Fig 1. Flow chart for sample selection (RRT: Renal replacement therapy).

Relative to the excluded patients, the included patients did not differ significantly in terms of gender and the prevalence of congestive heart failure but were older (+1 year, p<0.001), had a slight lower eGFR at initiation (-0.2ml/min, p = 0.03), were less likely to have diabetes (-2%, p<0.002) and had different patterns of initially treated conditions or primary renal disease (Table A in S1 File).