Title : Improving the decision-making for measures in a future pandemic
Background and purpose: In combating COVID-19, different strategies led to different public health and economic outcomes of COVID-19. For instance, Sweden took mild measures, yielding poor general mortality and COVID-19 mortality, in contrast to the other Nordic countries. Here, we discuss challenges in the decision-making for a future pandemic and the possible improvement.
Challenges in decision-making for measures in combating pandemic
- Only table data is immediately available. With table data, one usually performs descriptive analysis, which provides immediate evidence for decision-making but does not adjust for population characteristics and updating pandemic situation.
- Pandemic progression is complex: measure influences the pandemic outcome, which in turn influences the subsequent measures and pandemic outcomes. It is one of the most challenging analyses to provide statistical evidence for decision-making, such as the long-term influence of measure on a remote outcome.
- Pandemic progression is hardly repetitive: every wave is different due to the new variant of disease and the effectiveness of vaccination. Therefore, it is nearly impossible to develop an optimal decision-making mechanism for new measures.
- Politics, economics, culture, and resource play important roles in decision-making and are always challenging. They do not influence the pandemic outcome directly but via the decision. Thus, they are not the evidence for decision-making in combating a pandemic.
Here, we focus on how to acquire evidence, based on which the decision is made.
Possible solution and illustration: Instead of the impossible optimal decision-making mechanism (challenge 3), we illustrate how to provide statistical evidence for decision-making (challenges 1 and 2). We can use the sequential causal inference to study the short-term and long-term influences of measure on pandemic outcomes. (Reference: Wang, X. and Yin, L. (2020). New G-Formula for the Sequential Causal Effect and Blip Effect of Treatment in Sequential Causal Inference. Annals of Statistics 48, 138-160).
We illustrate how to apply the sequential causal inference to table data by showing that the very early Swedish measure led to significant and sizable long-term influence on general mortality and COVID-19 mortality in comparison to the common measure adopted by the other Nordic countries. This evidence was not revealed by descriptive analysis but could have improved the decision-making for measures in the second wave. (Reference: Wang, X., Wallentin, Y.F. and Yin, L. (2022). The statistical evidence missing from the Swedish decision-making of COVID-19 strategy during the early period: A longitudinal observational analysis. Social Science and Medicine – Public Health. Volume 18, 101083).
Audience Take Away:
- Possible way to improve the decision-making for measures in a future pandemic.
- Pro and cons for using table data to provide timely statistical evidence for decision-making.
- General methodology of estimating the long-term influence of measure on public health and economic outcomes.