Aspiration pneumonia is by far the most common form of hospital-acquired pneumonia, occurring in four to eight of every 1000 patients admitted to hospitals in the United States [1]. Because the reported mortality rates for patients with aspiration pneumonia have ranged from 20% to 62% [2, 3], aspiration pneumonia should be recognized as a life-threatening condition, especially in older patients. Some patients with aspiration pneumonia have antecedent clinical signs or symptoms of aspiration, including choking or coughing during swallowing, a chronic cough, or voice changes such as hoarseness [4]. However, as many as 40% of patients who aspirate have clinically silent aspiration, with no recognizable signs or symptoms [5]. As a result, the clinical history is a notoriously unreliable means of determining which patients are at risk for developing aspiration pneumonia [6, 7, 8].

A videofluoroscopic swallowing study often is obtained as part of the workup of patients with difficulty swallowing or with clinically suspected aspiration. This examination requires the patient to ingest barium boluses of varying sizes and consistencies so that the oral and pharyngeal phases of swallowing can be assessed. Depending on the radiographic findings and the degree of aspiration, the patient can be taught compensatory maneuvers to make during swallowing, the patient's diet can be modified, or, if necessary, oral feedings can be withheld. The videofluoroscopic study therefore is a potentially valuable tool not only for detecting aspiration but also for guiding the management of patients at risk for the development of aspiration pneumonia.

In the combined data from five previous studies, 53 (24%) of 221 patients who aspirated on barium studies developed aspiration pneumonia versus 20 (4%) of 493 patients in whom no aspiration was observed [9, 10, 11, 12, 13]. However, these studies did not attempt to correlate the degree of swallowing dysfunction with the likelihood of developing aspiration pneumonia. Also, all but one of these studies had small sample sizes [9, 10, 11, 12], and all but one consisted exclusively of stroke patients [9, 10, 11, 13]. The purpose of our investigation was to determine the relationship between the degree of dysfunction observed on videofluoroscopic swallowing studies and the likelihood of developing pneumonia in a large series of patients with a variety of medical conditions.

Our affiliated Veterans Administration medical center was selected as the site for this investigation because the center is a self-contained health care system in which we were more likely to obtain adequate clinical follow-up than at our university hospital. A computerized radiology database revealed that 678 videofluoroscopic swallowing studies had been performed during a 3-year period from January 1999 through December 2001.

A total of 297 (44%) of these 678 studies were excluded from analysis because of one or more of the following reasons: the patient's swallowing findings were not adequately described in the radiology reports; the videofluoroscopic swallowing studies were performed with water-soluble contrast agents rather than with barium; the patient had undergone more than one videofluoroscopic swallowing study during this period, so we included only the index examination in our study; the patient had undergone a prior laryngectomy or other pharyngeal surgery; the patient had died within a 6-month period after the videofluoroscopic swallowing study and thus could have developed pneumonia if he or she had lived longer; or the patient's clinical records were not available. The remaining 381 patients composed our study group.

The videofluoroscopic swallowing studies in our study population were obtained by having the patient ingest a high-density (250% weight/volume) barium suspension (E-Z-HD, E-Z-EM, Westbury, NY) in frontal, lateral, and, if necessary, oblique positions with video recordings obtained during swalowing. Subsequent spot radiographs were obtained during suspended respiration, phonation, and modified Valsalva maneuvers. Depending on the patient's condition, the studies were performed with the patient standing on the footrest of the fluoroscopy table or seated in a speech therapy chair. A speech therapist often was present to assist the radiologist performing the examination. Unless marked aspiration of high-density barium was observed, the patient subsequently was asked to swallow a low-density (50% weight/volume) barium suspension (Entrobar, Lafayette Pharmaceuticals, Lafayette, IN) as well as a barium paste or barium-impregnated food substances such as cookies or crackers. Depending on the patient's condition and clinical history and the findings on the initial portion of the swallowing study, some patients ingested an additional bolus of high-density barium (E-Z-HD) in the upright, left posterior oblique position or low-density barium (Entrobar) in the prone, right anterior oblique position for double-contrast or single-contrast images of the esophagus. All videofluoroscopic swallowing studies were obtained with conventional fluoroscopy equipment (400-speed RFXII, General Electric Medical Systems, Wauke-sha, WI). All examinations were performed by radiology residents or by one of three attending gastrointestinal radiologists, and all were interpreted by the attending radiologists.

