Title: Commentary on “Guidelines for the provision of nutrition support therapy in the adult critically ill patient: The American Society for Parenteral and Enteral Nutrition”
Abstract: As authors of recent meta-analyses evaluating lipids for parenteral nutrition1, 2 and/or the European Society for Clinical Nutrition and Metabolism Expert Group: Lipids in the intensive care unit (ICU),3 we welcome the updating of the American Society for Parenteral and Enteral Nutrition guidelines for clinical nutrition in adult critically ill patients,4 but would like to remark on certain aspects that cause us concern. Our main areas of concern are a lack of transparency in reporting the methodology and the inconsistencies and errors in the publication. Very few studies seem to be included in comparison with other recent meta-analyses on similar topics,1, 2, 5 but the reasons for this are unclear. The technical reporting of the guideline development process should be fully detailed and follow best practices.6 Because of the unclear screening process, it is impossible to duplicate or understand the authors' methodology; for example, Metry et al7 was included despite not being indexed in PubMed or Embase, nor having a statement about manual searching of the literature. We are also concerned by the search string itself, which does not reflect all of the research questions asked or the recommendations made. For example, it includes enteral tube feeding, proteins, probiotics, and antioxidants, but no key words relating to lipids. We would have particularly welcomed a flow chart giving reasons for excluding certain randomized controlled trials captured by the search strategy,6 and we were puzzled by why some studies were included for selected outcomes but not for others also reported and included in the recommendations. For question 5B, for example, why were8-10 excluded from the meta-analysis of intensive care unit (ICU) length of stay, yet included for other parameters such as infections and/or mortality? Furthermore, different methods were used to report different recommendations: for question 5A, the authors gave recommendations based on tabulated results and a text summary (but no meta-analysis), whereas meta-analyses were used as the basis for all other recommendations. If a meta-analysis was not performed for one question, then the reason for this should be detailed.11 Although errors inevitably occur, to prevent them it is customary for two or more researchers to work independently to extract data, with a final data-checking step.6, 12, 13 Despite extensive changes being made between the online and print versions, including the replacement of 11 of 25 figures and changes to 5 of 8 tables, major errors remain. For example, hospital mortality data in figure 24 for Grau Carmona et al are wrong, as ICU deaths are missing, resulting in the omission of 42 of 54 deaths during hospitalization.10 Finally, we would like to comment on more subjective aspects, such as the "slicing" of outcomes and lack of comparison with published literature. Although the decision to split outcomes into small and selected groups (eg, pneumonia and catheter-related infections considered separately rather than in aggregate) is not necessarily right or wrong, it reduces the likelihood of statistically significant results by considering fewer events. As decisions on whether to combine data are often complex and somewhat subjective, it is of utmost importance that the authors describe their rationale so that readers can understand their approach. Furthermore, there was lack of discussion about the update's conclusions in comparison with other systematic reviews/meta-analyses and previous guidelines, and where conclusions differed, reasons for this (eg, different eligibility criteria, search methods, or data synthesis approaches), as is customary when producing guidelines, were not provided.14 Lorenzo Pradelli is a Director and employee of AdRes, which has received project funding from Fresenius Kabi. Philip C. Calder has received research funding from B Braun and has undertaken consultant/advisory work for Fresenius Kabi. Nicolaas E. Deutz declares no conflicts of interests, but discloses that he is a coinventor of several patents (owned by others), has served on scientific advisory boards for Novartis and Baxter, and has been a consultant for Abbott Nutrition, Ajinomoto, OCERA, and VitaNext. Maurizio Muscaritoli has received speaker's fees from Fresenius Kabi. Michael Adolph, Teodoro Grau Carmona, Adina T. Michael-Titus, and Pierre Singer declare no funding, financial relationships, or conflicts of interest. These authors declare no conflict of interest: Michael Adolph, Teodoro Grau Carmona, Adina T. Michael-Titus, Pierre Singer.