Hunting the tiger - what can be learned from public surveillance data on
the population risk caused by SARS-COV-2?
Abstract
There is a large reservoir of publicly available data that could be used
to better understand the population risk caused by the novel corona
virus SARS-COV 2. However, important questions subject to public debate
have not been sufficiently addressed with empirical data so far.
Based on data published by official sources covering the period from
March 02 to May 31, the impact of general testing activity in Germany on
the number of confirmed cases was investigated with a linear regression
model. The model yielded an adjusted R square of .07, which was
statistically significant but numerically too small to explain a
substantial part of the variation observed.
For the same period, the relationship between changes in public mobility
and the number of confirmed cases was analyzed. A strong correlation
(-.51) was found for mobility and confirmed cases on the same day, which
decreased with an increasing time lag. The correlation was stronger
(-.68) when the date of reporting was used as a basis for confirmed
cases rather than the date of first symptoms. These findings suggest
that public mobility decreased in response to infection numbers reported
rather than mobility restrictions having an impact on case numbers.
Two important sources of bias are discussed that should be considered
for disease modelling based on public surveillance data. The strong
initial increase of case numbers observed in some countries might be an
artifact of the national testing policy. Furthermore, the numbers are
subject to a strong negative selection bias which does not allow for
valid conclusions on the population.
There is a continued and growing demand for representative data to
arrive at a more realistic picture of the true population-based hazard
potential of this novel virus