In this report, we will take a closer look at the detections we have observed over the course of the last year and compare the fever curve to other existing COVID-19-related public health data.
In order to more clearly visualize trends in the data, the following figures show the curves generated by aggregating the detections by calendar week. Correspondingly, we will be updating the online fever monitor within the next few days to display these weekly detections from now on.
Weekly detections compared to COVID-19 cases
In this first figure, we look at the breakdown of detections by age group compared to new confirmed COVID-19 cases per week. Of the active donors, the overwhelming majority fall within the ages of 35 to 59 at nearly 65% of donors, while the 15-34 age group is the least represented, containing only 10% of regular users. The remaining 25% belong to the 60+ group.
It is important to note that only 53% of active users have provided data pertaining to age and gender, so our estimates must be taken with a grain of salt. Regardless, the detection curves for these groups all show decent correspondence with the age-stratified case counts. Predictably, the least represented youngest age group shows the highest inconsistencies.
If we look at detections by gender, we see similar agreement with trends in the reported COVID-19 case numbers, but keep in mind that our donor population is biased towards the female population who make up 60% of our donors.
If we look at the regional distribution of the detections and our donors, we find that 89% of our donors are found in western Germany compared to 82% of the actual German population.
For more details on donor demographics, including a comparison of the age and gender distribution of the donors to the German population, please see our earlier post here: Donor demographics.
Comparison with other data sources
Next, we compare the overall weekly detection curve to three data sources:
i) the confirmed case counts, as before,
ii) the test positivity rate, and
iii) the number of COVID-19 patients in intensive care.
In all curves, we can see similar trends, characterized by two distinct waves, peaking around December 2020 and April 2021. While the case counts appear to be in a phase with the detections, the test positivity and ICU data are clearly shifted forward in time. Upon closer inspection, preliminary analyses suggest that the detections precede all three measures. If these findings hold true, a rise in detections may predict a rise in cases. We are currently evaluating the extent to which these detections can improve nowcasting estimates from current surveillance measures, which can be delayed by 1 to 3 weeks.