Background The role of enteral nutrition in Crohn's disease is controversial. Increasing research on the mechanisms by which nutritional therapy improves the clinical well being of patients with Crohn's disease has led to novel formula design and trials comparing two different forms of enteral nutrition. This meta-analysis aims to provide an update on the existing effectiveness data for both corticosteroids versus enteral nutrition and for one form of enteral nutrition versus another for inducing remission of active Crohn's disease. Objectives To evaluate the effectiveness of exclusive enteral nutrition ( EN) as primary therapy to induce remission in Crohn's disease and to examine the importance of formula composition on effectiveness. Search strategy Studies were selected using a computer-assisted search of the on-line bibliographic databases MEDLINE ( 1966-2006) and EMBASE ( 1984-2006), as well as the Science Citation Index on Web of Science. Additional citations were sought by manual search of references of articles retrieved from the computerized search, abstracts submitted to major gastroenterologic meetings and published in the journals: American Journal of Gastroenterology, Gut, Gastroenterology, Journal of Pediatric Gastroenterology and Nutrition, and Journal of Parenteral and Enteral Nutrition, and from the reviewers' personal files or contact with leaders in the field. Selection criteria All randomized and quasi-randomized controlled trials involving patients with active Crohn's disease defined by a clinical disease activity index were considered for review. Studies evaluating the administration of one type of enteral nutrition to one group of patients and another type of enteral nutrition or conventional corticosteroids to the other group were selected for review. Data collection and analysis Data were extracted independently by two authors and any discrepancies were resolved by rereading and discussion. For the dichotomous variable, achievement of remission, individual and pooled trial statistics were calculated as odds ratios ( OR) with 95% confidence intervals ( CI); both fixed and random effect models were used. The results for each analysis were tested for heterogeneity using the chi square statistic. The studies were separated into two groups: A. one form of enteral nutrition compared with another form of enteral nutrition and B. one form of enteral nutrition compared with corticosteroids. Subgroup analyses were conducted on the basis of clinical or disease criteria and formula composition. Sensitivity analyses were conducted on the basis of the inclusion of abstract publications, methodologic quality and by random or fixed effects models. Main results In part A, of the 15 included eligible trials ( one abstract) comparing different formulations of EN for the treatment of active CD, 11 compared one ( or more) elemental formula to a non-elemental one, three compared enteral diets of similar protein composition but different fat composition, and one compared non-elemental diets differing only in glutamine enrichment. Meta-analysis of ten trials comprising 334 patients demonstrated no difference in the efficacy of elemental versus non-elemental formulas ( OR 1.10; 95% CI 0.69 to 1.75). Subgroup analyses performed to evaluate the different types of elemental and non-elemental diets ( elemental, semi-elemental and polymeric) showed no statistically significant differences. Further analysis of seven trials including 209 patients treated with EN formulas of differing fat content ( low fat: < 20 g/ 1000 kCal versus high fat: > 20 g/ 1000 kCal) demonstrated no statistically significant difference in efficacy ( OR 1.13; 95% CI 0.63 to 2.01). Similarly, the effect of very low fat content (< 3 g/ 1000 kCal) or type of fat ( long chain triglycerides) were investigated, but did not demonstrate a difference in efficacy in the treatment of active CD, although a non significant trend was demonstrated favoring very low fat and very low long chain triglyceride content. This result should be interpreted with caution due to statistically significant heterogeneity and small sample size. Sensitivity analyses had no significant effects on the results. The role of specific fatty acids or disease characteristics on response to therapy could not be evaluated. In part B, eight trials ( including two abstracts) comparing enteral nutrition to steroid therapy met the inclusion criteria for review. Meta-analysis of six trials that included 192 patients treated with enteral nutrition and 160 treated with steroids yielded a pooled OR of 0.33 favouring steroid therapy ( 95% CI 0.21 to 0.53). A sensitivity analysis including the abstracts resulted in an increase in the number of participants to 212 in the enteral nutrition group and 179 in the steroid group but the meta-analysis yielded a similar result ( OR 0.36; 95% CI 0.23 to 0.56). There were inadequate data from full publications to perform further subgroup analyses by age, disease duration and disease location. Authors' conclusions Corticosteroid therapy is more effective than enteral nutrition for inducing remission of active Crohn's disease as was found in previous systematic reviews. Protein composition does not influence the effectiveness of EN in the treatment of active CD. A non significant trend favouring very low fat and/or very low long chain triglyceride content exists but larger trials are required to explore the significance of this finding.
