HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:
Guest Access | Sign In via User Name/Password

This Article Abstract Full Text (PDF) Free Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Article Archive Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (46) Citing Articles via Google Scholar Google Scholar Articles by Mutlu, G. M. Articles by Factor, P. Search for Related Content PubMed PubMed Citation Articles by Mutlu, G. M. Articles by Factor, P.

GI Complications in Patients Receiving Mechanical Ventilation^*

^* From the Section of Respiratory and Critical Care Medicine (Dr. G. Mutlu), University of Illinois at Chicago, Chicago; Division of Digestive Diseases and Hepatology (Dr. E. Mutlu), Rush-Presbyterian St. Luke’s Medical Center, Chicago; Division of Pulmonary and Critical Care Medicine (Dr. Factor), Evanston Northwestern Healthcare and Northwestern University Medical School, Evanston, IL.

Correspondence to: Phillip Factor, DO, FCCP, Pulmonary and Critical Care Medicine, Evanston Northwestern Healthcare, 2650 Ridge Ave, Evanston, IL 60201; e-mail: pfactor{at}northwestern.edu

	Abstract

TOP Abstract Introduction Interactions Between MV and... GI Hemorrhage Nonhemorrhagic Complications Summary References

 
Mechanical ventilation (MV) can be lifesaving by maintaining gas exchange until the underlying disorders are corrected, but it is associated with numerous organ-system complications, which can significantly affect the outcome of critically ill patients. Like other organ systems, GI complications may be directly attributable to MV, but most are a reflection of the severity of the underlying disease that required intensive care. The interactions of the underlying critical illness and MV with the GI tract are complex and can manifest in a variety of clinical pictures. Incorporated in this review are discussions of the most prevalent GI complications associated with MV, and current diagnosis and management of these problems.

Key Words: complications • GI • mechanical ventilation • review • stress ulcer

	Introduction

TOP Abstract Introduction Interactions Between MV and... GI Hemorrhage Nonhemorrhagic Complications Summary References

 
Mechanical ventilation (MV) is a lifesaving therapy with a myriad of organ-system complications that can significantly affect the outcome of critically ill patients.¹ Like other organ systems, GI complications may be directly attributable to MV, but most are a reflection of the disease process that required intensive care. Incorporated in this review are discussions of the most prevalent GI complications associated with MV, such as stress ulcer^{2 3 4} and GI hypomotility (Table 1 ).^{5 6} The current diagnosis and management of these conditions and other less common complications are also included.

	Interactions Between MV and Critical Illness

TOP Abstract Introduction Interactions Between MV and... GI Hemorrhage Nonhemorrhagic Complications Summary References

View larger version (21K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Proposed mechanisms for the development of GI complications during MV. MV can contribute to the pathogenesis of GI problems in much the same way as critical illness by affecting splanchnic blood flow and leading to increased release of proinflammatory mediators. SIRS = systemic inflammatory response syndrome; TNF- = tumor necrosis factor-.

MV with PEEP is also associated with increased plasma-renin-angiotensin–aldosterone activity and elevated catecholamines because of sympathetic activation.^{13 14 15} Moreover, these patients frequently receive catecholamine therapy for BP support. These neurohormonal alterations can contribute significantly to splanchnic hypoperfusion by leading to vasoconstriction and redistribution of blood away from the splanchnic vascular bed.^{16 17} Whether because of decreased cardiac output and/or increased vascular resistance, splanchnic hypoperfusion produces an imbalance between oxygen supply and demand (a relative oxygen shortage) that may contribute to the development of GI complications, such as mucosal damage (eg, stress ulcer) and/or altered GI motility (eg, ileus).^{16 17 18} Perhaps more concerning than splanchnic hypoperfusion itself is reperfusion injury and further damage to GI epithelial cells that may occur after restoration of blood flow after prolonged periods of hypoperfusion.¹⁹ Repetitive episodes of hypoperfusion followed by reperfusion may be responsible for acute nonocclusive mesenteric ischemia in the critical-care setting.²⁰

Recent advances in our understanding of the adverse effects of MV suggest an important role for cytokines in the pathogenesis of GI complications. Cytokines (eg, tumor necrosis factor-, interleukin [IL]-1, and IL-8) are inflammatory mediators that can affect many organs and elicit a variety of physiologic and biochemical responses to critical illness.²¹ They cause a series of intracellular signaling events via highly specific cell surface receptors that typically result in elaboration of other cytokines within the target cell. If these processes are not attenuated, excessive amplification of the inflammatory cascade and overproduction of proinflammatory mediators can occur with the consequent uncontrolled activation of the immune system.^{21 22} These processes can lead to a number of clinical sequelae in the GI tract as a part of multiple organ dysfunction syndrome (MODS).^{23 24} Cytokines may contribute to splanchnic hypoperfusion as well, and may also impair intestinal smooth muscle function.^{16 25 26 27} In animals, MV with "injurious" (large tidal volume, high end-inspiratory pressures) ventilatory strategies has been shown to cause an increase in production of pulmonary cytokines, as well as increasing alveolar capillary permeability that would increase the transfer of intrapulmonary cytokines from lungs to the systemic circulation.^{28 29 30 31} Recently, these data were confirmed in humans by Ranieri and colleagues,³² who demonstrated that injurious ventilatory strategies (high end-inspiratory pressures) are associated with greater increases in cytokine levels in both BAL and serum than a lung "protective" ventilatory strategy (low end-inspiratory pressures). Moreover, both high peak pressures as well as the absence of PEEP have been shown to increase bacterial translocation from the lung into the bloodstream in animal models of intratracheal instillation of bacteria, providing another mechanism by which MV can produce systemic manifestations.³³ Growing evidence in regards to increased cytokines during MV (particularly with injurious strategies) suggests a potentially critical role for MV in the initiation and propagation of a systemic inflammatory response that may include dysfunction and damage to the GI tract.

