Enteral Nutrition (EN) is considered by all medical societies to be the preferred route of providing nutrition to critically ill patients, including those who are intubated, and should be ideally be initiated within 48 h of hospitalization [1–7]. Enteric feeding has been associated with significant reduction in 28-day mortality, while being more accessible and less expensive than parenteral nutrition [3, 8, 9]. Early enteric feeding with a lipid and protein-rich formula has an anabolic effect essential to the healing process in critical illness and in preservation of the gut mucosa, maintaining gut associated lymphoid tissue, and promoting protection via the gut’s natural flora [10–12]. Parenteral nutrition (PN) has been associated with a higher degree of infection, especially nosocomial infection, and mortality than EN, particularly among patients with higher severity of critical illness as assessed by Acute Physiologic Assessment and Chronic Health Evaluation (APACHE II) score. The Impact of Early Parenteral Nutrition Completing Enteral Nutrition in Adult Critically Ill Patients trial (EPaNIC,) a 2011 RCT involving 4640 ICU patients evaluated early vs. late initiation of PN [11]. It demonstrated reduced rates of early alive discharge from the ICU and hospital with supplemented PN than with EN alone. Additionally, the earlier PN is initiated, the worse the outcome [3, 6]. With PN the patient does not benefit from stimulating the gut and endures complications associated with infusing a high glucose concentration and fat globules directly into the venous system [13]. A meta-analysis of PN compared with no nutrition found an almost doubled risk of dying in the PN group [14]. In the circumstance that early EN is impossible, nutrition should be withheld for up to 7 days, unless the patient has evidence of protein-calorie malnutrition [6]. This has become a Critical Care recommendation of the Choosing Wisely campaign. If the decision is made to provide PN, consider discontinuation once EN becomes feasible or when markers of nutritional status improve.

The Surviving Sepsis Campaign [15] advises that, if a septic patient cannot tolerate full caloric needs through EN, trophic feeds are acceptable for up to 6 days, and PN should not be initiated. This assumes the patient is not malnourished prior to ICU admission. Trophic feeds of at least 25 % of daily nutritional needs may have a protective effect on the bowel [2, 10, 16–18]. The EDEN trial, a multi-center RCT evaluating 1000 patients with acute lung injury, found no significant difference in mortality between trophic and full feedings, though it is important to note that all patients in the study were well-nourished, with a mean BMI of 30 [18]. Moderate obesity is shown to be protective during critical illness; however, increased protein supplementation does not appear to limit muscle wasting or loss of lean body mass [19–21].

Small enteral feeds are enough to cause enzyme secretion from the brush border and preserve gut epithelium and structure, thus preventing increased permeability [3]. Feeding also preserves commensal bacteria which further contribute to gut integrity by stimulating mucous production, while allowing competitive inhibition of pathogens and interfering with expression of virulence factors on pathogenic bacteria [3]. Enteric feeds have an anti-inflammatory effect on the gut via the gut-brain axis, and as such, should be considered a therapeutic modality amongst conditions of gastrointestinal (GI) inflammation such as pancreatitis or colitis [12].

Despite all of the benefits of early enteric feedings, this goal is not always accomplished. One major barrier to appropriate early nutrition is the practice of holding feedings for gastric residuals or decreased bowel sounds [6]. Patients in the ICU are often too ill to have swallowed enough air to allow for bowel sounds. Multiple studies have shown that gastric emptying and reflux has no correlation with incidence of ventilator-associated pneumonia (VAP) [22–24]. Aspiration pneumonia is likely more resultant of oropharyngeal/subglottic secretions with pathogenic microbes than reflux of gastric contents [23, 24]. If gastric residuals are checked, there appears to be no worsening of outcome with gastric residuals up to 500 mL [24]. Absence of gastric residual monitoring may not be inferior to monitoring residuals in terms of the likelihood of nosocomial pneumonia [23]. Therefore, in the absence of a known GI dysmotility, orogastric or nasogastric feeding should be initiated. If EN is poorly tolerated based on residuals or coughing, prokinetic agents such as metoclopramide, or erythromycin, can be started [25]. If these practices are still unsuccessful, post-pyloric feeds could be considered, with the understanding that the patient-important outcomes, such as VAP, emesis, aspiration, diarrhea, and mortality may not change [25].

In the CALORIES Trial [26], 2400 patients in 33 ICUs in England were randomized to either EN or PN for up to 5 days. The authors hypothesized that more patients in the PN arm would be able to achieve nutritional goals earlier, leading to a relative risk reduction in mortality. Although there was no difference in 30 day mortality between groups (33 % vs. 34 %) the patients in the PN group did not get more calories than those in the EN group in this pragmatic trial. In addition, although there was no increase in rate of infection in the PN group, this is likely because infection rate in PN is dose-dependent [1–3, 5]. Pragmatic trials are important in demonstrating the challenges of protocol implementation but are not explanatory in nature and the superiority of one method over another cannot be concluded from this trial.

The actual caloric goals in critical illness have not yet been defined. Most are estimated through calculations based on characteristics of the patient prior to illness onset and illness-severity scores [27]. A table of commonly used predictive equations can be found here [28]. Indirect calorimetry is suggested to be the gold standard in measuring resting energy expenditure [29], though its standard and frequent use is often impractical [30]. There can be significant differences between measured and predicted energy expenditures [28], though the effect on patient outcomes is not well defined.