Nutritional support currently accounts for about 1% of the total healt
h care costs in the USA. Interestingly, most of the prospective random
ized controlled trials to date have not been able to demonstrate that
this therapeutic intervention alters morbidity or mortality. In fact,
parenteral nutritional support may predispose the recipients to develo
ping systemic infections. There have been a few areas in which nutriti
onal support may be of benefit. Enteral supplements given to underweig
ht women who suffer hip fractures reduce the hospital stay and, presum
ably, overall cost. Preoperative parenteral nutritional support may pr
oduce a small absolute reduction in post-operative morbidity, but its
cost becomes prohibitive. Preoperative enteral nutritional support, es
pecially if carried out in the home, may be of benefit (using the most
optimistic interpretation of a small number of trials); if so, it is
an economically defensible intervention. Particular nutrients or diets
may have specific effects on certain disease processes. Indirect comp
arisons have suggested that elemental diets can be used to treat flare
s of Crohn's disease (perhaps because putative food antigens are remov
ed). However, corticosteroid therapy is more efficacious. Furthermore,
it is less expensive to employ 6-mercaptopurine as the next modality
in steroid failures. Branched-chain amino acid infusions may have some
effect on hepatic encephalopathy, but again, lactulose is less expens
ive. Nutritional support is one area of medicine in which there has be
en far more enthusiasm than the data justify. Disease-associated malnu
trition probably is a secondary phenomenon, not an important cause of
morbidity. The widespread use of this modality cannot be justified in
a cost-constrained health care system.