Guarding the Larynx

Laryngeal closure (sometimes less precisely termed glottal closure) refers to a series of coordinated motions that typically accompany deglutition. They include descent of the epiglottis and elevation of the arytenoid processes to its base, which combine to cover the laryngeal opening; adduction of the arytenoids and stiffening of the aryepiglottic fold; and adduction of the ventricular folds (false vocal cords) and the true vocal cords (actual glottic closure).

Laryngeal closure typically begins just before the start of the pharyngeal phase of deglutition, which is understandable because the principal muscle of the upper esophageal sphincter (UES), the cricopharyngeus, attaches anteriorly to the cricoid cartilage. Thus, delayed laryngeal closure is associated with delayed UES opening, leading to pooling of the bolus in the hypopharynx and possible entry into the laryngeal vestibule.

In the upright position, the larynx rises above a "moat" formed by the pyriform sinuses, providing a backup mechanism to catch pooled food or secretions. Due to considerable normal variation in pyriform sinus depth, individuals have different reserve capacity to resist overflow into the glottis. This protective mechanism is largely lost in the supine position (although the larynx does remain anterior to the hypopharynx).

Laryngeal closure is compromised by anatomic abnormalities (especially following treatment of cancer by surgery or radiation). Laryngeal closure changes minimally with normal aging¹ (unlike the oral phase of deglutition, which slows in the elderly), but requires a larger stimulus in the elderly.² Laryngeal adduction is paramount to minimize aspiration. Even unilateral vocal cord paralysis following thyroid or parathyroid surgery is associated with clinically evident aspiration in 24% to 61% of cases,^{3, 4} whereas clinical⁵ and experimental⁶ evidence implies that epiglottal descent is dispensable.

Tracheostomy impedes laryngeal closure by reducing laryngeal elevation during deglutition. However, the degree to which this leads to aspiration appears to depend more on other characteristics of the specific patient, especially level of consciousness, than on tracheostomy itself. The balloon of a cuffed tracheostomy tube impinges on the proximal esophagus in some patients, compromising bolus passage. Hence, cuff deflation in alert patients can often improve swallowing function.

Tracheostomy also appears to hamper airway protection in a more subtle fashion, by ablating key feedback from laryngeal mechanoreceptors that normally detect the rise in airway pressure during glottic closure. Dettelbach et al.⁷ studied 11 alert patients with tracheostomies and known aspiration before and after application of a Passy-Muir valve. Besides permitting speech, use of these one-way valves reduced or eliminated aspiration during swallowing in all 11 patients, compared to swallowing with an open tracheostomy. Improvement was seen with liquids, semisolids, and pureed consistencies. Concordant results were seen in three small studies using Passy-Muir valves in alert patients,^{8, 9, 10} and in three tracheostomized children by use of continuous positive air pressure (CPAP),¹¹ which would have the same effect on subglottic pressure. Less encouraging results have been reported,^{12, 13} urging caution and individualization about extrapolating these results. Importantly, however, use of Passy-Muir valves largely reverses the inability of patients with open tracheostomies to cough effectively, permitting them to clear secretions that do penetrate the airway.

Defective laryngeal closure may also accompany disorders of nerves [e.g., amyotrophic lateral sclerosis (ALS), Guillain-Barré syndrome] or muscles (e.g., scleroderma, polymyositis). However, neuromuscular disorders can also compromise respiratory function directly, in the former diseases by limiting lung inflation (i.e., restrictive ventilatory defect), in the latter diseases via associated interstitial lung disease. Thus, pulmonary dysfunction manifest by radiographic abnormalities or dyspnea in patients with known oral-pharyngeal motility disorders cannot be assumed to result solely from defective laryngeal protection and resulting aspiration.

"Don't Breath, Swallow"

The mechanical safeguards provided by laryngeal closure are supplemented during deglutition by brief cessation of ventilation called "swallowing apnea." Beginning slightly before the bolus enters the hypopharynx, this apnea interrupts exhalation in most studies of normal subjects.^{14, 15, 16} Normal subjects complete exhalation after swallowing. Teleologically, postdeglutition exhalation drives material away from the glottis. Swallowing apnea is compromised by neurologic dysfunction, as in stroke or cerebral palsy, and by increased respiratory rate, a key feature of almost all lung diseases.

The predominant concept of swallowing apnea has been "turn-taking" between inspiration and deglutition, regulated by adjacent "central pattern generators" in the medulla oblongata. Dorsal swallowing group neurons near the nucleus of the solitary tract (NTS) regulate the pharyngeal and esophageal phases of deglutition, by activating and inhibiting interneurons and motoneurons in the ventral swallowing group, which reside adjacent to the nucleus ambiguus and the hypoglossal nucleus. Dorsal respiratory group neurons, also in the NTS, integrate vagal inputs to control cough, hiccupping, and sighing. Reciprocal inhibition between these predominately inspiratory neurons of the dorsal respiratory group and nearby expiratory respiratory neurons controls the rhythm of respiration.

