OP10 Worse baseline risk factors explain the association of healthy obesity with increased mortality risk: whitehall II study. (September 2018)

• Record Type: Journal Article
• Authors: Johnson, W; Bell, JA; Robson, E; Norris, T; Kivimaki, M; Hamer, M
• Is Part Of: Journal of epidemiology and community health. Volume 72(2018)Supplement 1
• Page Start: A5
• Page End: A5
• Publication Date: 2018-09
• Subjects: healthy obesity -- risk factor trajectories -- mortality; Public health -- Periodicals; Epidemiology -- Periodicals; 614.4
• DOI: 10.1136/jech-2018-SSMabstracts.10
• Languages: English
• ISSNs: 0143-005X

Abstract

Abstract : Background: The extent to which dichotomizing variables to define metabolically healthy obesity (MHO) inherently results in different risk factor trajectories and mortality, relative to a healthy non-obese group, is unknown. We aimed to describe 20-year risk factor trajectories according to initial weight/health status and investigate the extent to which baseline differences explain greater mortality among MHO than healthy non-obese individuals. Methods: The sample comprised 6529 participants in the Whitehall II study who were measured at five assessments between 1991–1994 and 2012–2013. Baseline weight (non-obese or obese; body mass index (BMI) ≥30 kg/m 2 ) and health status (healthy or unhealthy; two or more of hypertension, low high-density lipoprotein cholesterol (HDL-C), high triglycerides, high glucose, and high homeostatic model assessment of insulin resistance (HOMA-IR)) were defined using a standard approach. The relationships of baseline weight/health status with 20-year trajectories summarizing ∼25 000 observations of systolic and diastolic blood pressures, HDL-C, triglycerides, glucose, and HOMA-IR were investigated using multilevel models. Associations of baseline weight/health status with all-cause mortality up until July 2015 were investigated using Cox proportional hazards regression. Analyses were adjusted for sex, ethnicity, and baseline covariates (age, alcohol, smoking, exercise, occupational grade, GHQ-rated psychological distress, and diet). Results: Approximately 90% of the sample was non-obese and 75% of these individuals were healthy. Conversely, among the 10% of the sample that was obese only 40% were healthy. Baseline levels and trajectories of metabolic risk factors tended to be consistently worse for the MHO group compared to the healthy non-obese group (e.g., HOMA-IR by 0.91 (95% CI 0.67, 1.14) units at 20-years of follow-up). Consequently, the MHO group had a greater risk of mortality when the referent group comprised either all healthy non-obese individuals in our sample (hazard ratio 1.57 (1.15, 2.15)) or a random sample of healthy non-obese individuals (2.11 (1.24, 3.58)). This estimate, however, attenuated (1.34 (0.85, 2.13)) when the referent group was matched to the MHO group on baseline risk factors (i.e., blood pressures, HDL-C, triglycerides, glucose, and HOMA-IR) using propensity scores. Conclusion: This paper demonstrates how dichotomising continuous variables results in different levels of cardio-metabolic disease risk factors over 20 years of follow-up between MHO and healthy non-obese individuals. The greater disease and mortality risk of MHO compared to healthy non-obese individuals, observed in large scale epidemiological studies, is likely largely explained by the more deleterious risk factor trajectories that result from crude stratification, further challenging the concept of healthy obesity.