Marjon Borgert

95 Potential risk factors in the delivery of enteral nutrition Analysis Descriptive statistics were used to summarize the baseline patient and feeding characteristics. Continuous variables that were normally distributed were expressed as means with standard deviations and not normally distributed variables as medians and inter-quartile ranges (IQR). To test two independent groups of not normally distributed continuous variables, the Mann-Whitney U test was used. Categorical variables were expressed as percentages, numerators and denominators and were compared with the Chi-square test or Fisher’s exact test. Statistical uncertainty was expressed by 95% con dence intervals as appropriate, and statistical signi cance was de ned at a P value of < 0.05. Adequate EN intake is de ned as the real EN intake at least equal to the ideal EN volume/calorie/protein intake or more. Ideal calorie intake is calculated as 25Kcal/ kg last known body weight before hospital admission/day 7 , success of calorie intake as percentage of ideal realized calorie intake. Ideal protein intake is calculated as 1.2 to 1.5 g/kg/last known body weight before hospital admission/day 7,15 , likewise success of protein intake as percentage of ideal realized protein intake. Patients with a body mass index of > 27-30 were excluded for analysis. To describe the factors of success or failure of feeding intake we used a multilinear regression model. We calculated the total feeding intake during ICU stay in terms of calorie as well as protein intake. We divided this number by the total duration of ICU stay (in hours) as a single outcome measure for overall success of feeding. In the multilinear regression model we included all patients that have been enterally fed during their ICU stay. As candidate predictors we selected age, gender, apache II score, type of admission (medical/surgical) planned admission (yes/no), mechanical ventilation (yes/no), type of feeding tube (nasoduodenal, nasojejunal, percutaneous endoscopic gastrostomy (PEG), naso-gastric tubes). The goal of the analysis by this prediction model was to nd the most valid subset of available predictors and the corresponding best tting regression model for describing the relationship between average EN intake of protein and calories during ICU stay and the predictors. A multiple linear regression model was used with forward selection (by hand) of predictors. Analyses were performed using R (version: 3.3.1; R Foundation for Statistical Computing, Vienna, Austria). Ethical statement The study was approved by the Medical Ethics Committee of the Academic Medical Center of Amsterdam, the Netherlands and the need for informed consent was waived.