In experiments using decerebrate rats, Saito et al.¹⁷ found evidence that synapses between the dorsal respiratory group and the dorsal swallowing group inhibited swallowing during inspiration. Conversely, interneurons that inhibited exhalation (decrementing-expiratory interneurons in the Bötzinger complex) were activated during stimulated swallowing.¹⁸ These data, in agreement with older findings in the cat, support the concept that swallowing apnea results from communication between central pattern generators for deglutition and respiration located in the brainstem. This conventional view of the relationship between deglutition and breathing is a "top-down" control model.

Most emphasis on the neurologic basis of this reciprocity between swallowing and breathing has focused on the impact of the gastrointestinal (GI) tract on respiration. This is the case not only for the motor patterns that determine the timing of the two processes, as in swallow apnea, but also for the reflex arcs that induce laryngeal closure. For example, injection of water into the pharynx normally induces vocal cord closure, the pharyngoglottal closure reflex.² Sudden inflation of a balloon in the esophagus does the same,¹⁹ evidence in humans for the esophagoglottal reflex characterized in the cat.²⁰ These latter findings, cited here in relationship to the gut's impact on respiration, are also one of many pieces of evidence in the debate over the association between gastroesophageal reflux and airways diseases, a controversial topic that is beyond the scope of this review.

Breathing and Swallowing: A Bidirectional Interaction

Rather than reflex turn-taking, other data imply that breathing and swallowing interact bidirectionally. Laryngeal closure, swallowing motions, and strong resetting of the respiratory pattern can all be induced experimentally using electrical stimulation of the internal branch of the superior laryngeal nerve, which carries afferent impulses from the supraglottic larynx and epiglottis.²¹ This finding is seen in humans and experimental mammals. Thus, laryngeal sensation appears to be crucial to protect the airway during deglutition.

This idea is supported by analysis of swallowing in 16 normal subjects following bilateral injection of 0.5% bupivacaine into the paraglottic compartment.²² Selective sensory blockade induced loss of airway protection due to incomplete closure of the larynx during the pharyngeal phase of swallowing. By contrast, laryngeal closure was robust during voluntary the Valsalva, Müller, and cough maneuvers. Interestingly, all subjects developed a sensation of globus and a perceived need to swallow forcefully. Sensory blockade did not alter the normal phase relationships between swallowing and breathing.²² These results bolster the clinical impression that laryngeal desensitization, as occurs particularly on extubation following prolonged translaryngeal intubation, might predispose to aspiration.

Support for regulation of swallowing by respiration comes from a study by Gross et al.²³ of the effect of varying lung volumes in 28 young healthy subjects on specific measures of swallowing physiology. Pharyngeal activity was prolonged during deglutition initiated at low (residual lung volume) relative to higher lung volumes (functional residual capacity or total lung capacity). Because pharyngeal transit times are a risk factor for aspiration pneumonia (in stroke patients²⁴), these findings provide a novel explanation for dysphagia in patients with restrictive ventilatory defects. This possibility needs to be studied prospectively. Changes in lung volume in the normal subjects in the study of Gross et al. did not alter esophageal bolus transit time or intramuscular electromyography of the superior esophageal constrictor.²³ None of these normal subjects aspirated during deglutition initiated at any lung volume. These results were interpreted as evidence that the interaction between deglutition and respiration is more complicated than simple turn-taking, and instead requires interactive ("bottom-up") bidirectional coordination.

This conceptual framework fits well with the clinical observation that successful oral feeding without aspiration depends not only on these reflexes, but on a host of factors including alertness, food consistency, dentition, and oral sensation. Regulation of swallowing also clearly depends on input from the cerebral cortex. One third to one half of patients with unilateral cerebral hemispheric strokes develop oral-pharyngeal dysphagia.^{25, 26, 27} Better basic understanding of the neurophysiologic basis of cortical input to swallowing could improve rehabilitation of stroke patients.

