Pulse Oximetry Does Not Reliably Detect Aspiration on Videofluoroscopic Swallowing Study

730 Pulse Oximetry Does Not Reliably Detect Aspiration on Videofluoroscopic Swallowing Study Tyng-Guey Wang, MD, Yeun-Chung Chang, MD, Ssu-Yuan Chen, MD, Tzu-Yu Hsiao, MD, PhD ABSTRACT. Wang T-G, Chang Y-C, Chen S-Y, Hsiao T-Y. Pulse oximetry does not reliably detect aspiration on videofluoroscopic swallowing study. Arch Phys Med Rehabil 2005;86:730-4. Objective: To examine the reliability of pulse oximetry for identifying aspiration by comparing it with the videofluoroscopic swallowing study (VFSS). Design: Nonrandomized, prospective, double-blind study. Setting: VFSS laboratory in a teaching hospital. Participants: Sixty patients from among 130 patients with clinically diagnosed dysphagia between September and December 2002. Interventions: Not applicable. Main Outcome Measures: Simultaneously monitoring the arterial oxygen saturation (SpO 2 ) by pulse oximetry while patients were performing VFSS. A decrease in SpO 2 exceeding 3% was considered as significant desaturation. Bolus or portion of bolus passing through the vocal cords and entering the subglottic space was defined as aspiration on VFSS. The results of pulse oximetry and VFSS were compared. Results: No significant correlation existed between desaturation measured by pulse oximetry and aspiration on VFSS ( 2 test, P.87). The positive predictive rate of pulse oximetry in detecting aspiration on VFSS was 39.1%, and the negative predictive rate was 59.4%. Conclusions: Aspiration occurring on VFSS cannot be predicted based on decrease in SpO 2 in pulse oximetry. The application of pulse oximetry to detect aspiration during regular meals requires further investigation. Key Words: Aspiration; Dysphagia; Oximetry; Rehabilitation; Swallowing. 2005 by American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation From the Departments of Physical Medicine & Rehabilitation (Wang, Chen), Medical Imaging (Chang), and Otolaryngology (Hsiao), National Taiwan University Hospital, College of Medicine, and National Taiwan University, Taipei, Taiwan. Supported by the National Science Council of the Republic of China, Taiwan (grant no. NSC91-2314-B-002-359). No commercial party having a direct interest in the results of the research supporting this article has or will confer a benefit on the author(s) or on any organization with which the author(s) is/are associated. Reprint requests to Tyng-Guey Wang, MD, Dept of Physical Medicine & Rehabilitation, National Taiwan University Hospital, 7 Chung-Shan S Rd, Taipei, Taiwan, ROC, e-mail: tgw@ha.mc.ntu.edu.tw. 0003-9993/05/8604-8764$30.00/0 doi:10.1016/j.apmr.2004.10.021 VIDEOFLUOROSCOPIC SWALLOWING study (VFSS) is generally regarded as the criterion standard for assessing oropharyngeal dysphagia, particularly in detecting laryngeal penetration or subglottic aspiration. 1-3 However, VFSS has several unavoidable limitations. VFSS can be performed only on cooperative patients, has radiologic exposure, and involves a single time measurement, yet aspiration may vary during feeding. 4 Various bedside clinical evaluations, such as neurologic examination, 5,6 water swallow test, 7-9 swallowing provocation test, 10 and cough reflex, 11 have been used to identify or predict aspiration. Comparing these techniques with VFSS reveals that they have aspiration prediction rates varying between 35% and 100%. Consequently, researchers continue to seek a noninvasive, long-term monitoring method for detecting aspiration during regular meals. Recently, pulse oximetry has been proposed as an alternative for detecting aspiration. 12-14 Pulse oximetry is a method of measuring arterial oxygen saturation (SpO 2 ) as measured by pulse oximetry, and is accurate to a standard deviation of 2% to 3% down to a saturation of 50%, although errors such as motion artifact, insufficient hemoglobin, and comprised peripheral perfusion may limit its accuracy. 15,16 Use of pulse oximetry to detect aspiration is based on the hypothesis that aspiration of food into the airways causes reflex bronchospasm and thus leads to ventilation-perfusion mismatch and oxygen desaturation. 