Cause of death diversity in multi-group settings: an application to Latin America and the Caribbean

Cause of death (CoD) diversity indices measure the extent to which some populations die from more similar or variegated causes than others. Higher CoD diversity implies higher unpredictability of the causes of individuals dying and greater challenges for health systems. In this paper, we propose a novel method to decompose overall CoD diversity as the sum of two interpretable parts: the within- and between-group components.

The novel approach is applied to Latin America and the Caribbean (LAC) region to illustrate its usefulness. We decompose overall CoD diversity, measured by the Simpson index of diversity, into between-country and within-country components. In addition, we provide further decompositions assessing how each cause of death and each country contributes to overall CoD diversity in the region.

The CoD diversity in the region followed a nonmonotonic trend. From 2000 to 2018, the CoD diversity increased from 0.81 to 0.83 for women, reaching approximately 0.84 for men. El Salvador, Peru, and Uruguay are the countries that contribute the most to explaining the differences in the mortality profile between countries, but for very different and opposing reasons. While the high diversity in El Salvador and Peru can be explained by causes of deaths related to the early stages of the epidemiological transition, such as communicable causes, respiratory causes, and external causes, Uruguay presents a high diversity because the deaths are very dispersed between chronic conditions. Cardiovascular deaths are the main contributor to both CoD diversity levels and their changes over time. As cardiovascular deaths decline, they give way to other chronic causes, which become more prominent and contribute to diversifying the corresponding mortality profiles. However, external causes also significantly contribute to forming uneven epidemiological profiles.

The decomposition proposed in this paper makes possible to assess whether some groups contribute more or less to the uncertainty around the causes of individuals' deaths and identify the sources of CoD diversity. In this way, this approach can contribute to a better understanding of contemporary mortality dynamics, especially in a context with large health inequalities.