Review Article A metaanalysis of treatment outcomes of early enteral versus early parenteral nutrition in hospitalized patients* John Victor Peter, MBBS, MD, DNB (Med); John L. Moran, MBBS, FRACP, FANZCA; Jennie Phillips-Hughes, RN C ritical illnesses, stress, and surgery place increased de- mands on the body’s nutri- tional requirements. These conditions promote a catabolic state and negative nitrogen balance. Prolonged bed rest and inactivity, per se, produce a neg- ative nitrogen balance in healthy individ- uals (1), and this effect is accentuated by exogenous steroids (2). Thus, a hyper- metabolic state, like in critical illness, trauma, or sepsis, in concert with bed rest and inactivity form a suitable envi- ronment for the occurrence of malnutri- tion. Nutritional problems are common in hospitalized patients with the incidence varying depending on the population screened and the type of screening tool used. In a prospective study of 995 con- secutive patients in Switzerland, malnu- trition, defined by the authors as a body mass index (BMI) of 20 kg/m 2 , was ob- served in 17.3% (3). In an Australian study, the incidence of malnutrition, as- sessed by subjective global assessment, was 36% in a cohort of 819 patients (4). Patients identified as malnourished using BMI criteria, or nutritionally depleted by subjective global assessment, tend to stay longer in the hospital with increased costs of care (4 – 6) and higher mortality (4). Nutritional supplementation affords the opportunity of slowing down or stop- ping the catabolic process, restoring ni- trogen balance, and preventing malnutri- tion. Total parenteral nutrition (PN) was popular in the 1970s and 1980s (7) when it was used indiscriminately to counter- act the metabolic problems associated with illnesses. However, within the para- digm of evidence-based medicine, a sig- nificant benefit of total PN has not been demonstrated. Heyland et al. (8), evaluat- ing 26 randomized, controlled trials (RCTs) involving 2,211 patients, com- pared the use of PN with standard care (usual oral diet plus intravenous fluids) in patients undergoing surgery. The re- sults of this metaanalysis, reported as risk ratio (RR), suggested no mortality benefit with PN (RR, 1.03; 95% confidence inter- val [CI], 0.81–1.31). There was a trend to a reduction in complication rates with PN (RR, 0.84; 95% CI, 0.64 –1.09), with sig- nificant benefits in malnourished pa- tients (RR, 0.52; 95% CI, 0.30 – 0.91). Enteral nutrition (EN), on the other hand, has been advocated as a means of reducing mucosal atrophy and increased intestinal permeability with consequent reduction in the incidence of gut trans- location and septic complications. More- over, EN is popular because it is cheaper, *See also p. 260. From the Intensive Care Unit, The Queen Elizabeth Hospital, Woodville, South Australia. Study funded by Unit Trust Funds, Intensive Care Unit, The Queen Elizabeth Hospital. Copyright © 2005 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/01.CCM.0000150960.36228.C0 Objective: Nutritional support as enteral or parenteral nutrition (PN) is used in hospitalized patients to reduce catabolism. This study compares outcomes of early enteral nutrition (EN) with early PN in hospitalized patients. Design: The authors conducted a metaanalysis of randomized, controlled trials (RCT) comparing early EN with PN. Studies on immunonutrition were excluded. Studies were categorized as medical, surgical, or trauma. Patients: RCTs of early EN/PN were identified by search of 1) MEDLINE (1966 –2002), 2) published abstracts from scientific meetings, and 3) bibliographies of relevant articles. Measurements and Main Results: Thirty RCTs (ten medical, 11 surgical, and nine trauma) compared early EN with PN. The effect of nutrition type on hospital mortality and complication rates was reported as risk difference (RD%) and hospital length of stay (LOS) as mean weighted difference (MWD days). Missing data, by outcomes, varied from 20% to 63%. As a result of heterogeneity of treatment effects, the DerSimonian-Laird random-effects esti- mator was reported. There was no differential treatment effect of nutrition type on hospital mortality for all patients (0.6%, p  .4) and subgroups. PN was associated with increases in infective complications (7.9%, p  .001), catheter-related blood stream infections (3.5%, p  .003), noninfective complications (4.9%, p  .04), and hospital LOS (1.2 days, p  .004). There was no effect of nutrition type on technical complications (4.1%, p  .2). EN was associated with a significant increase in diarrheal episodes (8.7%, p  .001). Publication bias was not demonstrated. Meta- analytic regression analysis did not demonstrate any effect of age, time to initiate treatment, and average albumin on mortality estimates. Cumulative metaanalysis showed no change in the mortality estimates with time. Conclusion: There was no mortality effect with the type of nutritional supplementation. Although early EN significantly re- duced complication rates, this needs to be interpreted in the light of missing data and heterogeneity. The enthusiasm that early EN, as compared with early PN, would reduce mortality appears misplaced. (Crit Care Med 2005; 33:213–220) KEY WORDS: metaanalysis; nutrition; hospital mortality; random effects; publication bias; metaregression; trim and fill 213 Crit Care Med 2005 Vol. 33, No. 1