Ubiquitous Artifactual Periodicity in Covid-19 Data Sets: Extracting the Artificial Teeth of the Pandemic

Tejal Schwartz

Accurate epidemiological data are crucial for modeling pandemics, and for designing and instituting public health measures to minimize spread and societal impact. Data for Covid-19 incidence, hospitalizations, and death, as presented in the most widely viewed publicly accessible sources, exhibit a remarkably constant pattern of recurring peaks and troughs with a periodicity averaging 7 days. This tightly constrained interval is seen across state, national, and international data sets, and does not conform to known or modeled super-spreader event bursts of transmission. The 2-14 day asymptomatic SARS-CoV2 incubation period yields localized outbreaks spanning weeks or months. I confirmed the non-random nature of the data periodicity, identified fixed temporal shifts in data sets when reported by secondary vs. primary sources, and demonstrated highly significant correlations between troughs and weekends for primary source data. In total, these findings support my hypothesis that the observed recurring 7 day interval between troughs is an artifact caused by deficits in testing and/or recording of data on weekends at primary sites of data generation (labs and hospitals). Additional weekend deficits in transmission of data between primary, secondary, and tertiary data aggregators may contribute to the artifactual patterns observed in almost all graphically presented data sets. Given the enormous political, economic, and public health ramifications of these data, and the unprecedented erosion of public confidence in current official sources issuing data and recommendations, it is important to minimize unintended, artifactual errors in data presentation and interpretation, and to clarify the reason for weekly oscillations in Covid-19 events.