Statistical analysis of three data sources for Covid-19 monitoring in Rhineland-Palatinate, Germany

Abstract

In Rhineland-Palatinate, Germany, a system of three data sources has been established to track the Covid-19 pandemic. These sources are the number of Covid-19-related hospitalizations, the Covid-19 genecopies in wastewater, and the prevalence derived from a cohort study. This paper presents an extensive comparison of these parameters. It is investigated whether wastewater data and a cohort study can be valid surrogate parameters for the number of hospitalizations and thus serve as predictors for coming Covid-19 waves. We observe that this is possible in general for the cohort study prevalence, while the wastewater data suffer from a too large variability to make quantitative predictions by a purely data-driven approach. However, the wastewater data and the cohort study prevalence are able to detect hospitalizations waves in a qualitative manner. Furthermore, a detailed comparison of different normalization techniques of wastewater data is provided.

Figure 1

Location of Rhineland-Palatinate in Germany and distribution of sewage plants (blue dots), cohort study cities (orange stars), and population density (grey shading). We created the maps ourselves with the python package geopandas. The maps are both based upon open data sources, which we retrieved from http://opendatalab.de/projects/geojson-utilities/ which itself uses geodata from the German Federal Agency for Cartography and Geodesy https://gdz.bkg.bund.de/ and population data from the Federal Statistical Office of Germany https://www.destatis.de/DE/Themen/Laender-Regionen/Regionales/Gemeindeverzeichnis/_inhalt.html. The coordinates from the referenced cities were manually collected from Google Maps.

Figure 2

Wastewater values (mean, N1, and N2) in dependence of normalization technique. Main trends are detectable in all types of genecopies and all normalization techniques. Normalizing by flow or PMMoV regularizes the curve.

Figure 3

Comparison of prevalence, genecopies, and hospitalizations. The main wave is observable in all three parameters.

Figure 4

Hospitalization prediction and corresponding feature importance derived by a random forest. The number of hospitalizations can be predicted well. The most important predictors are the cohort study prevalences. The term ‘lag’ defines the measurement ‘lag’ before the current measurement.

Figure 5

Prevalence prediction and corresponding feature importance derived by a random forest. The wastewater data do not allow a good prediction of the cohort study prevalence. The term ‘lag’ defines the measurement ‘lag’ before the current measurement.