Association of cumulative SPISE and two-point SPISE patterns with incident cardiovascular disease in middle-aged and older Chinese adults: a prospective landmark cohort analysis of CHARLS

Single-point insulin sensitivity estimation (SPISE) can be calculated from routine metabolic and anthropometric measures, but longitudinal evidence for cardiovascular disease (CVD) remains limited. We examined whether cumulative SPISE and outcome-blind two-point SPISE patterns were associated with incident CVD in middle-aged and older Chinese adults. We performed a 2015-landmark cohort analysis using the China Health and Retirement Longitudinal Study. Participants were required to have valid 2011 and 2015 SPISE measurements, be free of CVD during the exposure window, and have post-landmark outcome or death follow-up. Cumulative SPISE was calculated as the mean of the two values multiplied by each participant’s actual assessment interval. Cox models with 100 multiple imputations, restricted cubic splines, outcome-blind K-means clustering, subgroup analyses, and prespecified sensitivity analyses were used. Among 4,737 participants, 893 incident CVD events occurred during a median 5.0 years of post-landmark follow-up. In the primary Model 3, each 1-SD higher cumulative SPISE was associated with lower incident CVD (hazard ratio [HR] 0.79, 95% confidence interval [CI] 0.74–0.86), and the HR for the highest versus lowest quartile was 0.57 (95% CI 0.46–0.69; P for trend < 0.001). The four-knot spline showed an overall association ( P < 0.001) without evidence of nonlinearity ( P = 0.227). K-means selected two SPISE patterns; the higher versus lower SPISE pattern had an adjusted HR of 0.63 (95% CI 0.54–0.73). The main result was directionally consistent in prespecified sensitivity analyses. Higher cumulative SPISE and a higher two-point SPISE pattern were associated with lower subsequent self-reported incident CVD. These observational findings support cumulative SPISE as a longitudinal metabolic exposure for research, while two measurements do not justify causal or trajectory-based conclusions.