Best paper award of the Swedish Society of Medical Statistics (FMS) 2024

Winner of Swedish Society for Medical Statistics best student paper 2024 is Stina Zetterström for her article Bounds for selection bias using outcome probabilities¹, written at the Department of Statistics at Uppsala university. 

Stina addresses how causal bounds for selection bias can be estimated to highlight the extent to which selection to the study might impact the causal question. This is a theoretical paper, aimed to advance causal inference theory.

In this field, causal bounds are typically estimated based on assumptions on presumed values of some sensitivity parameter. Stina’s contribution is that her sensitivity parameters are based on probabilities for the outcome given the exposure level and unmeasured confounder, rather than on risk ratios. Her main argument for this approach to estimating a bound for the selection effect is that it might be easier to provide reasonable values for these probabilities, rather than for relative risks. She derives bounds for causal risk ratios and for risk differences and discusses the performance on both these scales. Lastly, she contrasts different forms of bounds in a simulation study and demonstrate the use of bounds in a practical example based on real data. 

Causal inference is an emerging field in medical statistics and with this prize FMS wishes to highlight the importance of methodological development and papers that help guiding the reader in their choice of methods. Stina’s work is a fine example of that and as sole author of the paper her own contribution to this field is clear. We hope that this recognition encourages Stina and others to further advance our field and thereby contribute to advancing also applied medical research through uptake of novel theoretical ideas.

Presentation slides

References

1. Zetterstrom S Bounds for selection bias using outcome probabilities; Epidemiol. Methods 2024; 13(1): 20230033 ↩︎