LAUSR

Supplemental Digital Content is available in the text. Objectives: To evaluate serum ferritin measured within 48 hours of admission as a prognostic marker and examine the association with unfavorable outcomes in a population of pediatric patients with sepsis and high prevalence of iron deficiency anemia in which this biomarker is routinely measured. Design: Retrospective cohort study. Setting: PICU of a tertiary care teaching hospital in a middle-income country in South America. Patients: All patients 6 months to 18 years old (n = 350) admitted with a diagnosis of sepsis, suspected or proven, were eligible for inclusion. Exclusion criteria were length of PICU stay less than 8 hours and inherited or acquired disorder of iron metabolism that could interfere with serum ferritin levels. Interventions: None. Measurements and Main Results: Three-hundred twelve patients had their ferritin levels measured within 48 hours, and only 38 did not. The prevalence of iron deficiency anemia (hemoglobin < 11 g/dL and mean corpuscular volume < 80 fl was 40.3%. The median of the highest serum ferritin level within 48 hours was 150.5 ng/mL (interquartile range, 82.25–362 ng/mL), being associated with mortality (p < 0.001; Exp(B), 5.170; 95% CI, 2.619–10.205). A 10-fold increase in ferritin level was associated with a five-fold increase in mortality. There was a monotonic increase in mortality with increasing ferritin levels (p < 0.05). Regarding the discriminatory power of ferritin for mortality, the area under the receiver operating characteristic curve was 0.787 (95% CI, 0.737–0.83; p < 0.0001). Conclusions: Serum ferritin at lower thresholds predicts mortality in children with sepsis admitted to the ICU in a middle-income country with high prevalence of iron deficiency anemia.