The potential contribution of MV to the development of GI complications is not limited to its indirect effects on the GI tract. Medications that are frequently used to facilitate MV such as opiates and sedatives, particularly benzodiazepines, may decrease GI motility and impair venous return via venodilation and/or diminution of responsiveness to vasopressor agents. Other commonly used medications that are frequently associated with GI complications in patients receiving MV include vasopressors (as discussed above), antibiotics, and additives in oral medications (eg, sorbitol).

Critical illness may promote GI complications via adverse effects on splanchnic blood flow and increased levels of proinflammatory mediators. In the last decade, alterations at the cellular level have become a focus of study as several investigators⁷ have suggested a role for altered gut barrier function in the pathogenesis of MODS. Decreased mucosal perfusion appears to play a pivotal role in intestinal mucosal injury; however, other consequences of critical illness such as malnutrition and altered intestinal microflora may also threaten GI epithelial cells. As a result of these unfavorable changes, gut barrier function can be compromised during critical illness. Gut barrier function is dependent on integrity of mucosal cells and intracellular junctions, mucus production, gut-associated lymphoid tissue, and secretory IgA production, all of which may be impaired during stressful events. Although not clearly established in humans, it is reasonable to presume that alterations in barrier function may allow the passage of proinflammatory mediators (eg, endotoxin) and possibly microorganisms from intestinal lumen to the bloodstream.¹⁹ This process can become self-sustaining if the underlying disease that initiates the cascade is not abbreviated.

As described above, the complexity of interactions between critical illness and MV on the GI tract necessitates that intensivists understand the pathogenesis of GI complications to allow appropriate management as well as prospective use of preventive practices. Conceivably, critical illness may serve as a "priming" factor that allows MV to affect the GI tract. Thus, the combination of effects of critical illness and MV may create an ideal environment for the development of these complications.

	GI Hemorrhage

TOP Abstract Introduction Interactions Between MV and... GI Hemorrhage Nonhemorrhagic Complications Summary References

 
Critically ill patients, especially those who are receiving MV, are prone to a spectrum of GI mucosal lesions that may result in GI hemorrhage. Acute respiratory failure requiring MV for > 48 h has been shown to be one of the two strongest independent risk factors for clinically important GI bleeding in the ICU.^{34 35} It is not clear, however, whether MV contributes the pathophysiology of GI bleeding or if it is simply a marker of severity of critical illness.

SRMD
Background and Clinical Significance:
SRMD is the most common cause of GI bleeding in patients receiving MV. Within a few hours of critical illness, macroscopic damage becomes evident as subepithelial petechiae progress to lesions ranging from superficial erosions to true gastric ulcers. These mucosal lesions tend to be multiple and occur predominantly in the fundus of the stomach, typically sparing the antrum.⁴ Distal (antral and duodenal) mucosal erosions and/or ulcers can also develop, although they typically appear later, tend to be deeper, and may be associated with a higher incidence of bleeding.^{3 36}

Most (74 to 100%) critically ill patients have endoscopically detectable mucosal erosions and subepithelial hemorrhage within 24 h of admission to the ICU.^{2 3 4} These lesions are generally asymptomatic and may or may not produce occult fecal blood. Symptomatic lesions have a wide spectrum of clinical presentations including occult, overt, or clinically significant bleeding. Overt bleeding includes frank hemorrhage, which is generally easy to detect based on the appearance of hematemesis, melena, coffee-ground–like material in nasogastic tube aspirates, or hematochezia. Clinically significant or life-threatening bleeding is defined as bleeding that causes hemodynamic changes or necessitates transfusion.³⁷ Patients receiving MV who develop clinically significant bleeding generally do so within the first 2 weeks of their ICU stay.³⁸

Because erosions are strictly mucosal lesions and therefore involve only small vessels, clinically detectable bleeding typically does not occur unless true ulcer develops. By definition, ulcers extend beyond the mucosa and into the submucosa and muscularis propria where they can erode into larger arteries (Fig 2 ).³⁹ Overt bleeding because of SRMD occurs in up to 25% of critically ill patients who do not receive prophylactic therapy.^{4 35 40} Approximately 20% of clinically evident hemorrhages (ie, 5% of all critically ill patients) are also clinically significant, such that they cause hemodynamic changes or necessitate transfusion.³⁷ Thus, the overall incidence of clinically significant GI bleeding in patients not given prophylactic treatment for SRMD is approximately 3 to 4%, ranging from 0.6 to 5%.^{35 41 42 43} Not surprisingly, clinically significant stress ulcer bleeding is associated with increased morbidity and has been shown to increase ICU length of stay (and cost) by as much as 11 days.³⁸ Similarly, mortality has also been shown to be several-fold higher in patients who develop stress ulcer bleeding compared with those who do not; importantly, these patients generally die of their primary disease process, not GI hemorrhage.^{34 35}

