Two-Time-Point Inflammatory Status Patterns and the Risk of Incident Diabetes: A Cohort Study Based on CHARLS.

Background: Chronic inflammation plays a central role in the pathogenesis of diabetes mellitus (DM). Aims: This study delineated two-time-point inflammatory status patterns and examined their relationship with incident DM risk using data from the China Health and Retirement Longitudinal Study (CHARLS). Methods: The temporal changes in the inflammatory score between Wave 1 (2011) and Wave 3 (2015), which was calculated as the sum of Z-scores for C-reactive protein (CRP) and white blood cell (WBC) counts, were specified as the exposure. Incident DM was assessed at Wave 3 (2015) and Wave 4 (2018). Associations between inflammatory status patterns and incident DM was analyzed using a longitudinal targeted maximum likelihood estimation (LTMLE) framework in a cohort study. Results: A total of 4305 participants were included, 2404 (55.84%) of whom were female, with a median baseline age of 58 years. During a 7-year follow-up period, 730 individuals (17.0%) developed incident DM. Four distinct inflammatory status patterns were identified: low-stable, high-stable, increasing, and decreasing. After adjusting for multiple covariates, LTMLE analyses showed that the high-stable pattern was significantly associated with an increased risk of DM compared with the low-stable reference group (risk ratio [RR] = 1.430, 95% confidence interval [CI]: 1.204-1.698), with an E-value 2.21. In a landmark analysis of incident DM at Wave 4 (n = 150), the estimates for the high-stable pattern remained directionally positive across the adjusted models but were imprecise and compatible with no association. Exploratory subgroup analyses showed no statistical evidence of effect modification, as neither multiplicative nor additive interaction tests were statistically significant. Conclusions: A persistently elevated inflammatory score is significantly associated with a heightened risk of incident DM.