Hemoglobin Levels Across the Pediatric Critical Care Spectrum: A Point Prevalence Study.

Objectives: To determine the prevailing hemoglobin levels in PICU patients, and any potential correlates.

Design: Post hoc analysis of prospective multicenter observational data.

Settings: Fifty-nine PICUs in seven countries.

Patients: PICU patients on four specific days in 2012.

Interventions: None.

Measurements And Main Results: Patients' hemoglobin and other clinical and institutional data. Two thousand three hundred eighty-nine patients with median age of 1.9 years (interquartile range, 0.3-9.8 yr), weight 11.5 kg (interquartile range, 5.4-29.6 kg), and preceding PICU stay of 4.0 days (interquartile range, 1.0-13.0 d). Their median hemoglobin was 11.0 g/dL (interquartile range, 9.6-12.5 g/dL). The prevalence of transfusion in the 24 hours preceding data collection was 14.2%. Neonates had the highest hemoglobin at 13.1 g/dL (interquartile range, 11.2-15.0 g/dL) compared with other age groups (p < 0.001). The percentage of 31.3 of the patients had hemoglobin of greater than or equal to 12 g/dL, and 1.1% had hemoglobin of less than 7 g/dL. Blacks had lower median hemoglobin (10.5; interquartile range, 9.3-12.1 g/dL) compared with whites (median, 11.1; interquartile range, 9.0-12.6; p < 0.001). Patients in Spain and Portugal had the highest median hemoglobin (11.4; interquartile range, 10.0-12.6) compared with other regions outside of the United States (p < 0.001), and the highest proportion (31.3%) of transfused patients compared with all regions (p < 0.001). Patients in cardiac PICUs had higher median hemoglobin than those in mixed PICUs or noncardiac PICUs (12.3, 11.0, and 10.6 g/dL, respectively; p < 0.001). Cyanotic heart disease patients had the highest median hemoglobin (12.6 g/dL; interquartile range, 11.1-14.5). Multivariable regression analysis within diagnosis groups revealed that hemoglobin levels were significantly associated with the geographic location and history of complex cardiac disease in most of the models. In children with cancer, none of the variables tested correlated with patients' hemoglobin levels.

Conclusions: Patients' hemoglobin levels correlated with demographics like age, race, geographic location, and cardiac disease, but none found in cancer patients. Future investigations should account for the effects of these variables.