Malnutrition is frequently seen in critically ill patients. As long as there is a functional gastrointestinal tract, enteral nutrition is preferred over parenteral nutrition due to lower mucosal atrophy and decreased translocation of bacteria and toxins [118]. Several delivery methods are currently available for a variety of patients’ conditions.

When supplemental feeding is expected to be needed for <30 days, nasogastric or nasoenteric tubes are the preferred delivery method. A nasogastric tube (NGT) is easy to place and can be performed bedside, but is associated with a high incidence of displacement, increased risk of aspiration, misplacement into the airway, and direct injury to the nasopharynx or oropharynx [119]. It is typically placed blindly at the bedside with radiologic verification of proper placement prior to the initiation of feeds. Despite its ease in placement, controversy persists as to the optimal site for enteral nutrition delivery (gastric versus small intestinal). Unlike the small intestine, the stomach commonly exhibits an ileus following surgery, major trauma, and during other critical illnesses. Ritz et al. demonstrated that 45% of mechanically ventilated patients showed delayed gastric emptying impeding adequate delivery of gastric enteral nutrition [120].

In order to avoid this, many clinicians advocate post-­pyloric feeding. However, randomized, controlled trials comparing gastric to post-pyloric feeding have produced varying results [121–127]. A possible explanation for this is that most post-pyloric feeding tubes are too short to go beyond the ligament of Treitz. Thus, enteral nutrition is being administered into the duodenum and studies have shown a high incidence of duodenogastric reflux in patients at risk for aspiration [122]. Heyland et al. documented an 80% rate of radioisotope-labeled reflux into the stomach, 25% into the esophagus, and 4% into the lung when radioisotope-labeled enteral formulas were fed through post-pyloric feeding tubes in mechanically ventilated ICU patients [125]. In postoperative patients, Tournadre et al. demonstrated gastroparesis and rapid discoordinated duodenal contractions with some 20% migrating in a retrograde fashion [128]. These studies provide compelling evidence that duodenogastric reflux is present in postoperative and critically ill patients. Thus, feeding into the duodenum is not significantly different than feeding into the stomach in these patients with regard to the aspiration risk.

The Canadian Critical Care Clinical Practice Guidelines Committee likewise studied this topic. They evaluated 11 Level 2 studies comparing enteral nutrition via the small intestine with gastric enteral nutrition [129]. The studies that reported nutritional delivery demonstrated more rapid and successful delivery in those patients fed via the small intestine. Nine studies documented infectious complications. When these were aggregated, small intestinal enteral nutrition was associated with a significant decrease in infectious complications (RR, 0.77; 95% CI, 0.60–1.00; p  =  0.05). In respect to mortality, there was no difference between the two routes of delivery (RR, 0.93; 95% CI, 0.72–1.20; p  =  0.6). Overall, the committee recommended small intestinal enteral nutrition when obtaining small bowel access was feasible. If this placement is unsuccessful blindly, a naso-jejunal feeding tube may be placed with endoscopic guidance.

A gastric tube has become the best alternative for enteral feeding in patients requiring nutrition for longer than 30 days. The major indication for gastric tube placement is the inability to maintain adequate oral nutritional intake as a result of a recent cerebrovascular accident, altered level of consciousness, dysphagia, oro-pharyngo-esophageal obstructing lesions, facial or esophageal trauma, or tracheal–esophageal fistula [130, 131].

Irrespective of the technique, gastrostomy tubes should not be placed in patients with rapidly progressing incurable diseases or those with a high likelihood of a rapid recovery [132]. Among other contraindications are signs of acute infection, peritonitis, coagulopathy, ulceration at future puncture site, and distal obstruction [133]. With percutaneous endoscopic gastrostomy (PEG), there are certain technique-specific contraindications that must be kept in mind. An absolute contraindication to PEG placement is any condition that prevents approximation of the stomach and anterior abdominal wall. Relative contraindications are obesity and hepatomegaly, which may prevent transillumination across the abdominal wall, as well as neoplastic and inflammatory diseases of the abdominal wall [131].