One such advance is provided by an exciting study of Fraser et al.,²⁸ suggesting that the adult brain might be able to be remodeled following injury to improve swallowing function. They found that specific patterns of transcranial magnetic stimulation of normal subjects induced stronger or weaker cortical activation, as assessed by functional magnetic resonance imaging. When applied to 16 acutely dysphagic stroke patients, the optimal patterns of stimulation improved acute swallowing corticobulbar excitability and short-term (2-hour) swallowing function. Fraser et al. found a highly significant correlation within the stroke patients between the total change in excitability and the change in aspiration before versus after cranial stimulation, compatible with a causal relationship. Although this study did not assess duration of improvement, results offer hope of eventual novel therapies to supplement such rehabilitation maneuvers as thermal-tactile stimulation of the faucial pillars, which have a very transient effect.²⁹

Obstructive Lung Disease and Deglutition

The lungs and esophagus share the thoracic cavity, and intrathoracic pressure is dominated by the pleural pressure swings associated with ventilation. In normal subjects, inhalation results from a subatmospheric drop in intrathoracic pressure generated largely by motion of the diaphragm. Normal exhalation is passive, allowing pleural pressures to rise to atmospheric levels. Obstructive lung diseases are characterized by hyperinflation, which flattens the diaphragm and exaggerates pleural pressure changes. In severe obstruction, exhalation is active and associated with positive intrathoracic pressures that can be large. These changes would be anticipated to compromise normal deglutition.

In fact, COPD appears be a significantly underrecognized cause of swallowing dysfunction in adults. It refers to a constellation of lung diseases characterized by varying degrees of mucus hypersecretion, peribronchiolar fibrosis, and parenchymal lung destruction.³⁰ It has recently been defined as "a disease state characterized by airflow limitation that is not fully reversible. The airflow limitation is usually progressive and associated with an abnormal inflammatory response of the lungs to noxious particles and gases."³¹ In industrialized countries, COPD results overwhelmingly from tobacco smoke exposure, but in developing countries indoor air pollution contributes significantly. Worldwide prevalence of COPD in 1990 was estimated at 9.34/1000 in men and 7.33/1000 in women (see first Web site listed in Table 2). However, these estimates include all ages and doubtless underestimate the true prevalence of COPD in older adults. Rapid increases in prevalence (e.g., 25% in men and 69% in women in the United Kingdom between 1990 and 1997) are not due to changes in case definition. The World Health Organization projects that COPD will be the third leading cause of human death worldwide by 2020.³² Therefore, care of the lung disease and associated airway malignancies in COPD patients could come to dominate medical care in the 21st century in the same manner that the 19th and 20th centuries were dominated by tuberculosis and ischemic heart disease, respectively.

Table 2: Web site links: sites extant as at May 2006

• Full size table

Dysphagia appears to be common in COPD patients.³³ In a prospective questionnaire-based cross-sectional survey of patients in Veterans Administration (VA) clinics, Mokhlesi et al.³⁴ found an increased proportion of dysphagia in 100 COPD patients compared to 51 medicine patients without respiratory symptoms or a diagnosis of COPD (17% vs. 4%). Dysphagia may be one major cause of the nutritional depletion that is common in COPD patients and that adversely impacts their exercise capacity.³⁵ However, dysphagia with decreased caloric intake is probably not the only cause of weight loss in advanced COPD, as both hypermetabolism³⁶ and anorexia³⁷ have been demonstrated. Defining the relative contribution of these independent and potentially additive factors will be important, because weight loss has been associated with significantly increased morbidity and death in advanced COPD.^{38, 39, 40}

One likely cause for dysphagia in COPD is that swallowing apnea cuts into the already compromised time available for breathing. Owing to an increase in wasted ventilation from lung damage, patients with mild to moderate COPD must sustain a higher than normal minute ventilation (the amount of air inhaled per minute) even at rest, simply to keep their blood carbon dioxide levels normal. As lung damage worsens, many patients with advanced COPD accommodate to CO₂ retention rather than further increasing their resting minute ventilation. Heightened ventilatory demands during exercise or respiratory infections pose a further dilemma. Due to limited ability to increase the amount of air inhaled per breath (tidal volume), COPD patients increase their minute ventilation chiefly by increasing respiratory rate. By reducing the relative time available for exhalation, this leads rapidly to trapping of excess air in their chests, stretching respiratory muscles to a further mechanical disadvantage. This cycle of air trapping and muscle weakness can quickly lead to hypercapnic respiratory failure.

These factors imply that dysregulation of swallowing apnea would be seen in COPD patients, especially when their breathing worsens. That was exactly the effect seen by Shaker et al.¹⁹ in a landmark analysis on the effect of COPD on deglutition. They studied respiration during voluntary deglutition in 26 COPD patients (46 to 72 years old) during an acute exacerbation of chronic bronchitis (AECB), and restudied 10 of these patients in remission. Their control groups consisted of 18 healthy young (18- to 34-year-old) and 11 healthy older (63- to 83-year-old) volunteers studied in the upright and supine positions. As seen previously in normal subjects,^{14, 15, 16} virtually all deglutitions in control subjects interrupted expiration, and this coupling to the expiratory phase was increased by the presence of a liquid bolus and tachypnea (rapid breathing). By contrast, COPD patients studied during AECB showed a markedly increased frequency of inspiratory swallows, with some improvement in the groups studied in remission. Shaker et al. defined a deglutition/respiration index, that is, the average duration of deglutition apnea divided by the duration of the respiratory cycle. This ratio fell in COPD patients studied during AECB. These data suggest that COPD patients are at increased risk of aspiration precisely when they are already undergoing stress of their respiratory system.