View larger version (43K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Illustration of the difference between an erosion and an ulcer. An erosion is a mucosal break that does not penetrate the muscularis mucosae, whereas an ulcer does penetrate the muscularis mucosae. Reprinted with permission from Weinstein.39

View larger version (25K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Proposed mechanisms for development of stress ulceration. SRMD results from the complex interaction of multiple systems. The specific relationships depicted remain somewhat speculative. Reprinted with permission from Bresalier.44

Another preventive strategy that seems to decrease the risk of overt GI bleeding in patients receiving MV is enteral feeding.^{38 81 82} The precise mechanism of this beneficial effect is not known and is probably multifactorial. Enteral feeding may prevent SRMD by providing cytoprotection by restoration of gastric epithelial energy stores and dilutional alkalinization of gastric fluid.^{38 83 84} Because its effects on gastric pH are variable, cytoprotection remains to be a more likely explanation (however not proven) for reduction in stress ulcer bleeding. Interestingly, parenteral nutrition alone has been reported to provide stress ulcer prophylaxis comparable to standard preventive therapy.⁸⁴ While a recent randomized multicenter trial³⁸ reported that H₂-blockers offer stress ulcer prophylaxis regardless of whether the patients receive enteral nutrition, to our knowledge, there has been no direct comparison of enteral feeding and stress ulcer prophylaxis. Further studies investigating relative effectiveness of enteral nutrition vs stress ulcer prophylaxis on GI bleeding outcomes are warranted.

Concerns About Prophylactic Therapy:
Side effects ascribed to antacids, H₂-receptor antagonists, and sucralfate are uncommon and occur in < 1% of patients, particularly when administered on a short-term basis.⁷⁴ However, concerns regarding the possibility of increased risk of pneumonia because of gastric colonization have led to numerous investigations regarding the use of antacids and H₂-blockers in critically ill patients.^{85 86 87} Stress ulcer prophylaxis with antacids and/or H₂-blockers raises gastric pH and increases colonization of the stomach with Enterobacteriaceae.^{88 89} Retrograde oropharyngeal contamination by colonized gastric contents and subsequent aspiration to the lower airways have been suggested to cause nosocomial pneumonia.^{85 90} Studies^{91 92} showing that supine positioning is an independent predictor of ventilator-associated pneumonia (VAP) support the importance of gastro-oropharyngeal colonization. In view of available evidence, the Centers for Disease Control and Prevention has recommended semirecumbent positioning to prevent nosocomial pneumonia.⁹³

While there is a wealth of data implicating the stomach as a reservoir for microorganisms causing VAP, there is an ongoing controversy about the contributory roles of gastric pH and colonization and subsequent aspiration.^{94 95} Confirming the role of gastric acidity and the importance of gastro-oropharyngeal route are studies^{96 97 98} reporting lower incidences of nosocomial pneumonia in patients who received sucralfate than patients who received pH-altering drugs. Gastric colonization may be particularly important in the pathogenesis of late-onset VAP (> 4 days of MV).⁹⁷ A recent meta-analysis and a large randomized controlled trial by Cook et al^{37 42} corroborate these previous investigations by showing a "trend" toward a lower incidence of pneumonia in patients who received sucralfate compared to other prophylactic measures that alter gastric pH. Other investigators^{94 99 100 101} have challenged the importance of the gastro-oropharyngeal route. In a randomized trial of 141 patients receiving MV, Bonten et al⁹⁴ found that antacids and sucralfate had similar effects on gastric acidity, colonization rates, and incidence of VAP. High gastric pH influenced colonization of the stomach but not of the upper respiratory tract or the incidence of VAP. When all studies that evaluated the sequential colonization from the stomach to the trachea were considered, gastric colonization preceded tracheal colonization in 4 to 24% and VAP in 0 to 15% of patients.¹⁰² A recent small prospective trial¹⁰³ of acidification of enteral feeding reduced the incidence of gastric colonization (2% vs 43%) but failed to show any beneficial effects on the incidence of VAP.

Although the gastro-oropharyngeal route may not be the predominant mechanism for nosocomial pneumonia, there is some evidence that supports its importance if certain precautions (eg, recumbent body position) are not undertaken. The variability of published data and the resulting controversy are most likely because of the differences in study design, measurement of gastric pH, dose of drugs administered, definition of VAP, body position (supine vs semirecumbent), gastric volume, and whether patients received simultaneous enteral feeding. More studies, particularly those controlled for the body position, are warranted to clarify the relative contribution that gastric colonization makes to the development of VAP. Until then, the risk of VAP attributable to stress ulcer prophylaxis with pH-altering drugs can be minimized if clinicians carry out preventive measures, including keeping the patient in a semirecumbent position, avoiding high gastric residuals, and administering the enteral feedings into the small bowel as opposed to the stomach. While placing the feeding tube beyond the stomach may prevent the undesirable effects of enteral feeding, such as gastric colonization and distention, it may hinder the beneficial effects of dilutional alkalinization on stress ulcer. At this time, to our knowledge, there is no study comparing the relative risks and benefits of such practice.