Because the videotapes of the videofluoroscopic swallowing studies had not been stored on a long-term basis, the original radiology reports for these 381 examinations were reviewed retrospectively by one of the authors to determine whether swallowing dysfunction (including difficulty controlling the bolus, abnormal tongue motion, nasopharyngeal regurgitation, decreased or delayed epiglottic tilt, pharyngeal paresis, or cricopharyngeal dysfunction) was observed and whether laryngeal penetration or tracheobronchial aspiration was present. Laryngeal penetration was defined as entry of barium into the larynx during swallowing without any passage of barium below the true vocal cords, whereas tracheobronchial aspiration was defined as entry of barium into the trachea or bronchial tree. Silent aspiration was defined as entry of barium into the trachea or bronchial tree without an associated cough reflex, so patients with silent aspiration represented a subset of all patients with aspiration observed on videofluoroscopic swallowing studies. On the basis of radiographic findings, each patient was classified into one of five groups: patients with normal swallowing; those with abnormal swallowing but no laryngeal penetration or tracheobronchial aspiration; those with laryngeal penetration; those with tracheobronchial aspiration; and those with silent tracheobronchial aspiration.

A computerized clinical database was then reviewed by one of the authors to determine whether any of the 381 patients had developed pneumonia during the 6 months before or after the videofluoroscopic swallowing study. The author who performed the review was unaware of the findings on videofluoroscopic swallowing studies at the time of review so as to avoid reviewer bias. Chest radiographs had been obtained in 259 (68%) of the 381 patients. As in previous studies, the diagnosis of pneumonia required the presence of an opacity visible on chest radiographs and two or more of the following clinical signs or symptoms: temperature greater than 100°F, WBC greater than 10 × 10³/μL, sputum cultures with positive results for bacteria (only 10 patients had sputum cultures), dyspnea, chest pain, or cough [13, 14, 15, 16]. The clinical database was also reviewed to determine the indications for the videofluoroscopic swallowing study.

We analyzed the data using statistical software (STATA version 7.0, STATA, College Station, TX). We performed a chi-square test for trends to determine whether the risk of developing pneumonia increased significantly with each level of swallowing dysfunction observed on the videofluoroscopic studies. A chi-square test was also performed to determine the odds ratio and 95% confidence interval for each level of swallowing dysfunction on the videofluoroscopic studies, using the patients with normal swallowing as the reference group.

The institutional review board at the Veterans Administration medical center approved all aspects of this retrospective study and did not require informed consent from patients whose records were included in our study.

Despite the frequent use of barium studies for evaluation of laryngeal penetration and tracheobronchial aspiration, the medical literature contains surprisingly little data about the relationship between the degree of swallowing dysfunction observed on videofluoroscopic studies and the likelihood of developing pneumonia. In our study, however, we found a significant correlation between the degree of swallowing dysfunction seen on videofluoroscopic studies and the relative risk of pneumonia. Patients with laryngeal penetration, tracheobronchial aspiration, and silent tracheobronchial aspiration on videofluoroscopic swallowing studies were approximately four times (p = 0.008), 10 times (p < 0.0001), and 13 times (p < 0.0001), respectively, more likely to develop pneumonia than those with normal swallowing. Thus, the likelihood of developing pneumonia was directly related to the degree of swallowing dysfunction seen on the videofluoroscopic studies.

Previous studies have shown that patients with swallowing abnormalities such as delayed initiation of the pharyngeal phase of swallowing and pharyngeal stasis have a higher risk of developing pneumonia than those with normal swallowing on videofluoroscopic swallowing studies [9, 16]. However, these studies did not directly correlate the risk of developing pneumonia with laryngeal penetration or tracheobronchial aspiration. In our study, no significant difference was found in the frequency of pneumonia in patients with abnormal swallowing but no laryngeal penetration compared with those with normal swallowing on the videofluoroscopic studies (p = 0.85). Our findings suggest that modifying a patient's diet or withholding oral feedings may not be necessary in patients with abnormal swallowing on videofluoroscopic swallowing studies, as long as neither laryngeal penetration nor tracheobronchial aspiration is detected on these examinations.

