Patients were stratified by PCS or MCS scores
(above or below median), and survival was analysed
using Kaplan-Meier plots. The magnitude of the associations
between diabetes and mortality risk were estimated
with Cox-proportional hazard models, using a
univariate model entering only diabetes as a variable,
then adjusting for age, coronary artery disease, duration
of dialysis, and either PCS or MSC.