Currently, there is no consensus on the drug of choice for stress ulcer prophylaxis. Despite recent conflicting evidence, both pH-altering medications and sucralfate appear to effectively prevent overt stress ulcer bleeding. The choice of drug depends on the availability of enteral route for drug administration. Until parenteral proton pump inhibitors become available, H₂ antagonists remain the only option for IV use in North America. When enteral administration is feasible, both H₂ antagonists and sucralfate can be administered for prophylaxis. Increasing use of duodenal tubes limits the use of sucralfate because it needs to be administered into the stomach in order to be effective. Antacids remain an alternative, but frequent administration makes their use cumbersome. Proton pump inhibitors are reasonable options; however, they are expensive, lack well-designed controlled studies, and are therefore second-line agents. Nevertheless, it is important to mention that a small, double-blind, placebo-controlled study¹⁰⁴ showed that omeprazole may decrease the rate of further bleeding and the need for surgery in patients with recent ulcer-related upper-GI tract bleeding, particularly in those with visible vessels or adherent clots. Because of the small study size and the lack of therapeutic endoscopy, which is contrary to the current practice, these findings cannot be easily applied to the current management of ulcer-related hemorrhage in Western countries. More importantly, the study¹⁰⁴ failed to show a significant difference in mortality between the omeprazole-treated group and the placebo-treated group. It is also noteworthy that in an earlier study,¹⁰⁵ when all patients, not only those with evidence for recent bleeding, were evaluated, omeprazole therapy provided no advantage with respect to rates of rebleeding, transfusion requirements, need for surgery, or mortality.

The identification of patients who might benefit from prophylactic therapy appears to be more important than the particular medication used. Although it is widely practiced, not all critically ill patients need prophylaxis for SRMD.^{58 106} Moreover, no evidence is available (to our knowledge) to suggest that stress ulcer prophylaxis improves mortality in critically ill patients.⁷⁶ This is probably because most deaths in patients with stress ulcer bleeding are not because of GI hemorrhage. In unselected ICU populations, the contribution of stress ulcer bleeding to overall ICU mortality does not appear to be significant; however, this may not be the case in high-risk patients. In a study that evaluated the cost-effectiveness of stress ulcer prophylaxis, Ben-Menachem et al¹⁰⁷ concluded that "the cost of prophylaxis is substantial, and may be prohibitive in ICU patients at low-risk of developing stress-related hemorrhage." Current evidence^{34 35 108 109} suggests that patients with respiratory failure requiring MV for > 48 h and those with coagulopathy (defined as a platelet count of < 50,000/µL, an international normalized ratio of > 1.5, or a partial thromboplastin time more than twice the control value) are at the highest risk and should receive prophylactic therapy. The incidence of clinically significant stress ulcer bleeding in patients without respiratory failure or coagulopathy appears to be negligible (0.1%).³⁴ Among patients receiving MV, those who develop organ dysfunction, particularly renal failure, at any time during their ICU stay appear to be at especially high risk for stress ulcer bleeding.³⁸ Additional risk factors for which stress ulcer prophylaxis should be considered include sepsis, hypotension, hepatic failure, renal failure, major trauma, extensive burns, and intracranial hypertension (Table 5 ).^{35 52 110}

Nasogastric tubes cause mechanical irritation and interfere with normal esophageal motility and sphincter function. Patients with nasogastric tubes also have a higher incidence of GER compared to those without it.¹¹³ Although supine body position contributes to the increased incidence of GER in patients receiving MV, restoration to semirecumbent position does not provide complete protection. In a prospective study of 15 patients who had both intratracheal and nasogastric intubations, Orozco-Levi et al¹¹⁴ showed the presence of GER irrespective of body position. This study emphasized a pivotal role of the nasogastric tube in the development of GER. In another recent prospective study,¹¹⁵ the use of smaller-sized nasogastric tubes (external size, 2.85 mm vs 6 mm) did not affect the incidence of GER in ICU patients receiving MV. These results corroborate studies¹¹⁶ from healthy volunteers, which also show that the size of a nasogastric tube is not an important determinant of GER in normal subjects during short-term nasogastric intubation. While GER is common in patients receiving MV, the standard use of stress ulcer prophylaxis makes acid-induced mucosal damage a less likely explanation for esophagitis.^{113 117} In a study¹¹¹ of 25 patients receiving MV, only 1 of 12 patients (8%) with esophagitis was found to have pathological acid reflux, whereas 9 patients (75%) had evidence of bile reflux, suggesting that chemical injury induced by duodenogastroesophageal reflux and direct trauma caused by nasogastric tubes are the most important factors in the pathogenesis of esophageal mucosal injury. The severity of esophagitis correlates with the volume of residual gastric contents. Gastric colonization of bacteria, which alters bile composition and increases the percentage of injurious unconjugated bile, may contribute to esophageal damage.^{88 118}

To minimize the occurrence and severity of esophageal injury, patients should be kept in a semirecumbent position, nasogastric tubes should be used judiciously, and strategies that improve gastric emptying and prevent both GER and duodenogastric reflux (eg, metoclopramide) should be instituted. Until the controversy about the role of gastric colonization in nosocomial pneumonia resolves, measures that minimize GER and hence microaspiration of contaminated gastric contents remain reasonable approaches to minimize VAP.^{92 95}

	Nonhemorrhagic Complications

TOP Abstract Introduction Interactions Between MV and... GI Hemorrhage Nonhemorrhagic Complications Summary References

 
Hypomotility
GI hypomotility manifesting as decreased bowel sounds or abdominal distention is common and has been reported in up to half of patients with respiratory failure.⁶ In a recent multicenter study, Montejo¹¹⁹ prospectively investigated the frequency of nonhemorrhagic GI complications in 400 ICU patients receiving enteral feeding. Almost two thirds of subjects developed one or more GI complications; high gastric residuals (39%) and constipation (15.7%) were most common. Patients with GI complications had longer ICU stays (20.6 ± 1.2 days vs 15.2 ± 1.3 days) and higher mortality (31% vs 16%) compared to the group without GI complications.

