Modeling even simple aspects of the COVID-19 pandemic is more than challenging in the United States because of the sparsity of data.There is no comprehensive testing, and an active debate is currently taking place regarding the accuracy of the limited testing we have. Adding to the uncertainty: we lack information on the total number of people infected; we do not know whether those who have been infected can be re-infected; we don’t know how many are infected but asymptomatic; and we lack information on when, and for how long, those infected may be contagious. In short, this is a forecaster’s nightmare.
The issue with a lack of data is further compounded because most, if not all, epidemiological models are complex. They are designed to provide a lot of needed information in the face of a pandemic. And even though at the national level some data may be available, applying a typical epidemiological model to a county is virtually impossible without a very heavy assumption burden.
The driving force for my Whatcom County forecasting was the fact that very little was known about a possible peak during the initial surge. Epidemiological models have been noticeably absent, not only in Whatcom County, but virtually every other county that was not in a heavily populated urban area (although models for many urban areas have been absent as well). All these issues leave local officials and residents literally in the dark when it comes to trying to get a picture of what may be coming and how to prepare for it. Unfortunately, it is in these small, local areas that the battles are being fought.
That said, the table to left shows the three-parameter logistic model has done a reasonable job of identifying the path of the pandemic in Whatcom County from April 16th to the 30th. (Review earlier models here, here, here and here) All the reported cases fell within the 95% prediction intervals during this time. The table shows that from the 16th to the 30th, the model erred, on average, by only 1.31%, and the model underestimated, on average, the number of cases by less than 1 percent, -0.88%
I didn’t ever intend to provide post-peak forecasts for two reasons: (1) it was expected the epidemiological and other complicated models should have had sufficient data by the time the initial surge peaked; and (2) the urgency is not on the downhill slope, it is on the uphill slope. But a word of caution here is appropriate. If , moving forward, the daily number of new cases stays above zero for a long period of time, occasionally punctuated by spikes due to data lags and dumps or, worse, because hotspots break out, COVID-19 will have evolved from being an acute condition to being a chronic condition.
Moreover, this chronic condition is likely to be common throughout the country, brought about because our widely varying containment policies and their efficacy will affect different areas at different times. If the history of the 1918-20 Spanish Flu world pandemic is any guide, new surges of COVID-19 are coming for many of the same reasons they occurred in 1918-20, lack of both information and comprehensive, universal containment measures.
All of these issues are aggravated even further by: (1) the absence of uniform, comprehensive, containment measures for the U.S. (2) the length of time containment measures are/have been in effect; (3) the lack of federal support for states; and (4) the fate of those without incomes due to containment measures, and how long they can hold out without any source of support. Due to the toxic national political environment, and fueled by misinformation, propaganda, and outright lies, those of us who live to see 2023 will bear witness to one of the most turbulent times in U.S. history.
I will leave the tracking and monitoring of these likely events to those with the experience needed to run epidemiological models. Hopefully, the data needed to do this will finally come online, including universal test results, accurate reporting, and the correct coding of COVID-19 deaths. However, it is equally likely that while states, counties, and cities struggle to make this happen, we, as a nation, will struggle with a steady drumbeat of political messaging telling us that testing is not needed, reports are perfect, and there are no COVID-19 deaths because the novel corona virus has been banished from the land.