Facebook
Twitter
LinkedIn
Email
 |
Checking for direct PDF access through Ovid | |
Excerpt
In their recent, prospective, randomized trial, Dr. Bower and colleagues [1] concluded that septic intensive care unit patients who received enteral feedings supplemented with arginine, nucleotides, and fish oil (Impact Registered Trademark, Sandoz Nutrition, Minneapolis, MN) experienced a substantial reduction in length of stay and in the frequency of acquired infections compared with the group who received a routine enteral formula (Osmolite Registered Trademark, Ross Laboratories, Columbus, OH). They also stated that "both formulas were safe and well tolerated." I would like to offer a different perspective on the data presented.
Of the 33 deaths, 23 occurred in the group receiving the experimental formula. Although the authors correctly asserted that there were no significant differences in deaths between the groups, a chi-square test demonstrates that the trend toward increased death in the experimental group approaches statistical significance (p equals .058). There is only a 6% chance that the observed increase in mortality rate in the experimental group is due to chance. It is generally thought to be preferable to provide the actual p value in this situation and allow the reader to draw his/her own conclusion regarding the potential significance of the trend [2].
The trend toward an increased mortality rate also has important implications for the main end point in this study, length of stay. It is not clear from the study how length of stay was counted for those patients who died. If these patients were included in the calculation for length of stay, the greater number of deaths would result in a shorter length of stay in the experimental formula group because 14 of 23 deaths in that group occurred early (before feedings could be switched on day 7). Even if patients who died were not included in the length of stay calculation, removing them would also skew the length of stay data. Since those patients who died were sicker than those patients who did not, removing a greater number of sicker patients from the experimental group leaves a select healthier population that would be expected to have a shorter length of stay.
Although the study by Dr. Bower and colleagues [1] is interesting and important, I am not sure that the length of stay was shortened in a beneficial manner, and I am very concerned that the experimental formula may lead to an increase in the mortality rate. I urge caution in prescribing Impact until these issues are resolved by an appropriate study design in the subset of patients who appear to be septic.