Using manometric evaluation, it has been reported that the motility of the upper GI tract in patients receiving MV is severely impaired.⁵ Contractile activity was completely lost in the stomach and diminished to a lesser degree in the duodenum. Subsequently, Heyland et al¹²⁰ and Bosscha et al¹²¹ confirmed the presence of impaired gastric emptying with reduced but persistent duodenal activity during MV. These abnormalities may be related to dysfunction of interstitial cells of Cajal that are concentrated in the antrum and act as the pacemaker and controller of GI motor activity.¹²² Clinically, most patients with hypomotility present with intolerance to enteral nutrition and high gastric residuals. This contraction abnormality may also favor duodenogastric reflux and colonization of the stomach by enteric Gram-negative pathogens.¹²³ In a recent study, Dive and colleagues¹²⁴ showed the presence of duodenogastric reflux in 10 of 11 patients receiving MV who were receiving nasojejunal tube feedings.

Correction of electrolyte abnormalities (eg, hypokalemia, hypomagnesemia) and avoidance of medications (particularly opiates) that impair GI motility are important for the prevention of ileus and bowel dilatation.¹²⁰ Like opiates, dopamine has been shown to impair GI motility. This negative effect can be seen at doses as low as 5 µg/kg/min and worsens with increasing rates of infusion.^{125 126} Other commonly used medications that cause GI hypomotility are phenothiazines, diltiazem, verapamil, and drugs with anticholinergic side effects. If necessary, nasogastric suction and/or rectal tubes and, in intractable cases, colonoscopy can be used to decompress the GI tract.¹²⁷ Rectal tubes have been associated with complications including discomfort, local ulceration, infection, and perforation of rectum.¹²⁸ Prokinetic agents, such as erythromycin, have been shown to promote gastric emptying in patients receiving MV and should be considered once mechanical obstruction is excluded. Erythromycin, 200 mg once daily, can improve gastric motility in these patients by increasing the amplitude of antral contractions and improving antroduodenal coordination.^{129 130 131} While erythromycin acts via motilin receptors, an intact vagal pathway has been shown to be necessary for its GI effects.^{132 133} Metoclopramide is another prokinetic agent that is useful in the treatment of gastroduodenal hypomotility.^{134 135} The precise mechanism of action is unclear, but metoclopramide improves antroduodenal coordination and reverses the inhibitory effects of dopamine on GI motility.^{125 136 137} Similarly, cisapride stimulates myenteric cholinergic nerves with consequent increase of acetylcholine release and has been used extensively in ICU patients to promote motility.^{138 139} However, > 300 reports showing its association with cardiac arrhythmia, including 80 deaths, have led to withdrawal of cisapride from the US market, although it will remain available through a limited-access program to patients for whom other therapies are not effective. There are insufficient data directly comparing the relative potency of prokinetics in critically ill patients with GI hypomotility.¹⁴⁰ The only relevant study¹⁴¹ in critically ill patients looked at the effects of single doses of cisapride (10 mg), erythromycin (200 mg), and metoclopramide (10 mg) administered sequentially q12h to critically ill patients intolerant to enteral nutrition. Metoclopramide and cisapride were more effective in accelerating gastric emptying compared to erythromycin. In addition, metoclopramide had a faster onset of action than cisapride. Limitations of this work include small study size (10 patients) and results that are in contrast to meta-analysis data¹⁴⁰ from patients with chronic gastroparesis (eg, diabetic gastroparesis), which suggested faster gastric emptying and improvement in GI symptoms with erythromycin compared to metoclopramide. Further investigation in larger populations for longer durations is required to define the precise roles of these agents in critical illness. Interestingly, to our knowledge, there are no studies comparing promotility agents with correctly positioned postpyloric (ie, duodenal) feeding tubes, which might obviate the need for these agents.

A recent study¹⁴² has suggested that neostigmine may be effective in patients with intestinal pseudo-obstruction; although not tested in patients with respiratory failure, it may become a therapeutic tool for colonic hypomotility in critical illness. Major concerns with the use of neostigmine are bradycardia, increased airway secretions, and bronchial reactivity. Concomitant treatment with neostigmine and the anticholinergic agent glycopyrrolate has been reported to diminish the central cholinergic effects of neostigmine without diminishing the improvement in colonic motility.¹⁴³ Further studies that examine the effects of combination therapy with neostigmine and glycopyrrolate are warranted.

