Non-linear associations of HOMA2-IR with all-cause and cardiovascular mortality in obese individuals: insights from NHANES 1999-2006

Abstract

Abstract Background The association of homeostatic model assessment (HOMA2-IR) and mortality among obese and non-obese population has not been clearly explained so far. Methods A total of 7,085 individuals aged ≥ 20 years from the National Health and Nutrition Examination Survey 1999–2006 were included in the study. The study endpoints were all-cause and cardiovascular mortality. Multivariate Cox proportional hazards regression models, restricted cubic spline analysis and Kaplan-Meier curve were employed for the evaluation. Results In a study comprising 59,815,733 weighted eligible participants, a total of 1666 all-cause deaths and 555 cardiovascular (CV) deaths were recorded over a mean follow-up period of 195.53 months. Notably, a significant difference in obesity was observed in relation to the association between HOMA2-IR and mortality. Upon adjusting for multiple variables, the highest group of HOMA2-IR exhibited an increased risk of all-cause and CV mortality among non-obese individuals when compared to the middle group. Conversely, the lowest group of HOMA2-IR was linked to higher CV mortality in obese participants compared to the middle group (adjusted hazard ratio, 1.82, 95% confidence interval, 1.14–2.92, P = 0.013). The results from the restricted cubic spline analysis displayed a J-shaped correlation between HOMA2-IR and all-cause, as well as CV mortality. Additionally, a nonlinear U-shaped correlation was observed in the obese population, with inflection points identified at 1.64 and 1.85. Below these inflection points, a negative relationship between HOMA2-IR and mortality was observed. Conclusions Elevated HOMA2-IR demonstrated a noteworthy correlation with escalated risks of both all-cause and CV mortality. It was noteworthy that excessively diminished levels of insulin resistance exhibited a distinct association with heightened mortality among obese individuals.