What if dependent causes of death were independent?

Abstract In the prevailing era of cardiovascular and chronic diseases, the causes of death seldom operate in isolation. We aim to assess the impact of distortions in statistically significant associations between causes of death on the overall structure of cause of death dependencies. Focusing on the causes of death for USA in 2019, we study all possible scenarios for independence of diseases by optimizing the marginal distributions in the respective contingency tables. We select the independence scenario that corresponds to the minimal intervention in the contingency table and study the effect of setting up independence between diseases i and j on the relationships between i and k (or j and k ). Disrupting associations between leading cause of death pairs would generally lead to decreases in the strength of relations these diseases have with others outside the disrupted pair. The most pronounced effects are observed in the loss of association between mental and behavioral disorders with other respiratory diseases; heart diseases with endocrine, nutritional, and metabolic disorders; and heart diseases with mental and behavioral disorders. Changes in odds ratios between leading cause of death pairs do not occur in isolation from other diseases. Our results may serve as a reference point for optimizing strategies for multimorbidity prevention.