Long-Term Weight Variability Is Associated with Higher Triglyceride Concentrations After Accounting for Long-Term Weight Slope

Background: Long-term fluctuations in body weight may reflect changes in energy balance, illness, or intentional weight loss. However, their association with subsequent triglyceride concentrations remains uncertain. Objective: To examine the association between long-term weight variability and subsequent triglyceride concentrations while accounting for directional weight change. We also assessed the robustness of the findings to adjustment for adiposity, missing data, and socioeconomic and lifestyle covariates. Methods: We analyzed adults from the China Health and Nutrition Survey who had at least three measured weights during 1989–2006 and a triglyceride measurement in 2009. The core complete-case analysis included 5191 participants and adjusted for baseline age, sex, weight slope, number of weight measurements, and working status in 2009. Weight variability was assessed using the coefficient of variation (CV; per 1 SD increment, standardized among 9500 eligible participants) and average successive variability (ASV; per 5 kg increment). Results: In the core model, each 1 SD increment in CV was associated with 2.7% higher triglyceride concentrations (β = 0.026, 95% CI: 0.007 to 0.045; p = 0.007). Each 5 kg increment in ASV was associated with 10.7% higher triglyceride concentrations (β = 0.102, 95% CI: 0.057 to 0.147; p < 0.001). After further adjustment for baseline BMI in the same BMI-complete sample, the corresponding estimates were 3.6% (95% CI: 1.7% to 5.6%) and 5.6% (95% CI: 1.0% to 10.3%), respectively. Multiple imputation yielded similar estimates. A four-knot restricted cubic spline analysis supported an overall association between CV and triglyceride concentrations (p = 0.016), with no strong evidence of nonlinearity (p = 0.101). Marginal standardization yielded adjusted differences in arithmetic mean triglyceride concentrations of 0.048 mmol/L per 1 SD increment in CV and 0.200 mmol/L per 5 kg increment in ASV. Conclusions: Greater long-term weight variability was associated with modestly higher subsequent triglyceride concentrations. Because of the observational design, the heterogeneous causes of weight change, and the possibility of residual confounding, the findings should not be interpreted as causal or independent of obesity.