The primary techniques for gastrostomy tube placement are surgical and percutaneous. Surgical gastrostomies are performed in the operating room and require general anesthesia. With the introduction of PEGs, the surgical technique is now relegated to cases where the percutaneous option is unavailable, contraindicated, or unsuccessful. PEG tubes are the most common procedure performed in establishing an enteral route for nutrition due to their lower cost and faster recovery time [132]. They are often performed bedside utilizing either the “pull” or “push” method. The “pull” technique is more widely used due to a lower complication rate and cost [134]. That being said, more recent data documents the “push” technique to be safe with a lower insertion site infection rate [135–137].

During PEG placement via the “pull” technique, the anterior abdominal wall insertion site must meet certain criteria: (1) adequate transillumination via the endoscope within the stomach, and (2) appreciation of a distinct finger ballottement inside the stomach via the abdominal wall. The Seldinger technique is then utilized in inserting the guidewire into the stomach. It is snared and pulled retrograde through patient’s mouth at which point the gastrostomy tube is attached to the guidewire. The guidewire (gastrostomy tube in tow) is then pulled into the stomach and through the anterior abdominal wall, where it is affixed to the abdominal wall [138].

The “push” technique starts with identification of the insertion site via transillumination with the endoscope. Next, a specially designed gastropexy device is used to pexy the stomach to the anterior abdominal wall. A trocar with an introducer sheath is then inserted through the anterior abdominal wall and into the stomach under endoscopic visualization. The trocar is withdrawn while the tip of the sheath remains inside the stomach and the gastric tube is inserted through the introducer sheath. The gastrostomy tube balloon is inflated and the peel away sheath is removed [135, 138]. The “push” technique should be considered in patients with a decreased oropharyngeal diameter that is insufficient for the gastrostomy tube’s bumper (8–9 mm in diameter) to pass through if the “pull” technique were utilized. In such circumstances, the “pull” technique can still be used, but will often require dilatation with a bougie, which is associated with an added risk of perforation [136].

Procedure-related mortality for PEG tube placement is minimal, with a 30-day overall mortality rate of 19%. This has increased from 8% in the 1990s due to the significantly sicker patient population referred for PEGs [133]. PEG tube placement complications include PEG site leakage, PEG tube blockage or dislodgement, peri-stomal infection, necrotizing fasciitis, gastro-colic fistula, and peritonitis. The latter is most frequently secondary to separation of the stomach and anterior abdominal wall or injury to adjacent visceral organs [133]. The most common complication is peri-stomal infection with a reported rate of 5–30% [135, 139, 140].

A recent Cochrane analysis recommends administration of preoperative antibiotics to all patients undergoing “pull” gastrostomies [130]. Both penicillin-based and cephalosporin antibiotics have been used with significantly improved infection rates. Kulling et al. documented a ­favorable cost analysis for prophylactic antibiotics in “pull” gastrostomies; however, the antibiotics used as the comparison were not methicillin-resistant Staphylococcus aureus (MRSA) sensitive [141]. There are several other peri-stomal infection prevention strategies that proven useful. Radhakrishnan et al. showed that a combination of preoperative intravenous antibiotics with antiseptic spray decreased peri-stomal infection rate in comparison to antibiotic or antiseptic spray alone [142]. Another study showed that a single dose of trimethoprim/sulfamethoxazole per tube postoperatively had an equivalent peri-stomal infection rate to the group who received preoperative antibiotics [143]. MRSA-positive patients have a higher peri-stomal infection rate [139]. In an attempt to address this problem, Horiuchi et al. revealed that decolonization of MRSA positive patients with intranasal antibiotics significantly decreased the peri-stomal infection rate [144].

While it is clear that preoperative antibiotics should be administered to all “pull” gastrostomies, the same recommendation cannot be made for the “push” technique [130]. Shastri et al. recently reported an equivalent 7-day infection rate for “push” gastrostomy with and without prophylactic antibiotics [135]. As mentioned earlier, some data document a significantly lower infection rate with the “push” technique in comparison to the “pull” technique [137]. One of the theories is that during the “pull” technique, there is seeding of the gastrostomy tube as it passes through the heavily colonized oropharynx.

Gastrointestinal bleeding is frequently seen in ICU patients and endoscopy has become the major diagnostic and therapeutic tool. Upper gastrointestinal bleeding is defined as bleeding proximal to the ligament of Treitz and can be broken down into two major categories: variceal and non-variceal. The major cause of non-variceal bleeding is peptic ulcer disease with an incidence of 35–50% [145]. Another cause is transpapillary hemorrhage, which could be due to either hemobilia or pancreatic duct hemorrhage as a result of trauma, pancreatitis, malignancy, or arteriovenous malformation [145]. Variceal bleeding is one of the complications of portal hypertension and is associated with a mortality rate of 40% with an early re-bleed rate of 30–50% [146].

