New Article in “Intereconomics” • 02.07.2020Pitfalls in the Interpretation of COVID-19 Data

Source: monsitj via Getty Images

The problems already begin with the choice of the figures used to measure the mortality of SARS-CoV-2. In addition, the interpretation of the data is made more difficult by the different registration and classification systems of COVID-19 cases in a country comparison of infection figures. In this interview, Dr. Andreas Backhaus talks about key findings of his analysis and points out what needs to be considered.

Dr. Backhaus, what traps can you fall into if you want to interpret the pandemic events appropriately?

With the spread of data in the course of the pandemic, a number of traps have emerged which can have a negative impact on the interpretation of the data and the actions taken as a result. This starts with the use of different epidemiological measures. In the public discourse, for instance, the Case Fatality Rate (CFR), the infectious mortality rate and the mortality rate are referred to. These three measures differ from each other and provide different values, which further adds to the confusion.

Another point addresses the basis of comparability of data between countries. For example, a comparison of the CFR between Italy and South Korea shows that in Italy this value exceeded that of South Korea at all times during the pandemic. It could therefore be assumed that the virus in Italy has more deadly consequences than in South Korea. However, this interpretation ignores the fact that first of all the comparability of the case mortality rates in different countries must be ensured. Only when the confirmed cases included in the calculation are sufficiently similar, for example with regard to their age, is a comparison possible. For these two countries, non-comparability is shown by the different age structure of the confirmed cases: Italy, for example, with a strongly increased proportion of older people among the confirmed infections, also reported more deaths than South Korea with a comparatively high proportion of young people among the confirmed cases.

What role do gaps in the data play?

Incomplete data transmission increases the risk of misinterpretation of the pandemic event. The extent to which data gaps can lead to distortions is demonstrated by the updates and revisions of the data material. It should therefore always be kept in mind that new data are often incomplete and often still subject to substantial revisions. Therefore, they are not suitable for immediate use as a basis for political evaluation.

How can we avoid the misinterpretation of data?

The examples shown demonstrate that the approaches and measures used must be understood, clearly defined and marked appropriately. If the same statistical measures are used, for example in a country comparison, it must be ensured that the fundamental data are sufficiently comparable. If there is any doubt about the accuracy of the data in the specific corona context, other independently collected data should also be used to support validation. Attention should be paid to the fact that data publications are already interpreted as final. There is always a high probability that they will be updated frequently. Furthermore, when interpreting data and statistics, the extent to which selection effects have influenced the composition of the basic sample should be considered.

Finally, when comparing the effectiveness of pandemic control policies, one should always ask why a particular measure has been taken in a specific context. A hard lockdown, for example, was often taken where the situation was already catastrophic at the time, while countries with milder initial histories often only took milder measures. The result would therefore be to incorrectly compare two measures that have been applied to completely different situations.

Backhaus, Andreas (2020): Common Pitfalls in the Interpretation of COVID-Data and Statistics. Intereconomics 55: 162–166.