widely expected. The extent of it, as measured by CDC forecasts as recent as a month ago after the surge, as its also known, began was not.

An ABC27 analysis found the CDC, in early November looking ahead then toward early December underestimated the now-known actual death counts by two-thirds, in terms of what the CDC considered a likely scenario.

Even its near-worst-case scenario (technically the upper end of its 95-percent confidence interval) was less than two-thirds of reality. The CDCs forecasts, in turn, rely on the work of 36 other organizations from Ivy League universities to companies like Microsoft that meet the CDCs data quality requirements.

As of Nov. 2, 2020, the CDCs midpoint expectation for weekly deaths now was about 225. It was 95 percent confident weekly deaths wouldnt exceed about 375. The real number, counting the 194 new deaths reported for Tuesday alone? 661.

Why were the forecasts off by so much? Possibly because the biggest variable of all isnt molecular.

In essence, models that do this kind of prediction, you have to predict human behavior, which is hard, said Dr. Mark Roberts, a professor of health policy and management at the University of Pittsburghs Graduate School of Public Health and also director of the universitys Public Health Dynamics Laboratory.

And human behavior, Roberts said, has been unhelpful.

The cases and the spread is because were, you know, COVID-tired. Were not social distancing. Were losing the intensity with which were supposed to socially distance, Roberts said.

He said this tool, developed by Massachusetts General Hospital at Harvard Medical School, is useful for gauging different pandemic outcomes based on different inputs public policy, human behavior and so forth and would have done a good job predicting the current state of affairs, given the correct assumptions about behaviors such as Thanksgiving travel.

Asked Tuesday whether the degree of the surge has caught the PA Department of Health by surprise, Michael Huff, the departments director of testing and contact tracing, said were not surprised by anything we see anymore but agreed with the idea that the surge has been supercharged by conducive behavior, including Thanksgiving gatherings.

Ominously, considering the current numbers, he said we wont see those cases for another week, maybe 10 days.

Hospitalizations and deaths, in turn, typically lag diagnoses by another week or two. This is a very dangerous time, Huff added.

Here are Pennsylvania COVID-19 deaths per day for the most recent seven days available:

Day-to-day variance occurs partly because not all deaths are reported the day they occur. Epidemiologists consider seven-day moving averages, on the other hand, to be more stable representations of reality.

One major difference between the current surge and the first surge, in March and April, is that the majority of deaths during the first surge occurred in long-term care facilities.

Indeed, despite a recent reversal of this trend, fully 64-percent of COVID-19 deaths in Pennsylvania 6,885 out of 10,757 deaths since the pandemic began have occurred in what the state classifies as nursing and personal care facilities.

But now? Of the 180 deaths reported Monday, just 30 or 17 percent occurred in such facilities.

Although a Department of Health spokeswoman couldnt confirm this, the ABC27 analysis found a high probability that the state is now setting new COVID-era highs for daily deaths due to community spread outside long-term care facilities.

View original post here:
ABC27 Exclusive: Current COVID-19 deaths in Pennsylvania more than triple what CDC expected a month ago - ABC27