Objective: The U-shaped association between body mass index (BMI) and mortality may depend on other traits with permanent health effects. Whether the association between BMI and mortality depends on levels of health-related traits known to be inversely associated with mortality throughout adult life such as height, intelligence, and education was investigated.
Method: The study was based on a cohort of young men with data on weight, height, intelligence test score, and education from the Danish Conscription Database. In total, 346,500 men born 1939–20195965ya were followed until December 2013. The association between BMI and mortality was analyzed using Cox-regression models including interactions between BMI and height, intelligence, and education, respectively.
Results: BMI and mortality showed the U-shaped association from the start of the follow-up period, and it persisted through the subsequent 56 years. As expected, the mortality was inversely associated with height, intelligence, and education, but the U shape of the association between BMI and mortality was unaffected by the levels of these traits except at higher BMI values, where the slopes were steeper for men with higher levels of height, intelligence, and education.
Conclusions: High and low BMI was associated with higher mortality throughout life regardless of the levels of height, intelligence, and education.
…Figures 3–5 illustrate the HRs of mortality for the interaction between BMI as a restricted cubic spline and the respective levels of height, intelligence, and education (reference values: BMI of 25.0 kg⁄m2 and medium levels of height, intelligence, and education, respectively). All the models with the interaction parameters fitted the data statistically-significantly better than the models without these interaction parameters. Overall, BMI and mortality had a U-shaped association that was elevated to higher HR levels for lower levels of height, intelligence, and education. The slopes of the curves that illustrate the HRs for the interaction between BMI and the 3 levels of height, intelligence, and education, respectively, were similar in the lower range of the BMI scale. However, at the higher range of the BMI scale, the slopes of the curves were steeper for the medium and high levels of height, intelligence, and education. Thus, on a multiplicative scale and based on multiplicative interactions, the influence of overweight and obesity for mortality was relatively stronger for the men with the medium and high levels of height, intelligence, and education.
Figure 3: Hazard ratios (HR) of the association between body mass index (BMI) as a cubic spline with 7 knots and interaction with short (solid), medium (dash), and tall (short dash dot) height for mortality. A BMI value of 25.0 kg⁄m2 and medium height was the reference, and the analysis was adjusted for strata of birth cohort and geographical area. Likelihood ratio test showed that the model with the interaction parameter between BMI and height had the best fit of data (p < 0.001).