Dysphagia in COPD patients could have several other causes. In a study using systematic videofluoroscopic evaluation of oropharyngeal swallowing, Mokhlesi et al.³³ compared 20 consecutive outpatients with COPD of at least moderate severity [mean forced expiratory volume in 1 second (FEV₁) 40 14% predicted; mean total lung capacity 128 19% predicted] to 20 age- and sex-matched historical control subjects. They excluded those with a history of conditions that might affect oropharyngeal swallowing, but not those with gastroesophageal reflux (GER). Although there was no evidence of tracheal aspiration in either group, and only four of the patients with COPD reported dysphagia, the COPD group showed a range of swallowing disorders, including abnormalities in tongue strength and motion, delayed pharyngeal swallowing, and slowed or delayed vestibule closure. Importantly, maximal laryngeal elevation during swallowing was significantly reduced in patients with COPD.³³ Patients with COPD also more frequently used spontaneous protective maneuvers (longer duration of airway closure and earlier laryngeal closure relative to cricopharyngeal opening) during swallowing. Interestingly, despite the marked hyperinflation in the COPD patients, which hypothetically might lead to caudal retraction of the larynx, there was no difference between groups in the laryngeal position at rest relative to the cervical vertebrae.³³

Stein et al.⁴¹ described severe cricopharyngeal dysfunction in 17 of 22 nonrandomly selected elderly patients with COPD who were referred for frequent exacerbations of their chronic respiratory symptoms. Ten of the patients underwent cricopharyngeal myotomy, which in eight patients improved not only deglutition but also reported frequency of AECB. Given the availability of endoscopic botulinum toxin injection as a minimally invasive means of addressing cricopharyngeus dysfunction, this is an area that merits controlled investigation.

Whether cricopharyngeal dysfunction is actually increased in unselected COPD patients is unknown. The paucity of data from controlled trials on dysphagia in COPD should not be surprising. Tellingly, a systematic review of the literature using the Cochrane registries was unable to identify any controlled trials of therapy of dysphagia in adults or children with chronic muscle diseases,⁴² a problem that has been recognized much longer than has dysphagia in COPD.

Acute Airway Occlusions

Acute obstruction of the upper airway by food (the "café coronary" syndrome of Haugen⁴⁵) can lead to death by asphyxia within minutes. A recent retrospective series analyzing almost 300 autopsies supported the popular association with alcohol intoxication and large pieces of meat in previously healthy middle-aged adults.⁴⁶ Considering all ages, obstructing material was supraglottic or within the glottis itself in 74% in these fatal cases. However, in the elderly (>65 years of age), a single large chunk of food caused obstruction in only 4%, and semisolid foods predominated over solids (58 vs. 38 cases). By contrast to younger subjects, neurologic impairment (45 of 105 individuals) and defective dentition appeared to be more crucial predisposing factors for fatal foreign-body asphyxiation in the elderly. A large mass of food was found in the mouths of 46% of patients aged 65 years; 27% of the elderly died in hospitals, and asphyxiation was suspected before autopsy in only four of 75 (5%) of cases in this age group.⁴⁶ Thus, foreign-body asphyxia in the elderly appears to be a consequence of an oral-pharyngeal motility disorder. The unique epidemiology in this age group needs to be publicized to health care workers and the general public.

The question of whether aspiration of modest amounts of liquid could induce ultimately fatal reflex laryngospasm ("dry drowning") is relevant to the discussion of pulmonary complications of dysphagia. By convention, the term drowning is reserved for episodes that are fatal out of the hospital, whereas near drowning refers to cases in which patients survive to receive emergency treatment (even if they die later). More recently, the more neutral term immersion injury has gained favor for the majority of cases that involve swimming pools and other total body exposures. Animal studies performed before the 1950s were interpreted to support the concept of dry drowning during immersion events, and it was held that up to 15% of drowning victims have no water distal to the larynx and die of asphyxia apparently triggered by reflex laryngospasm.⁴⁷ The entire concept of dry drowning has been reappraised and questioned,^{48, 49} with the suggestion that such cases may represent deaths by other natural causes or disposal of dead bodies.

