Do data support nutrition support? Part I: intravenous nutrition.

Intravenous (parenteral) nutrition has been advocated widely as adjunctive care in patients with a variety of underlying diseases. However, the enthusiasm for this therapeutic intervention was based largely on expert opinion. Because the best way to assess the efficacy of any treatment is to test it in a randomized controlled trial, this review will focus on data that was derived from such studies. Using established search strategies, randomized controlled trials were sought that compared one of two forms of intravenous nutrition: parenteral nutrition (nitrogen and >or=10 kcal/kg/day of non-protein calories for >or=5 days) or protein-sparing therapy (nitrogen and fewer non-protein calories) with no type of artificial nutrition beyond regular food and/or standard (5%) dextrose. The randomized controlled trials were stratified by the underlying disease state. The clinical outcomes of interest were mortality, morbidity (total/infectious complications), and/or duration of hospitalization. More than 100 randomized controlled trials failed for the most part to demonstrate that intravenous nutrition had any effect on clinical outcome. There were a few exceptions. In patients undergoing attempted curative surgery for upper gastrointestinal cancer, the use of preoperative parenteral nutrition seemed to reduce the incidence of major postoperative complications. However, this benefit was only found in low-quality randomized controlled trials. Findings conflict regarding the use of parenteral nutrition in patients with acute pancreatitis or undergoing bone marrow transplantation. Parenteral nutrition was harmful when provided to patients undergoing radiation or chemotherapy for cancer. Although no randomized controlled trials exist, it is assumed that parenteral nutrition is useful in patients with an inadequate gastrointestinal tract ("short gut"). Thus, for the most part, randomized controlled trials comparing intravenous nutrition to no artificial nutrition have not shown that this medical intervention is of benefit.