Gastric versus post-pyloric feeding: a systematic review.

Abstract

Our objective was to evaluate the impact of gastric versus post-pyloric feeding on the incidence of pneumonia, caloric intake, intensive care unit (ICU) length of stay (LOS), and mortality in critically ill and injured ICU patients. Data sources were Medline, Embase, Healthstar, citation review of relevant primary and review articles, personal files, and contact with expert informants. From 122 articles screened, nine were identified as prospective randomized controlled trials (including a total of 522 patients) that compared gastric with post-pyloric feeding, and were included for data extraction. Descriptive and outcomes data were extracted from the papers by the two reviewers independently. Main outcome measures were the incidence of nosocomial pneumonia, average caloric goal achieved, average daily caloric intake, time to the initiation of tube feeds, time to goal, ICU LOS, and mortality. The meta-analysis was performed using the random effects model. Only medical, neurosurgical and trauma patents were enrolled in the studies analyzed. There were no significant differences in the incidence of pneumonia, percentage of caloric goal achieved, mean total caloric intake, ICU LOS, or mortality between gastric and post-pyloric feeding groups. The time to initiation of enteral nutrition was significantly less in those patients randomized to gastric feeding. However, time to reach caloric goal did not differ between groups. In this meta-analysis we were unable to demonstrate a clinical benefit from post-pyloric versus gastric tube feeding in a mixed group of critically ill patients, including medical, neurosurgical, and trauma ICU patients. The incidences of pneumonia, ICU LOS, and mortality were similar between groups. Because of the delay in achieving post-pyloric intubation, gastric feeding was initiated significantly sooner than was post-pyloric feeding. The present study, while providing the best current evidence regarding routes of enteral nutrition, is limited by the small total sample size.