I was disappointed that Schurch and associates' comparison was made between protein supplements and an isocaloric carbohydrate feeding [1]. The carbohydrate feeding may have depressed patients' appetites so that protein intake was less than the patients normally would have consumed. I think a more appropriate comparison would have been between a protein supplement and no supplement; this is really the critical clinical question.

In the Results section, the authors implied that levels of albumin and prealbumin increased significantly and that the increase was due to the protein supplementation. This implication is not correct, of course, because the elevation in albumin and prealbumin levels was the same regardless of whether the protein or the carbohydrate supplement was administered. This supports other data showing that low levels of serum albumin and probably other serum markers do not reflect adequate nutrition but rather reflect the disease process in the host. I suspect that the relatively low serum albumin levels at the initiation of the study reflected the fact that the patients had just undergone the trauma of hip surgery.

Twelve serum and urine variables are reported in Table 3. Which of these variables was considered the most important outcome measure at the outset? With each additional variable, there is an increasing chance of having a difference between the protein and carbohydrate feeding. Was the difference seen in the insulin-like growth factor-I level simply statistical? Should a Bonferroni correction have been made in this analysis?