Analysing premature cardiovascular disease mortality in the United States by obesity status and educational attainment

Abstract

Background: In the United States (US), premature cardiovascular disease (CVD) mortality rates (35-74 years) have exhibited increases in recent years, particularly in younger adults, and large differentials by educational attainment. This trend has occurred concurrently with high and increasing obesity prevalence, which also show significant differences by education. This study aims to jointly model premature CVD mortality trends in the US according to obesity status and educational attainment.

Methods: We used multiple cause of death data from the National Center for Health Statistics, obesity prevalence data from the National Health and Nutrition Examination Survey (NHANES), and educational attainment data from the American Community Survey and NHANES. We applied Bayes' theorem to these data to calculate the conditional probability of premature CVD mortality given obesity status and educational attainment for 2003-2019. We then projected this conditional probability for 2020-2029 using the Lee-Carter model.

Results: The probability of premature CVD mortality was greatest for obesity and low education (not graduated high school) and was substantially higher (females 6.7 times higher, males 5.9) compared with non-obesity and high education (Bachelor's degree or higher) in 2019. There was a widening of the gap in premature CVD mortality from 2003 to 2019 between the obese and non-obese populations, which occurred at each education level and was projected to continue in 2020-2029, especially for males. The conditional probability of premature CVD death for obesity and middle education (finished high school but no Bachelor's degree) increased substantially and was projected to surpass the level for non-obesity and low education in coming years for males and in younger age groups. At high education, the conditional probability of premature CVD death for the obese population was projected to increase to 2029, while for non-obesity it was projected to remain steady for females and fall for males; this projected widening is greatest at older age groups.

Conclusions: The findings demonstrate the public health challenge to reduce premature US CVD mortality posed by continued high obesity prevalence, especially for younger ages, lower education groups and males. The relative importance of obesity in influencing premature CVD mortality trends has risen partly due to the decline in CVD mortality attributable to other risk factors.

Keywords: Cardiovascular diseases; Education; Mortality; Multiple causes of death; Obesity; United States.

Fig. 1

CVD mortality rate trends (log scale), by sex, 35–74 years, US, 2003–2019. Note: This chart shows Pr(CVD) in the conditional probability equation. The vertical axis is on a log scale. Source: NCHS data

Fig. 2

Proportion of CVD deaths at each education level that were obesity-related, females, 35–74 years, US, 2003, 2010 and 2019. Source: NCHS data

Fig. 3

Proportion of CVD deaths at each education level that were obesity-related, males, 35–74 years, US, 2003, 2010 and 2019. Source: NCHS data

Fig. 4

Conditional probability of CVD deaths given obesity status and educational attainment level, by education and sex, 35–74 years, US, 2003–2019 (actual) and 2020–2029 (projection). Note: This chart shows Pr(CVD|O,E) in the conditional probability equation. The vertical axis is on a log scale. Shaded areas are 95% prediction intervals

Fig. 5

Conditional probability of CVD death given obesity status and educational attainment level, by age group and education, females, 35–74 years, 2003–2019 (actual) and 2020–2029 (projection). Note: This chart shows Pr(CVD|O,E) in the conditional probability equation. The vertical axis is on a log scale. Shaded areas are 95% prediction intervals

Fig. 6

Conditional probability of US CVD deaths given obesity status and educational attainment level, by age group and education, males, ages 35–74, years 2003–2019 (actual) and 2020–2029 (projection). Note: This chart shows Pr(CVD|O,E) in the conditional probability equation. The vertical axis is on a log scale. Shaded areas are 95% prediction intervals