Health research during the COVID-19 pandemic

Published in El Pais, 27 February 2021.

Decision-making to respond to COVID-19 is critical for both business and government. The ideal is to take them based on the evidence given by research. But do we do it? Many do, and the key is to follow a good procedure. In the case of the daily COVID-19 numbers on cases and deaths, it triggered my curiosity to know if they really helped making good decisions or many with high errors. I would like to mention some aspects on this topic.

1. Sampling. When they mention the confirmed cases, are they from sick people who took the test or from healthy people? Many only test people with symptoms and thus it is normal that they give high number of positives (70% of the tests were positive). Others use tests that have a high error level (rapid tests) so that “40%” of all are false positives. And when the laboratories do not have tests or it is a holiday, a decrease in positive cases is reported.
2. Collection and processing of data. How much does COVID-19 impact the population? With results without details of age, sex, underlying disease, workplace, or study, it is not possible to know how much COVID-19 impacts society or if the measures taken are exaggerated or not for the population. Already many cities in different countries are publishing detailed data to take particular measures to the city. For example, if it was heavily hit in the first wave, the impact on the second wave may not be as severe, and the measures will be less drastic. Another example of this is that, if children were not affected, they could return to face-to-face classes without the need for masks and take immediate vaccination of teachers as measures. The right decisions depend on the detail of the data collection and how it is processed. Many models put the population the same way regardless of the region, or age profile. But certainly there are cities with their particularities that have prevalent underlying diseases, or are young populations, etc. This is important to include in the models.
3. Validity of the results. The information we have will give some validity to the results. If the information is detailed, it will generate particular results and with less error in these particular cases. But if the information is collected irregularly and without greater detail, it will generate greater error in the forecasts.
4. Conclusions and recommendations. Errors in the data generate errors in decision-making and measures that can be taken for COVID-19.

Today shows the great importance of having the most detailed information possible and available about the population so that it reaches the people who want to investigate enabling them to suggest better measures to take for the future.