Association Between Enteral Feeding, Weight Status, and Mortality in a Medical Intensive Care Unit.

The study was observational, not randomised, so indication bias could explain the association. Researchers used an order for enteral feeding as a proxy, not direct records of feeding volume or tolerance. The database did not show when feeding started, how much was given, whether it was trophic or to goal, or why patients were or were not fed. Patients with an order had much longer mechanical ventilation than patients without an order, so the groups differed in care intensity. The analysis excluded some visits for missing data, leaving 811 hospitalizations for 777 patients from 892 hospitalizations for 857 patients. It was a single 24-bed medical ICU in 2013, so generalisability may be limited. BMI was an average hospital stay value and may be a poor surrogate for nutritional status. There were few underweight patients, so the study could not detect benefit in that group. The study did not know which patients received vasopressors. Sensitivity analyses reduced some bias by excluding total parenteral nutrition patients, ICU stays under 24 hours, and deaths within 24 hours, but unmeasured confounding remained possible.

The supplied text does not report a conflict-of-interest statement or funding source. The publication types list Research Support, N.I.H., Extramural.

Do not conclude that enteral feeding caused lower mortality. This was a retrospective observational study using an order as a proxy, and patients judged unlikely to survive or planning withdrawal of life support may have been less likely to receive an order. The study also did not know feeding timing, dose, tolerance, or clinical reasons for feeding, and it cannot separate feeding from other ICU care decisions.