Projects per year
Introduction: Weight loss in cancer patients is a worrisome constitutional change predicting disease progression and shortened survival time. A logical approach to counter some of the weight loss is to provide nutritional support, administered through enteral nutrition (EN) or parenteral nutrition (PN). The aim of this paper was to update the original systematic review and meta-analysis previously published by Chow et al., while also assessing publication quality and effect of randomized controlled trials (RCTs) on the meta-conclusion over time. Methods: A literature search was carried out; screening was conducted for RCTs published in January 2015 up until December 2018. The primary endpoints were the percentage of patients achieving no infection and no nutrition support complications. Secondary endpoints included proportion of patients achieving no major complications and no mortality. Review Manager (RevMan 5.3) by Cochrane IMS and Comprehensive Meta-Analysis (version 3) by Biostat were used for meta-analyses of endpoints and assessment of publication quality. Results: An additional seven studies were identified since our prior publication, leading to 43 papers included in our review. The results echo those previously published; EN and PN are equivalent in all endpoints except for infection. Subgroup analyses of studies only containing adults indicate identical risks across all endpoints. Cumulative meta-analysis suggests that meta-conclusions have remained the same since the beginning of publication time for all endpoints except for the endpoint of infection, which changed from not favoring to favoring EN after studies published in 1997. There was low risk of bias, as determined by assessment tool and visual inspection of funnel plots. Conclusions: The results support the current European Society of Clinical Nutrition and Metabolism guidelines recommending enteral over parenteral nutrition, when oral nutrition is inadequate, in adult patients. Further studies comparing EN and PN for these critical endpoints appear unnecessary, given the lack of change in meta-conclusion and low publication bias over the past decades.