Characterization of upper gastrointestinal bleeding is best performed with endoscopy. In preparation for endoscopy, erythromycin has been shown by meta-analysis to be of benefit for emptying the stomach to allow better visualization of the bleeding area [147]. However, the International Consensus on Non-variceal Upper Gastrointestinal Bleeding recommends administration of promotility agents only to those patients suspected to have a large volume of blood in the stomach or those who have recently eaten [148].

While the general endoscopic technique for the management of variceal and non-variceal hemorrhage is similar, the medical treatment varies some. Meta-analyses have documented that octreotide, a somatostatin analog capable of decreasing portal hypertension, is comparable to sclerotherapy for control of variceal bleeding with fewer side-effects [149, 150]. Sclerotherapy involves endoscopic identification of the bleeding area and injection with sclerosing agents. Endoscopic variceal ligation, which involves placement of several bands over the varix, has previously been shown to be superior to 48-h somatostatin infusion in controlling acute variceal bleeding without increased complications [151]. However, more recent studies document that a combined treatment with endoscopic variceal ligation and somatostatin significantly improves initial control of variceal bleeding [151–153]. For non-variceal bleeding successfully placed hemoclips were equivalent to thermocoagulation and had significantly decreased re-bleeding rates and need for operation in comparison with injection of sclerosing agents alone [154].

If during endoscopic evaluation a clot is seen, attempts should be made to remove it in order to better visualize the area of concern. There is a controversy for appropriate treatment when the clot is not dislodged with irrigation. While some studies document that the incidence of re-bleeding from adherent clots is low, some authors recommend removal of the clot with a cold guillotine snare technique after pre-injecting the clot with epinephrine, especially in high risk patients [148]. If the bleeding cannot be controlled endoscopically, a clip can be left in the area of hemorrhage to serve as a guide for embolization by interventional radiology [155]. Transarterial embolization is considered equivalent to surgery in the case of failed endoscopic treatment with success rates of 60–90% [145, 148].

If endoscopy demonstrates bleeding esophageal varices, but the hemorrhage is too severe to manage endoscopically, balloon tamponade can be performed with a Sengstaken–Blakemore tube or one of its modifications. When properly placed, balloon tamponade has been reported to effectively control hemorrhage in 80–94% of patients; however, proper placement and maintenance are difficult and patients require close monitoring [156]. The tube is perfect for controlling variceal bleeding at the gastroesophageal junction, but will not be effective in distal gastric or duodenal bleeding.

In both variceal and non-variceal bleeding, recurrent hemorrhage is associated with increased mortality [157, 158]. Several factors have been identified that predict re-bleeding including hemodynamic instability, active bleeding (especially “spurting”) on endoscopy, an ulcer size >2 cm, and a posterior duodenal wall bleed [159]. While routine repeat endoscopy is not warranted in all patients, if a combination of these factors is present, repeat endoscopy may be beneficial. Intravenous proton pump inhibitors are usually started prior to endoscopy in both variceal and non-variceal bleeding, which has been shown to decrease the necessity for endoscopic treatment and in some studies decreased the rate of re-bleeding [148, 160, 161]. In acute variceal bleeding, the addition of isosorbide mononitrate alone or in combination with beta-blockers to endoscopic variceal ligation did not affect the re-bleeding rate [162]. However, another meta-analysis definitively showed that a combination of beta-blockers and endoscopic variceal ligation significantly decreases re-bleeding and mortality rates in comparison to endoscopic therapy alone [163].

Lower gastrointestinal bleeding is defined as bleeding distal to the ligament of Treitz and accounts for 20% of all major gastrointestinal bleeds, diverticular disease representing the most common cause. Other causes of lower gastrointestinal bleeding in the ICU include inflammatory disease, vascular ectasia, post-polypectomy bleeding, malignancy, and anorectal disease [164–166]. Most diverticular bleeding (80%) will cease spontaneously; however, right-sided diverticular bleeding is often more severe and is more likely to require surgical intervention [167–170].