Previous studies have also shown that the cough reflex is depressed in patients with aspiration pneumonia [17, 18] and that the risk of developing pneumonia is substantially higher in patients with silent aspiration compared with those in whom aspiration elicits a cough reflex [10]. Results of our study corroborate the latter observation; patients with silent aspiration on videofluoroscopic swallowing studies had a higher frequency of pneumonia than those who coughed when they aspirated the barium. Our findings underscore the clinical importance of detecting silent aspiration and the value of the videofluoroscopic swallowing study for its detection. Conversely, no significant difference was found in the frequency of pneumonia in patients with tracheal aspiration compared with those with aspiration into the bronchial tree (p = 0.49). Thus, when patients aspirate during barium studies, the depth of penetration of barium into the tracheobronchial tree does not seem to be nearly as important a factor in the development of pneumonia as is the presence or absence of a cough reflex in these individuals.

A definitive diagnosis of aspiration pneumonia is often difficult to establish on clinical grounds because of problems in differentiating this condition from other forms of pneumonia. Aspiration pneumonia usually occurs in patients with compromised states of consciousness or neurologic conditions that predispose these individuals to swallowing dysfunction and aspiration. Ideally, an aspiration episode should be witnessed by medical personnel before pneumonia develops, but such events are difficult to document in most cases. Also, sputum cultures from patients with aspiration pneumonia often reveal anaerobic organisms aspirated from the oropharynx or, even more commonly, polymicrobial infections containing aerobic and anaerobic organisms [19, 20], whereas cultures from patients with other forms of pneumonia are more likely to yield findings of a single organism.

In our study, aspiration events either were not observed or not entered in the medical records, and sputum cultures were obtained in only a few patients. It therefore is difficult to prove that the episodes of pneumonia in our patients were caused by aspiration. Nevertheless, patients with laryngeal penetration, tracheobronchial aspiration, and silent tracheobronchial aspiration were approximately four times, 10 times, and 13 times, respectively, more likely to develop pneumonia than those with normal swallowing. This direct relationship between the degree of swallowing dysfunction seen on the videofluoroscopic studies and the frequency of pneumonia (compared with the baseline frequency of pneumonia in patients with normal swallowing) strongly suggests that aspiration was the underlying cause of the pneumonia in most cases. We therefore believe that the degree of swallowing dysfunction observed on videofluoroscopic swallowing studies directly correlates with the relative risk of developing aspiration pneumonia.

Our study has other limitations. Nearly one third of the patients in our study population did not undergo chest radiography during the 6 months before or after the videofluoroscopic swallowing studies, and we cannot exclude the possibility that one or more of these patients could have had episodes of pneumonia that were not suspected on clinical grounds. This is a legitimate concern because our patients had a mean age of 66 years, and the diagnosis of pneumonia is a particular problem in older patients in whom clinical signs and symptoms can be minimal or absent [21, 22].

Logistic considerations also limited our study. Unfortunately, we are not able to provide long-term storage of the videotapes used for the videofluoroscopic swallowing studies, so we had to rely on the actual radiology reports for the swallowing findings. Because the videofluoroscopic swallowing studies were evaluated by three radiologists, differences in the interpretation of these examinations could have led to inconsistencies in the results. Some of the reports also provided less detailed information about the nature of the swallowing dysfunction than others, limiting our ability to classify these patients with the same degree of confidence.

The videofluoroscopic swallowing studies were excluded from analysis if the patient's swallowing findings were not adequately described in the radiology reports, creating a potential selection bias because these individuals may have been less likely to have swallowing dysfunction. Our study population represents another potential selection bias: patients at our Veterans Administration medical center may have been more likely to develop pneumonia than those in the general population.

Finally, the density of the aspirated barium was not routinely mentioned in the radiology reports, so the relationship between aspiration of high- versus low-density barium and the development of pneumonia could not be evaluated.

Because of the limitations of our study, prospective investigations using uniform radiographic techniques are needed to further elucidate the usefulness of the videofluoroscopic swallowing study as a diagnostic test for identifying patients at risk for the development of pneumonia. Additional studies are also needed to determine whether compensatory maneuvers during swallowing, modification of diet, or even withholding of oral feedings can prevent the development of aspiration pneumonia in these patients.

In conclusion, our findings indicate that the likelihood of developing pneumonia is directly related to the degree of swallowing dysfunction observed on videofluoroscopic studies. Patients with no laryngeal penetration—regardless of whether they have normal or abnormal swallowing—have the lowest risk of developing pneumonia. In contrast, patients with laryngeal penetration, tracheobronchial aspiration, or silent tracheobronchial aspiration are, in increasing order of magnitude, significantly more likely to develop pneumonia than those with normal swallowing. Therefore, findings on videofluoroscopic swallowing studies can be used to guide management of patients potentially at risk for pneumonia.

Address correspondence to M. S. Levine.