Cause-of-Death Dependencies: Structure, Evolution by Age and Sensitivity to Disruptions

Elizabet Ukolova , University of Southern Denmark, Interdisciplinary center for Population Dynamics, Odense
Trifon I. Missov, Interdisciplinary Centre on Population Dynamics (CPop), University of Southern Denmark

Typically, mortality analysis by causes is performed for only 35% of the death records listed on death certificates. This limitation primarily stems from research's predominant focus on the single underlying cause of death, despite the fact that almost 80% of deaths in the USA result from multiple causes. Consequently, this gap in understanding the relationships between causes of death has emerged. We are left with essential questions: What is the structure of cause of death dependencies? How does this structure evolve with age? How would it change if hypothetical disruptions in flows of deaths between diseases would be embedded into the structure of cause of death dependencies? To address these questions, we use US Multiple Cause of death Data from National Center for Health Statistics. Our preliminary findings highlight 20 leading causes of death pairs to which nearly 65% of deaths in the US are attributed. We observe a strong age-dependent and cause-specific pattern in the strength of associations, as measured by odds ratios, cause-of-death association indicator, and conditional probability of dying with cause j given a person had cause i. Furthermore, we explore the impact of dissolving significant associations between causes i and j on the strength of associations between causes i and k.

See extended abstract

 Presented in Session 29. Flash session Morbidity