Temporal stability of cardiometabolic risk factors from the perspective of primordial disease prevention in children.

Abstract

Excess adiposity and circulating lipoprotein lipids in childhood are predictive of cardiometabolic disease in adulthood. We analysed temporal tracking for body mass and lipid biomarkers using population-based time-series to assess the statistical feasibility of primordial prediction and prevention. Individuals born between 1962-1992 were included from four European local community cohorts. Serial measurements of weight, height, waist and lipoprotein lipids were analysed for 12,212 participants (ages 5-49 years, follow-ups 15-31 years depending on cohort). Longitudinal tracking was quantified by autocorrelation R2 over the same individuals between any two visits in the series; the weakening of R2(Δt) with increasing follow-up time was modelled by exponential decay. Based on the decay model, we calculated autocorrelation for two surveys fifteen years apart as a simplified tracking indicator: R2(15) = 35.7% for body-mass index, R2(15) = 28.1% for waist-height ratio and R2(15) = 32.5% for low-density lipoprotein (LDL) cholesterol. Total and high-density lipoprotein cholesterol were comparable to LDL cholesterol. Triglycerides exhibited weak tracking. The detection rate (DR5) at 5% false positive rate (FPR) of children who were in the top risk decile as adults was DR5(15) = 38.9% for body mass index and DR5(15) = 30.9% for LDL cholesterol. Individualised biomarker screening at low FPR misclassifies the majority of children who will have high values as adults (low statistical feasibility). Biomarker tracking is more useful for guiding broad population-based primordial prevention programmes that can be extended to all children at risk without harm from false positives.