Diarrhea
Among nonhemorrhagic complications, diarrhea is the most distressing to patients and nursing staff. Up to 50% of critically ill patients develop diarrhea during their ICU stay, and those with acute respiratory failure appear to be particularly at risk.^{6 144 145 146} Although many factors have been implicated, the etiology of diarrhea in ICU is unknown and probably multifactorial (Table 6 ^{6 144 146 147 148 149} ). While controversy about the role of each risk factor continues, it has also been suggested that diarrhea may be a reflection of the severity of underlying illness that leads to gut dysmotility.¹⁵⁰

Recently, relative luminal excess of bile acids has been offered as a cause of diarrhea in ICU patients.¹⁴⁹ Animal studies have shown that prolonged starvation causes diffuse atrophy of the gut, including the terminal ileum.^{157 158} Hernandez et al¹⁵⁹ performed duodenal biopsies in 15 critically ill patients after at least 4 days of fasting (mean, 7.8 days) and confirmed the presence of mucosal atrophy. Theoretically, if mucosal damage extends to the terminal ileum, abnormal bile acid homeostasis can occur. To test this hypothesis, DeMeo and colleagues¹⁴⁹ measured stool bile acid concentrations in critically ill patients who underwent fasting for at least 5 days. Eighteen of 19 critically ill patients (95%) developed diarrhea when enteral feedings were instituted after 5 days of fasting. Eighty-five percent of the subjects had fivefold to 10-fold increases in stool bile acid as compared to normal volunteers.¹⁶⁰ The administration of bile acid-binding agents improved diarrhea in this study.

Liberal use of antibiotics in ICU patients predisposes patients to antibiotic-associated diarrhea, which accounts for 20 to 50% of all cases of nosocomial diarrhea.¹⁶¹ Five to 38% of patients receiving antibiotics develop antibiotic-associated diarrhea.¹⁶² The incidence has increased fivefold over 10 years, probably because of increasing use of cephalosporins in the early 1990s.¹⁶³ Fifteen to 25% of antibiotic-associated diarrhea is caused by Clostridium difficile infection. Antibiotic-associated diarrhea that is not due to C difficile is probably caused by the direct effect of the antibiotic on intestinal motility and by a reduction of intestinal carbohydrate fermentation.¹⁶¹ It is usually self-limited and resolves with the discontinuation of antibiotic therapy. C difficile, however, is associated with significant morbidity and even mortality if fulminant colitis or toxic megacolon develops because of delay in diagnosis.¹⁶³ The clinical presentations of C difficile infection include—in increasing order of severity—asymptomatic carriage, antibiotic-associated colitis without pseudomembrane formation, pseudomembranous colitis, and fulminant colitis. Fortunately, the most severe forms are also the least common. C difficile diarrhea increases hospital length of stay by an average of 3 weeks.¹⁶⁴ The number and duration of the antibiotics seem to be determinant for C difficile diarrhea. In addition to frequent antibiotic use, patients receiving MV have other risk factors for C difficile diarrhea, including advanced age, prolonged hospitalization, and severe underlying illness.^{165 166} Diagnosis of C difficile diarrhea requires a high index of suspicion and is frequently made by detection of cytotoxins in the stool. The tissue culture assay remains the "gold standard," but it is expensive and requires overnight incubation of samples. The new rapid enzyme immunoassays (EIAs) can detect C difficile with fair sensitivity (69 to 87%) and good specificity (99 to 100%).¹⁶⁷ The major advantages of EIAs are that they are less expensive and quicker to perform than tissue culture assay, do not require specialized training of laboratory personnel, and provide reasonable sensitivity and specificity particularly compared to the initial assays, which were based on latex particle agglutination. Owing to the lack of sensitivity, it may be necessary to repeat the EIA.¹⁶⁸ While there are no guidelines as to how many assays should be performed before C difficile can be excluded, repeat testing may be helpful when clinical suspicion is high. Clinical and laboratory features that predict a positive assay are the onset of diarrhea 6 days after the administration of antibiotics, hospital stay > 15 days, the presence of fecal leukocytes, the presence of semiformed (as opposed to watery) stools, and cephalosporin use.¹⁶⁸ Interestingly, a commonly used drug, sucralfate, has been suggested to interfere with C difficile cytotoxin-B assays,¹⁶⁹ by either direct binding to the toxin itself or through its antibacterial effects.^{170 171}

Hypoalbuminemia has been implicated as a predisposing factor for diarrhea in critically ill patients.^{148 172 173} However, another study¹⁴⁹ has questioned its precise role as a risk factor. Earlier investigations have suggested that low albumin levels can lead to gut edema and impaired nutrient absorption. Brinson and Kolts¹⁴⁸ reported that all patients with a serum albumin level < 2.6 g/dL developed diarrhea, while no diarrhea was seen in those with a level > 2.6 g/dL. Subsequently, Hwang et al¹⁷³ confirmed the association between hypoalbuminemia (albumin < 2 g/dL) and diarrhea. For the same degree of hypoalbuminemia, subjects with chronic malnutrition had a higher incidence of diarrhea compared to those with acute hypoalbuminemia (eg, burn patients). These results suggested that it is not the severity, but rather the chronicity of malnutrition, that is more important in the development of diarrhea.

Treatment of diarrhea depends on the underlying cause. The inability to identify the exact cause often complicates the picture and limits optimal care. C difficile should always be considered in the differential diagnosis and therefore initial workup should include stool cytotoxin assays. The first step in managing diarrhea in association with confirmed or suspected C difficile infection is to discontinue antibiotic therapy, if possible. Patients should be placed on regimens of enteric precautions and empiric antibiotic therapy while the laboratory tests are pending. Oral metronidazole remains the drug of choice, with oral vancomycin being reserved for patients who cannot tolerate or do not respond to metronidazole or for those who are pregnant. Isotonic tube feedings can minimize diarrhea because of hyperosmolar formulas, but there is no evidence to support enteral nutrition with a hypo-osmolar formula (by diluting isotonic tube feedings)to decrease diarrhea in critically ill patients. Studiesthat evaluated the effects of peptide-based enteral formulas with standard tube feedings that contain wholeprotein do not show any difference in incidence of diarrhea.^{174 175 176} Similarly, the addition of fiber to promote the development of colonic flora does not offer any benefit over standard formulas in terms of reducing diarrhea.^{177 178} Although not shown to be beneficial in all patients, formulas composed of small peptides (eg, Peptamine; Nestle; Deerfield, IL) may be better tolerated in patients with severe hypoalbuminemia (albumin < 2.6 g/dL) and diarrhea.¹⁷²

