The prevalence of disease-related malnutrition in Western European hospitals is estimated to be about 30%. There is no consensus whether poor nutritional status causes poorer clinical outcome or if it is merely associated with it. The intention with all forms of nutrition support is to increase uptake of essential nutrients and improve clinical outcome. Previous reviews have shown conflicting results with regard to the effects of nutrition support.
To assess the benefits and harms of nutrition support versus no intervention, treatment as usual, or placebo in hospitalised adults at nutritional risk.
We searched Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library, MEDLINE (Ovid SP), Embase (Ovid SP), LILACS (BIREME), and Science Citation Index Expanded (Web of Science). We also searched the World Health Organization International Clinical Trials Registry Platform (www.who.int/ictrp); ClinicalTrials.gov; Turning Research Into Practice (TRIP); Google Scholar; and BIOSIS, as well as relevant bibliographies of review articles and personal files. All searches are current to February 2016.
We include randomised clinical trials, irrespective of publication type, publication date, and language, comparing nutrition support versus control in hospitalised adults at nutritional risk. We exclude trials assessing non-standard nutrition support.
We used standard methodological procedures expected by Cochrane and the Cochrane Hepato-Biliary Group. We used trial domains to assess the risks of systematic error (bias). We conducted Trial Sequential Analyses to control for the risks of random errors. We considered a P value of 0.025 or less as statistically significant. We used GRADE methodology. Our primary outcomes were all-cause mortality, serious adverse events, and health-related quality of life.
We included 244 randomised clinical trials with 28,619 participants that met our inclusion criteria. We considered all trials to be at high risk of bias. Two trials accounted for one-third of all included participants. The included participants were heterogenous with regard to disease (20 different medical specialties). The experimental interventions were parenteral nutrition (86 trials); enteral nutrition (tube-feeding) (80 trials); oral nutrition support (55 trials); mixed experimental intervention (12 trials); general nutrition support (9 trials); and fortified food (2 trials). The control interventions were treatment as usual (122 trials); no intervention (107 trials); and placebo (15 trials). In 204/244 trials, the intervention lasted three days or more.We found no evidence of a difference between nutrition support and control for short-term mortality (end of intervention). The absolute risk was 8.3% across the control groups compared with 7.8% (7.1% to 8.5%) in the intervention groups, based on the risk ratio (RR) of 0.94 (95% confidence interval (CI) 0.86 to 1.03, P = 0.16, 21,758 participants, 114 trials, low quality of evidence). We found no evidence of a difference between nutrition support and control for long-term mortality (maximum follow-up). The absolute risk was 13.2% in the control group compared with 12.2% (11.6% to 13%) following nutritional interventions based on a RR of 0.93 (95% CI 0.88 to 0.99, P = 0.03, 23,170 participants, 127 trials, low quality of evidence). Trial Sequential Analysis showed we only had enough information to assess a risk ratio reduction of approximately 10% or more. A risk ratio reduction of 10% or more could be rejected.We found no evidence of a difference between nutrition support and control for short-term serious adverse events. The absolute risk was 9.9% in the control groups versus 9.2% (8.5% to 10%), with nutrition based on the RR of 0.93 (95% CI 0.86 to 1.01, P = 0.07, 22,087 participants, 123 trials, low quality of evidence). At long-term follow-up, the reduction in the risk of serious adverse events was 1.5%, from 15.2% in control groups to 13.8% (12.9% to 14.7%) following nutritional support (RR 0.91, 95% CI 0.85 to 0.97, P = 0.004, 23,413 participants, 137 trials, low quality of evidence). However, the Trial Sequential Analysis showed we only had enough information to assess a risk ratio reduction of approximately 10% or more. A risk ratio reduction of 10% or more could be rejected.Trial Sequential Analysis of enteral nutrition alone showed that enteral nutrition might reduce serious adverse events at maximum follow-up in people with different diseases. We could find no beneficial effect of oral nutrition support or parenteral nutrition support on all-cause mortality and serious adverse events in any subgroup.Only 16 trials assessed health-related quality of life. We performed a meta-analysis of two trials reporting EuroQoL utility score at long-term follow-up and found very low quality of evidence for effects of nutritional support on quality of life (mean difference (MD) -0.01, 95% CI -0.03 to 0.01; 3961 participants, two trials). Trial Sequential Analyses showed that we did not have enough information to confirm or reject clinically relevant intervention effects on quality of life.Nutrition support may increase weight at short-term follow-up (MD 1.32 kg, 95% CI 0.65 to 2.00, 5445 participants, 68 trials, very low quality of evidence).
There is low-quality evidence for the effects of nutrition support on mortality and serious adverse events. Based on the results of our review, it does not appear to lead to a risk ratio reduction of approximately 10% or more in either all-cause mortality or serious adverse events at short-term and long-term follow-up.There is very low-quality evidence for an increase in weight with nutrition support at the end of treatment in hospitalised adults determined to be at nutritional risk. The effects of nutrition support on all remaining outcomes are unclear.Despite the clinically heterogenous population and the high risk of bias of all included trials, our analyses showed limited signs of statistical heterogeneity. Further trials may be warranted, assessing enteral nutrition (tube-feeding) for different patient groups. Future trials ought to be conducted with low risks of systematic errors and low risks of random errors, and they also ought to assess health-related quality of life.