Nevertheless, this concept has long raised concerns in speech pathologists and others dealing with patients with known or suspected swallowing disorders. The theoretical possibility that laryngospasm might be induced in dysphagic patients by aspiration of test liquids has been discussed,⁵⁰ largely to point out the absence of clear evidence. Even in X-linked spinobulbar muscular atrophy (Kennedy disease), a condition in which laryngospasm is frequent [23 of 49 patients (47%)], there has been no association with deglutition.⁵¹ Searching the National Library of Medicine PubMed database (Table 2) (using the terms laryngospasm and laryngeal dystonia does not reveal any cases that would support an increased risk of laryngospasm induced by ingested liquids in dysphagic patients. Well-documented cases, to the contrary, would be useful to report, but at present, clinical swallowing evaluation using liquids appears safe and prudent.

Toxic Aspiration Syndromes

Aspiration of toxic fluids causes a chemical pneumonitis that in its most severe form can lead to diffuse alveolar damage and the adult respiratory distress syndrome (ARDS). Although this syndrome generally relates to alterations in consciousness rather than true oral-pharyngeal motility disorders, it is important to understand in relation to the final two categories of aspiration syndromes.

Mendelson⁵² described the classic form as a syndrome that followed aspiration of acidic gastric contents in obstetric patients, and it is with this syndrome that most surgeons and anesthesiologists link the term aspiration. There were no fatalities due to acid aspiration alone in Mendelson's series, in agreement with some^{53, 54, 55} but not all^{56, 57} series examining the perioperative setting.

Experimentally, the extent of lung injury induced by aspirated fluids depends on their volume and pH. Significant lung injury required 1 to 4 mL/kg in dogs⁵⁸ and 1 mL/kg in a primate model,⁵⁹ which would correspond to at least 50 mL in adult humans. It is commonly held that a pH of below 2.5 is necessary to cause significant acid injury, motivating preoperative prophylaxis to raise gastric pH. This assumption has been challenged based on experimental grounds.⁶⁰ Guidelines of the American Society of Anesthesiologists (Table 2) do not recommend preoperative agents to alter gastric pH in patients who have no apparent risk of aspiration.

The clinical syndrome induced by aspiration of gastric contents may be spectacular, with rapid respiratory failure, or it may be silent.⁵³ Injury occurs in two phases, with an initial peak at 1 to 2 hours likely due to direct caustic injury, and a second peak at 4 to 6 hours.⁶¹ The pathophysiology of other toxic aspirations, including petrochemical hydrocarbons, is believed to parallel that of acid injury but is less well characterized.

Current treatment of toxic aspiration pneumonitis is supportive, and similar to ARDS of other etiologies.^{62, 63} Bronchoscopic lung lavage, nonsteroidal antiinflammatory agents, and corticosteroids are all of unproven benefit and may be detrimental.^{64, 65, 66, 67, 68} Prophylactic antibiotics in toxic aspirations remains controversial. Cultures obtained shortly after aspiration are usually sterile. In a prospective study of 69 patients who sustained 98 aspiration events in a chronic care setting, only 56% of episodes progressed to nosocomial pneumonia.⁶⁹ There are no controlled trials that show a benefit for immediate use of antibiotics, which have the theoretical risk of selecting for resistant organisms. However, the stomach can be colonized with enteric organisms when gastric pH is elevated, with tube feedings, and with intestinal obstruction. Presence of these risk factors might plausibly justify prophylaxis to many clinicians. Established pneumonia should be treated aggressively, as discussed in the next section.

Based on these principles and their own retrospective data, Mylotte et al.⁷⁰ have suggested an approach that involves withholding antibiotics for aspiration pneumonitis (lower respiratory tract signs and symptoms plus history of a definite or suspected aspiration event plus radiographic demonstration of a lower lobe infiltrate) of <24 hours' duration. In this approach, antibiotics would be indicated immediately if any of the following factors were found at presentation: cough, pleuritic pain, fever 100°F, purulent sputum, respiratory rate 25 breaths/minute, or localized auscultatory findings of consolidation. However, the authors point out that their algorithm requires prospective validation before its use is generalized.

Recent experimental analysis of acid aspiration has highlighted the roles of the alternative complement pathway, neutrophils, and oxidants in the second phase of injury.^{71, 72} Following these leads, some promise of ameliorating lung injury from acid aspiration has been seen using intraperitoneal administration of perfluorocarbons in rats⁷³ and recombinant surfactant protein C in rabbits.⁷⁴ Improvements in surrogate markers has also been seen using an antiendothelial selectin approach in a murine model^{75, 76}; the contrast of this result with previous endothelial selectin-independence shown by the same group⁷² may relate to the very high basal neutrophil numbers seen in the knockout strain used in the earlier study. These novel experimental therapies are interesting as they appeared to work even when applied within up to an hour, but their efficacy awaits confirmation in humans.