Effects on GI Hemodynamics
MV has a number of adverse effects on splanchnic hemodynamics, particularly when PEEP is used (Table 7 ). It is noteworthy that most of the available evidence regarding how MV affects hemodynamics in the different vascular beds of the GI tract comes from animal studies. While MV may compromise GI hemodynamics in humans in a similar fashion, there are insufficient data to indicate how clinically relevant this problem is. Evidence from experimental studies suggests that PEEP decreases mesenteric blood flow in parallel with reductions in cardiac output. Love and colleagues¹² studied the effects of increasing levels of PEEP on mesenteric perfusion in rats. Addition of 10 cm H₂O of PEEP resulted in reductions in cardiac output and mesenteric blood flow by 31% and 75%, respectively. Although IV fluids improved cardiac output, mesenteric blood flow remained 45% below baseline. These authors¹² noted that decreased arteriolar diameter suggested reflex vasoconstriction. Supporting this hypothesis, dopexamine, a potent ß₂-adrenoceptor and dopaminergic agonist, has been shown to selectively improve mesenteric blood flow during MV.^{179 180} Results from the studies of the effects of dopamine on PEEP-induced mesenteric hypoperfusion have been controversial. Two studies^{181 182} report that dopamine and dobutamine at low and high doses (2.5 µg/kg/min and 12.5 µg/kg/min, respectively) failed to improve PEEP-induced mesenteric hypoperfusion. Within the splanchnic bed, PEEP may decrease blood flow to the pancreas and stomach to a greater extent than intestinal perfusion.¹⁸³ Hemodynamic consequences of PEEP in the pancreas parallel reductions in cardiac output, but occur even when mean arterial pressure is maintained.¹⁸⁴ In animals, high levels of PEEP (15 cm H₂O) have been shown to cause pancreatitis, evidenced by inflammation, vacuolization, necrosis, and hemorrhage on histology and increased serum amylase and lipase levels.^{185 186} Histopathologic changes were evident within the first 24 h of MV with PEEP and were more pronounced with simultaneous stimulation of the gland (using a cholecystokinin analog). Similar to animals, MV may lead to a rise in lipase and amylase levels in humans,¹⁸⁶ but whether these findings represent clinically significant pancreatitis is unknown. An autopsy study¹⁸⁷ has demonstrated major pancreatic injury in patients dying after shock. When patients were examined prospectively, only 4 of 13 patients (30%) with elevated pancreatic amylase and lipase levels had clinical manifestations of acute pancreatitis.¹⁸⁷ No histologic data from patients receiving MV without hemodynamic collapse are available (to our knowledge) to clearly indicate if MV is associated with clinically evident pancreatitis. Thus, concerns about detrimental effects of PEEP on the pancreas remain theoretic but worthy of consideration in critically ill patients with otherwise unexplained signs of pancreatitis.

Positive-pressure ventilation with PEEP mediates its adverse effects on portal blood flow by raising downstream pressure (right atrial, inferior vena cava),^{188 197} by increasing hepatic sinusoidal resistance via mechanical compression of the liver by the descending diaphragm,^{189 191} and by diminishing arterial inflow (mesenteric) into the gut.¹⁸⁸ Conversely, alterations in hepatic arterial flow during PEEP are due, in part, to elevations of downstream pressure.¹⁹¹ While the precise role of arterial resistance is still controversial, the expected increase in the hepatic arterial resistance during PEEP¹⁹⁸ has been shown to be counterbalanced by vasodilatation, described as "hepatic buffer response," which compensates for reduced portal blood flow.^{191 199}

Studies^{189 191 200} addressing the clinical consequences of MV on portal hemodynamics have provided conflicting results. Nevertheless, a mismatch between the hepatic metabolic demand and the blood supply can result in abnormal liver function.²⁰¹ Indeed, reduction in hepatic venous oxygen saturation has been associated with subsequent hyperbilirubinemia and elevation in transaminase levels in humans.²⁰² In patients with septic shock, incremental rise in PEEP induces a drop in hepatic glucose production (a marker for hepatic metabolic performance) in parallel to reductions in cardiac output and hepatic venous oxygen saturation.²⁰³ Furthermore, hepatic clearance of drugs that are highly extracted at the hepatic level and therefore primarily depend on hepatic blood flow (eg, lidocaine) can be impaired by positive-pressure ventilation.^{200 204} In view of current evidence, it is reasonable to hypothesize that PEEP causes liver dysfunction in the presence of hypoxemia, hypotension, or any other condition that further compromises hepatic oxygen supply and that abolishes the hepatic arterial buffer response.

