Study overview

The present study is an analysis of a subset of a large international point prevalence survey of nutrition practice in intensive care units (ICUs) (www.criticalcarenutrition.com/ins) conducted in 2007, 2008, 2009, 2011, 2013 and 2014. Details of the survey are provided in the Additional File and elsewhere [13]. In total, 21,100 adult patients from 785 ICUs had been included into the survey.

Using a secure web-based data collection tool, the following information was collected: date of ICU admission, admission category (elective surgery, emergency surgery, medical), primary admission diagnosis (nine categories), sex, age, body mass index (BMI), duration of mechanical ventilation/propofol therapy, and APACHE II score on admission day. Treating physicians recorded daily the amount of calories, and type (enteral, parenteral) and amount of amino acids or protein received from parenteral nutrition (PN) and/or enteral nutrition (EN). Daily protein intake was collected from the day of ICU admission (partial day) to a maximum of additional 11 days after admission date. In the current analysis, we ignored protein intake received on the day of ICU admission, and referred to the subsequent discrete calendar days as “day on diet #1 to #11”.

For the first three days on diet, we recorded the number of days on which a patient had, at any time, been mechanically ventilated, or had received PN or oral feeding. Patients were followed for a maximum of 60 days after ICU admission. We registered time until in-hospital death or live hospital discharge. Patients remaining alive in hospital for more than 60 days were considered censored for either risk at that time.

A patient’s continuous survival time or time until live hospital discharge was calculated as “days after ICU admission”, where a “day” was defined as a 24-h period starting on the exact date and time of ICU admission. Consequently, days on diet, and days after ICU admission differed to the extent that the former always started/ended at 12 a.m. (midnight), whereas the latter could start/end at any time of the previous calendar day.