Bacterial Lung Infections Associated with Aspiration

By contrast to the toxic aspiration pneumonitis just described, to internists the term aspiration pneumonia generally refers to bacterial infections of the lungs (which could be a delayed consequence of aspiration pneumonitis). In one sense, aspiration is probably important in the pathogenesis of virtually all pneumonias. Aside from a few organisms capable of causing infection on inhalation (tuberculosis, anthrax, endemic fungi, possibly Legionella), most pneumonias are believed to result by spread of organisms from the oropharynx.^{56, 77, 78}

Small but detectable amounts of aspiration have been shown in sleeping normal subjects,^{79, 80} yet they do not all develop pneumonia. These results imply that failure of normal airway defenses (laryngeal closure, cough, and cellular immune factors) is only one factor in the development of pneumonia. A continuum can be imagined: pneumonia can develop in a relatively normal host due to a markedly virulent organism and a momentary lapse in defenses; less virulent organisms could suffice if defenses are compromised repeatedly; and an immunocompromised host might develop pneumonia with minimal changes in airway protection.

Two distinct spectra of such infections can be distinguished clinically and epidemiologically, with important differences that impact treatment: (1) indolent pleuropulmonary syndromes due largely to anaerobic organisms; and (2) nosocomial pneumonias, in which aerobic bacteria predominate. Once again, use of the single term aspiration to describe both types of infections has often led to prescription of antibiotics that are not totally appropriate for the microbiology.

Anaerobic Pleuropulmonary Infections

Anaerobic lung infections apparently begin as focal pneumonia, which, if not adequately handled by cellular immune defenses, can progress to an abscess, a usually spherical mass-like lesion, frequently showing an air-fluid level. Additionally, if the initial infection abuts the edge of the lungs, it can break into the pleural space (i.e., between the lung and the inner chest wall), and spread rapidly around the surface of the lung before becoming loculated in what is known as an empyema. Elements of all three conditions (pneumonia, abscess, empyema) can coexist in a given patient.

Anaerobic lung infections of all types present indolently. Patients frequently have only low-grade fevers and night sweats, whereas constitutional symptoms such as weight loss, fatigue, and malaise are frequently profound. This constellation of symptoms may lead to a mistaken diagnosis of malignancy.

Classic anaerobic infections of the lower respiratory tract are strongly associated with poor dentition, and are distinctly uncommon in edentulous patients. The other major predisposing factor, intermittent loss of consciousness (e.g., from epilepsy, alcoholism, substance abuse, or transient ischemic attacks), likely works, at least in part, by reducing airway protection and suppressing cough.

Nosocomial Pneumonias

Pneumonias in the health care setting are a heterogeneous group of infections with high case fatality. Causative organisms are also believed to come from the oropharynx in most nosocomial pneumonias,^{56, 77, 78} but the roles of oral-pharyngeal motility disorders and frank aspiration events are even less clear-cut than with the anaerobic infections just described. A very large percentage of hospital-acquired pneumonias occur within the first week after endotracheal intubation, which combines the potential for leakage of oropharyngeal secretions into the trachea via creases in the endotracheal cuff, deep sedation, and impairment of cough efficiency.

A joint statement from the American Thoracic Society and the Infectious Disease Society of America⁸¹ on clinical practice guidelines for the management of nosocomial pneumonias, and a critical review on methods to prevent them⁸² have been published recently. As stated in the former document, "nearly all of the evidence-based data on risk factors for bacterial HAP [hospital-acquired pneumonia] have been collected from observational studies, which cannot distinguish causation from noncausal association."⁸¹

Nosocomial pneumonias are frequently polymicrobial. Aerobic gram-negative bacilli and gram-positive cocci predominate, and viruses and fungi are uncommon causes in immunocompetent hosts. Anaerobic organisms appear to be far less important in nosocomial pneumonias than in pneumonias acquired in the outpatient setting, although the two studies that used bronchoscopic techniques and quantitative cultures disagreed on their exact role. Marik et al.⁸³ isolated no pathogenic anaerobes in 25 episodes of suspected nosocomial aspiration pneumonia and 185 cases of suspected ventilator-acquired pneumonia, whereas Doré et al.⁸⁴ did in 30 of 130 cases of suspected ventilator-acquired pneumonia.

What Is the Role of Dysphagia in Aspiration Pneumonia?

