METHODS: Patients were selected based on the availability of selenium biomarker information. Four subgroups were defined according to the patient's baseline status, including those with normal kidney function, reduced kidney function, selenium deficiency, and submaximal GPx3 activity.

RESULTS: Two hundred and forty-four patients were included in this analysis. Overall, higher serum concentrations of selenium, selenoprotein P (SELENOP) and GPx3 were correlated with less organ injury. GPx3 activity at baseline was predictive of 6-month survival (AUC 0.73; p = 0.03). While selenium treatment elevated serum selenium and SELENOP concentrations but not GPx3 activity in the full patient cohort, subgroup analyses revealed that GPx3 activity increased in patients with reduced kidney function, selenium deficiency and low to moderate GPx3 activity. Clinical outcomes did not vary between selenium treatment and placebo in any of these subgroups, though the study was not powered to conclusively detect differences in outcomes.