The second wave of SARS-CoV-2 infections and COVID-19 deaths in Germany – driven by values, social status and migration background? A county-scale explainable machine learning approach

ABSTRACT

There is a general consensus that SARS-CoV-2 infections and COVID-19 deaths have hit lower social groups the hardest, however, for Germany individual level information on socioeconomic characteristics of infections and deaths does not exist. The aim of this study was to identify the key features explaining SARS-CoV-2 infections and COVID-19 deaths during the upswing of the second wave in Germany.

We considered information on COVID-19 diagnoses and deaths from 1. October to 15. December 2020 on the county-level, differentiating five two-week time periods. We used 155 indicators to characterize counties in nine geographic, social, demographic, and health domains. For each period, we calculated directly age-standardized COVID-19 incidence and death rates on the county level. We trained gradient boosting models to predict the incidence and death rates with the 155 characteristics of the counties for each period. To explore the importance and the direction of the correlation of the regional indicators we used the SHAP procedure. We categorized the top 20 associations identified by the Shapley values into twelve categories depicting the correlation between the feature and the outcome.

We found that counties with low SES were important drivers in the second wave, as were those with high international migration and a high proportion of foreigners and a large nursing home population. During the period of intense exponential increase in infections, the proportion of the population that voted for the Alternative for Germany (AfD) party in the last federal election was among the top characteristics correlated with high incidence and death rates.

We concluded that risky working conditions with reduced opportunities for social distancing and a high chronic disease burden put populations in low-SES counties at higher risk of SARS-CoV-2 infections and COVID-19 deaths. In addition, noncompliance with Corona measures and spill-over effects from neighbouring counties increased the spread of the virus. To further substantiate this finding, we urgently need more data at the individual level.