Given this background on the importance of oral-pharyngeal colonization, dental health, and airway protection, it might seem intuitive that oral-pharyngeal motility disorders should strongly predispose to aspiration pneumonias. However, some of the better controlled prospective studies have found no association between objective findings on swallowing evaluation and aspiration pneumonia.^{85, 86}

One reason for this comes from a prospective study by Langmore et al.,⁸⁷ who examined risk factors in a range of elderly patients, including those who were acutely ill and hospitalized, those in nursing homes, and relatively healthy outpatients. One hundred eighty-nine subjects 60 years old who were recruited from a single VA Medical Center underwent systematic evaluation of oropharyngeal swallowing function, and then were followed for evidence of aspiration pneumonia for up to 3 years. A complaint of dysphagia was not an inclusionary or exclusionary criterion for enrollment, although patients with a current or past history of head and neck cancer were excluded. Aspiration pneumonia was common, occurring in 21.7% at a median time from entry of 6 months. Dysphagia was an important risk factor for aspiration pneumonia in bivariate analysis (81% vs. 47%), but not on logistic regression analysis, where (in rank order) dependent for feeding, dependent for oral care, number of decayed teeth, and tube feeding were all significant factors.

Hence, Langmore et al.⁸⁷ showed that dysphagia is one important risk for aspiration pneumonia, but is generally not sufficient in the absence of other factors. Although stroke and other neurologic disease have historically been identified as risks for aspiration pneumonia, this study found similar high cumulative rates (26–33%) in neurologically intact elderly patients with COPD, GI disease, and congestive heart failure. This study was important for considering functional status, dental health, and a variety of medical conditions in addition to swallowing function. We will return to this issue in the final section.

Chronic Noninfectious Responses to Repeated Aspiration of Food

Longstanding aspiration of food has traditionally been identified as one cause of interstitial pulmonary fibrosis. Experimentally, granulomatous pulmonary inflammation can be induced by intratracheal instillation of saline containing food particles.⁶⁰ Pathologic confirmation of aspirated food particles may be difficult unless careful serial sections are performed.⁸⁸

However, some such cases of fibrosis would be classified today as the delayed complication of ARDS triggered by an initial single aspiration event.⁸⁹ In the absence of biopsy confirmation, it is difficult to exclude the possibility that other older cases would be recognized today as idiopathic pulmonary fibrosis (IPF). Chest radiographs may be nonspecific in this prevalent condition, but it can now be diagnosed with high certainty when high-resolution computed tomography shows a characteristic appearance.^{90, 91} There is no compelling evidence that recurrent aspiration predisposes to IPF. An association of IPF with abnormal esophageal acid exposure has been noted in 17 consecutive patients with biopsy-proven IPF,^{92, 93} but it has not been proven that the GER is causative, and these cases did not show foreign body granulomata.

Matsuse et al.⁹⁴ proposed the term diffuse aspiration bronchiolitis (DAB), based on autopsy findings in 23 elderly Japanese patients (81.9 8.3 years, mean standard deviation) of a chronic inflammatory reaction to aspirated foreign particles. They commented that DAB resembled diffuse panbronchiolitis radiographically and pathologically, and like it, was characterized by productive cough, bronchospasm, and dyspnea. Given this presentation, it is difficult to exclude the possibility that occult aspiration caused the foreign-body reaction but not the entire syndrome.

Do Enteral Feeding Tubes Protect Against Pulmonary Complications?

Collectively, this diversity of syndromes encompassed by the term aspiration, and the variety of clinical settings in which enteral feeding tubes are used, explain why there is no simple answer to the question of whether they protect against pulmonary complications of dysphagia. Until recently, the clinical literature on the subject has been dominated by poorly controlled studies consisting of heterogeneous end points and etiologies for dysphagia, and likely case-selection bias. Nonrandomized studies on diverse groups of dysphagic patients showed equally high rates of pulmonary complications with jejunostomy tubes and gastrostomy tubes,^{95, 96} or even lower rates with oral feeding than with tube feeding.⁸⁶ A recent systematic review of the Cochrane Database on trials of gastrostomy or jejunostomy versus oral feeding alone in children with feeding disorders due to cerebral palsy failed to identify any studies that met inclusion criteria,⁹⁷ and a similar review in preterm infants found no evidence of benefit and some evidence of increased GI disturbance for transpyloric tubes in comparison with gastric tubes.⁹⁸

Moreover, as the controversy surrounding the Terry Schiavo case in 2005⁹⁹ illustrated, strongly held ethical viewpoints enter into any discussion about the use of enteral feeding in the principal target population, the neurologically impaired. Hence it is not surprising that there are no published controlled trials of enteral feeding in advanced dementia. Two critical reviews on that subject found scant evidence of objective improvement with enteral feeding.^{100, 101}

