Governor Cooper claims that “North Carolina is relying on data and science to protect our health and safety.”
I have my doubts.
Since the beginning of the COVID-19 pandemic, a drumbeat of “experts” peddling flawed models predicted a catastrophic death toll. Some of these statistical models even reached a level of notoriety, such as the Imperial College model, before “amateurs” exposed the many defects in what is best described as statistical alchemy. After months of relying on these models to guide policy, New York Governor Andrew Cuomo and others acknowledged that all of the national experts and their projection models were unequivocally wrong.
Statistical modeling, secrecy, and the Cooper administration
North Carolina Department of Health and Human Services (DHSS) Secretary Dr. Mandy Cohen jettisoned available models and relied on one developed by a consortium of North Carolina academics and researchers to guide the Cooper administration coronavirus response policy. That model predicted 250,000 more infections by the end of May, even with continued social distancing and stay-at-home restrictions. North Carolina has approximately 24,000 infections confirmed by lab tests now. Recognizing that asymptomatic individuals have not been tested, the model appears to have been unequivocally wrong. Secretary Cohen is no longer relying on that model to inform the reopening policy.
If the models are not to be trusted, what is Governor Cooper referring to when he says he will rely on “data” and “science” to move forward? Data from DHHS have been notoriously difficult to obtain. The General Assembly required the agency to reveal certain basic data in exchange for supplemental funding, and state and national media outlets have filed a lawsuit this week to obtain public records that the Cooper administration has been unwilling to provide. Unfortunately, Secretary Cohen appears to believe that only DHHS should collect, analyze, and report selected data.
Better measures of COVID-19 trends
Recall the fact that restrictions on our movements were only intended to prevent overwhelming our health care system. The goal of “flattening the curve” was never to prevent infections. The most serious indicator of a pandemic is, of course, the number of deaths. Identifying the precise cause of death can be difficult, leading the Centers for Disease Control (CDC) to use the concept of “excess deaths” to avoid such ambiguities. Once again, Secretary Cohen has withheld the data necessary for such a calculation.
As a result, we are forced to accept DHHS estimates for deaths from COVID-19. Deaths are a lagging indicator of the progression of coronavirus because the time from infection to death can be two to four weeks. Nevertheless, simply looking at a 7-day rolling average (to smooth out reporting inconsistencies) of daily reported deaths is revealing.
The data appear to show a rapid rise, a leveling, and then a decline in the deaths from COVID-19 in North Carolina. In fact, the state appears to have turned the corner somewhere around May 6. Again, this is a lagging indicator.
After the General Assembly forced Secretary Cohen to provide recovery data, she agreed that the CDC’s 14-day period is a reasonable estimate for the recovery of patients that do not require hospitalization. Using that simple estimate, the number of current active cases can be estimated by subtracting new positive cases each day from the cases that were reported 14 days before. This ignores the number of hospitalizations, but that is, thankfully a small and essentially constant number that does not change the result. Doing this yields the following:
This is not a complete picture, however, because the number of tests performed in North Carolina has risen dramatically. The more people that are tested, the more positives will be found, even if the percentage does not change. To correct for this, the same 14-day calculation was made, but this time dividing by the total tests reported for each day. This serves to correct for days when a lot of tests were taken. In addition, a 7-day rolling average was used to smooth out reporting irregularities.
For comparison, the same metric was calculated for New York. The results show what you might expect. After the initial increase in infections, a negative value shows that more people are recovering than are being newly infected. Eventually, the metric should return to zero. This metric would indicate that North Carolina turned the corner around April 28 or so.
Both of these metrics indicate that North Carolina surpassed the threat of increased infections roughly between the last week in April and the first week in May. Hence, the threat of overwhelming our health care system passed many weeks ago. The threat of economic hardship, however, is still ahead of us.