Acute Acalculous Cholecystitis
Acute acalculous cholecystitis (AAC), defined as acute inflammation of the gallbladder in the absence of stones, is an insidious complication that has been increasingly recognized in the ICU. The incidence in critically ill patients ranges from 0.2 to 3%.^{205 206 207} MV (72 h) has been implicated among other risk factors, including shock, sepsis, multiple transfusions, dehydration, prolonged enteral fasting, total parenteral nutrition, and medications (eg, sedatives and opiates; Table 8 ).^{205 208 209 210} The pathophysiology of ACC is probably multifactorial, involving both ischemic and chemical (bile) injuries to the gallbladder epithelium. Prolonged fasting interferes with normal emptying of the gallbladder and leads to stagnation of highly concentrated bile in its lumen.²¹¹ Redistribution of blood away from splanchnic because of critical illness, MV, and use of vasopressors may affect the gallbladder epithelium directly by causing hypoperfusion and ischemia of the gallbladder wall and indirectly by leading to poor contractility with consequent biliary stasis and sludge formation.²¹² For the same duration of fasting after major abdominal surgery, subjects who remained intubated have been shown to have a higher degree of gallbladder atony compared to those who were spontaneously breathing.²¹³ Motility changes were detected as early as 24 h after admission to the ICU.²¹³ The abundance of risk factors that can impair mucosal resistance (visceral ischemia) against injurious effects of bile makes critical illness a perfect setup for ACC.

Although cholecystectomy has been the traditional approach, it is not always feasible because of the severity of underlying disease in ICU patients. In subjects who represent high risk for general anesthesia, drainage via percutaneous cholecytostomy has been shown to be an acceptable option with low procedure-related risk and success rates between 59% and 88%.^{215 223 224} Transpapillary endoscopic cholecystostomy is another treatment option suggested to be useful in those who are also poor candidates for a percutaneous approach.²²⁵

	Summary

TOP Abstract Introduction Interactions Between MV and... GI Hemorrhage Nonhemorrhagic Complications Summary References

 
MV is a lifesaving tool, but it is not without limitations. There are numerous GI complications seen in critically ill patients receiving MV. Although it remains unclear if these complications are the direct effect of MV, current knowledge suggests that MV may contribute to physiologic changes that may impair the function of the GI tract. These changes can lead to common complications, such as SRMD and associated GI hemorrhage and hypomotility, some of which can occur in up to 50% of patients receiving MV. It is unclear to what extent GI complications contribute to the mortality of critically ill patients, but undoubtedly they are associated with significant morbidity that impacts the care of these patients by increasing length of stay and costs. Nevertheless, it is quite likely that GI complications also lead to increased mortality in patients receiving MV. Currently, there exists a broad knowledge base to guide the application of preventive therapies to prevent SRMD, and new data are evolving that may prove helpful for disordered motility. As our understanding of the systemic effects of MV and lung protective ventilatory strategies improves, it can be expected that complications associated with MV will become less common. Our goal in patient care is to not only provide treatment, but to recognize the potential complications related to any given therapy. Better understanding of the limitations and consequences of MV will help us to identify and minimize these complications.

	Acknowledgements

 
The authors thank Ali Keshavarzian, MD (Chief, Division of Digestive Diseases at Rush Presbyterian-St. Luke’s Medical Center, Chicago, IL), for his critical review of the article.

	Footnotes

 
Abbreviations: ACC = acute acalculous cholecystitis; EIA = enzyme immunoassay; GER = gastroesophageal reflux; H₂ = histamine type 2; IL = interleukin; MODS = multiple organ dysfunction syndrome; MV = mechanical ventilation; PEEP = positive end-expiratory pressure; SRMD = stress-related mucosal damage; VAP = ventilator-associated pneumonia

This work is supported by the American Heart Association and Evanston-Northwestern Healthcare Research Institute.

Received for publication July 14, 2000. Accepted for publication November 6, 2000.

	References

TOP Abstract Introduction Interactions Between MV and... GI Hemorrhage Nonhemorrhagic Complications Summary References

 

1. Mutlu, GM, Factor, P (2000) Complications of mechanical ventilation. Respir Care Clin North Am 6,213-252[CrossRef]
2. Lucas, CE, Sugawa, C, Riddle, J, et al (1971) Natural history and surgical dilemma of "stress" gastric bleeding. Arch Surg 102,266-273[ISI][Medline]
3. Czaja, AJ, McAlhany, JC, Pruitt, BA, Jr (1974) Acute gastroduodenal disease after thermal injury: an endoscopic evaluation of incidence and natural history. N Engl J Med 291,925-929
4. Peura, DA, Johnson, LF (1985) Cimetidine for prevention and treatment of gastroduodenal mucosal lesions in patients in an ICUs. Ann Intern Med 103,173-177
5. Dive, A, Moulart, M, Jonard, P, et al (1994) Gastroduodenal motility in mechanically ventilated critically ill patients: a manometric study. Crit Care Med 22,441-447[ISI][Medline]
6. Dark, DS, Pingleton, SK (1989) Nonhemorrhagic gastrointestinal complications in acute respiratory failure. Crit Care Med 17,755-758[ISI][Medline]
7. Bion, JF (1999) Multiple organ failure. Webb, AR Shapiro, MJ Singer, Met al eds. Oxford textbook of critical care ,923-926 Oxford University Press New York, NY.
8. Antonsson, JB, Engstrom, L, Rasmussen, I, et al (1995) Changes in gut intramucosal pH and gut oxygen extraction ratio in a porcine model of peritonitis and hemorrhage. Crit Care Med 23,1872-1881[CrossRef][ISI][Medline]
9. Bohle