There are theoretical reasons to believe that enteral feeding tubes may actually promote lung infection rather than protect against it. Tubes that traverse the esophagus compromise the integrity of the upper and lower esophageal sphincters, and provide a potential pathway for bacterial migration to the oropharynx. Food in the stomach may facilitate bacterial colonization of the stomach by increasing gastric pH, a clear risk factor for gram-negative pneumonia in the mechanically ventilated patient. Probably most importantly, enteral feeding tubes, regardless of anatomic site of entry, do not prevent aspiration of saliva, the principal source of bacteria entering the lungs.^{102, 103}

A beneficial effect of percutaneous endoscopic gastrostomy (PEG) tubes over feeding via nasogastric tubes (NGTs) in dysphagic patient with acute strokes was reported by Norton et al.¹⁰⁴ in a small randomized prospective study from two centers. The study group consisted of 30 patients with persisting dysphagia at 14 days after stroke. Six-week mortality was significantly lower in the PEG groups (two deaths, 12% mortality) than in the NGT group (eight deaths, 57% mortality). Other outcomes (serum albumin concentration, length of stay) also favored PEG.

A favorable but less dramatic difference in favor of PEG was seen in a more heterogeneous group of subjects reported by Dwolatzky et al.¹⁰⁵ They performed a nonrandomized clinical study involving elderly patients (>65 years) from six acute geriatric units and long-term-care hospitals who were thought on clinical grounds to require long-term enteral feeding. Patients were assigned to NGT (90 patients) or PEG (32 patients) at the physician's discretion. Aspiration was defined by the need for suctioning to remove gastric contents or development of pneumonia; systematic evaluation by videofluoroscopy or endoscopy was not performed. Despite a greater mean age and a higher prevalence of dementia, patients receiving PEGs had significantly greater survival by Kaplan-Meier analysis, as well as reduced aspiration [hazard ratio 0.48, 95% confidence interval (CI) 0.22-0.76]. Crossover between treatments occurred more often from NGT to PEG (34.4% vs. 6.7%). These positive results contrast with negative results in previous studies of shorter duration, but interpretation is limited by the nonrandomized nature. Randomized controlled trials in patients with dementia would be very helpful.

The recent publication of two papers by the FOOD (Feed or Ordinary Diet) Trial Collaboration introduce marked improvements in power and study design. The FOOD trials consisted of a novel approach that permitted simultaneous performance of three multicenter international trials in which separate but interrelated hypotheses can be tested while sharing trial resources such as centralized randomization and follow-up and data-collection systems. All three trials examined patients with acute strokes (first-ever or recurrent, excluding subarachnoid hemorrhage) hospitalized between November 1996 and July 2003. All three trials were terminated early due to funding problems, but the largest (after randomization of 4023 of a planned 6000 patients) did show that routine oral supplementation in well-nourished stroke patients (92% of those entering) was unlikely to have clinical benefit.¹⁰⁶

The other two studies by the FOOD Trial Collaboration, published jointly, provide important data on the timing and route of administration of early feeding in dysphasic stroke patients.¹⁰⁷ In the early tube versus avoid tube trial, patients were allocated to start enteral tube feeding (429 patients) (via the clinician's preferred tube) as soon as possible or to avoid any enteral tube feeding for at least 7 days (430 patients). Patients who were not tube fed were given parenteral fluids but not nutrition. The allocated method was continued as long as it remained practical, or as the patient's condition dictated. The primary outcomes were death or survival with poor outcome (assessed as a score of 4 to 5 on a modified Rankin scale). The early tube group showed a nonsignificant reduction in absolute risk of death of 5.8% (95% CI -0.8 to 12.5, p = .09), offset by an 4.7% excess of survivors with poor functional outcome.¹⁰⁷

In the PEG versus nasogastric trial from the FOOD Trial Collaboration,¹⁰⁷ patients were allocated to enteral tube feeding via PEG (162 patients) or nasogastric tube (159 patients) within 3 days of enrollment. Allocation to PEG feeding was associated with a nonsignificant increase in the absolute risk of death of 1.0% (95% CI -10.0 to 11.9, p = 0.09) but an increase of borderline significance in absolute risk of death or poor outcome of 7.8% (95% CI 0.0% to 15.5%, p = .05).

Neither FOOD trial showed a difference between groups in the incidence of pneumonia.¹⁰⁷ These data are consistent with the conclusions of a meta-analysis of three much smaller previous trials,¹⁰⁸ and disagree with the optimistic results of Norton et al.¹⁰⁴ regarding PEG. Intriguingly, allocation of acute stroke patients to PEG in the FOOD trial was associated with more frequent pressure sores and a greater fraction of patients still on tube feeds at 6 months. These finding suggest that this route of feeding might foster dependency, an interesting topic for further study.