Attribution of diabetes to the development of severe liver disease in the general population.

Abstract

Diabetes is associated with advanced liver disease and predicts mortality regardless of the primary aetiology of the liver disease. Even a family history of diabetes has been linked to advanced liver fibrosis in non-alcoholic fatty liver disease (NAFLD). However, the fraction of liver-related outcomes in the general population that are attributable to diabetes remains unclear. The population attributable fraction (PAF) of diabetes for liver disease as a time-dependent exposure was estimated in the Finnish FINRISK study (n=28 787) and the British Whitehall II study (n=7855). We also assessed the predictive ability of a family history of diabetes for liver-related outcomes. Incident diabetes data were from drug purchase/reimbursement and healthcare registries (FINRISK) or follow-up examinations (Whitehall II). Incident severe liver outcomes were identified through linkage with national healthcare registries. Diabetes was associated with a two-fold risk of liver-related outcomes in both the FINRISK (HR, 1.92; p< .001) and Whitehall II (HR, 2.37; p< .001) cohorts, and this remained significant after adjusting for multiple confounders. PAF analyses demonstrated that diabetes explained 12-14% of the risk for severe liver-related outcomes after 10 and 20 years of follow-up. Also, maternal diabetes increased the risk of liver-related outcomes in the FINRISK (HR, 1.43; p= .044) and Whitehall II (HR, 2.04; p= .051) cohorts. Approximately 12%-14% of severe liver-related outcomes are attributable to diabetes at the population level. The association between maternal diabetes and liver disease might suggest a mitochondrial genetic mechanism.