17,18 Pulse oximetry has numerous advantages in assessing swallowing function if its accuracy in aspiration detection can be proved. Pulse oximetry is noninvasive without exposure to radiation, is readily portable and widely available, requires minimal patient cooperation, and provides objective quantitative measures. Moreover, pulse oximetry provides continuous monitoring of a patient s swallowing condition during the meal, which can be further stored and downloaded to a computer for detailed analysis. However, the accuracy of pulse oximetry for detecting aspiration is debatable. Some researchers have reported a predictive rate of up to 81.5% for pulse oximetry in defining the aspiration, 19,20 but others questioned the correlation between the desaturation measured by oximetry and aspiration on VFSS. 21,22 Our work aimed to characterize the relationship between oxygen desaturation and radiographically detected aspiration to clarify broadly the role of pulse oximetry in assessing swallowing function. METHODS Participants The study had a prospective, double-blind design. Between September and December of 2002, approximately 130 adult patients from inpatient and outpatient populations were referred for VFSS examination because of a clinical suspicion of swallowing disorder. Sixty of these 130 patients were included in our study. Exclusion criteria included presence of peripheral vascular disease (PVD), chronic lung disease, smoking history, and incomplete VFSS study. The sample included 43 men and 17 women, whose mean age was 61 years (range, 19 89y). The main causes of dysphagia were cerebrovascular accident and nasopharyngeal cancer postradiation (table 1). The research was approved by the ethics committee of our hospital. The study procedures were explained to and informed consent was obtained from each patient. Pretest To understand the fluctuation of SpO 2 on normal swallowing, 40 healthy subjects (23 women, 17 men; mean age,

PULSE OXIMETRY DETECTS ASPIRATION, Wang 731 Table 1: Etiology of Swallowing Disorders Etiology n (N 60) Cerebrovascular accident 27 Nasopharyngeal cancer postradiotherapy 10 Motoneuron disease 4 Parkinsonism 4 Brain tumor 2 Multiple sclerosis 2 Traumatic brain injury 2 Esophageal cancer 2 Vocal cord palsy 2 Scleroderma 1 No definite diagnosis 4 40.8 11.7y) underwent monitoring of pulse oximetry while drinking water. None of the patients had a history of smoking, lung disease, PVD, or dysphagia. Subjects had probes fitted to their fingers. The fingers were clean and warm. Continuous SpO 2 readings, taken at 12-second sampling intervals, were obtained via pulse oximetry. a The data were recorded and stored on a compact flash memory card, and off-line analysis was conducted on a personal computer. Baseline SpO 2 was recorded for 10 minutes with subjects sitting comfortably. The subjects were then asked to drink 20mL water twice at a comfortable speed, and SpO 2 was then continuously recorded for 5 minutes. Instrumentation and Methods Before VFSS, all patients received pulse oximetry monitoring, a as described in the previous paragraph. Additionally, patients were instructed to keep still the arm with the probe, to prevent movement artifacts. The timer of pulse oximetry and recorder of VFSS were synchronized before SpO 2 monitoring and VFSS examination. Baseline SpO 2 was determined for a minimum of 5 minutes before the initiation of VFSS. SpO 2 was continuously recorded throughout the VFSS examination and for 5 minutes or more thereafter. A research assistant with no information on the VFSS results read the episodes of desaturation of SpO 2, including the time and duration of occurrence. Our experience and the manufacturer s literature suggested that the error of pulse oximetry is 2% or less. Additionally, the result of the pretest revealed that fluctuation of SpO 2 in healthy subjects was between 1% and 3%. Therefore, a decrease of more than 3% in SpO 2 between the baseline and the lowest level of saturation was considered significant and was recorded as an episode of desaturation. A radiologist used a fluoroscope to perform standardized VFSS with a high-resolution super-vhs recorder. The videotape recorder provided a frame rate of 30 frames per second and could display both real-time dynamic images and frameby-frame static images. Patients were told to sit in a specially designed chair to obtain both lateral and frontal anteroposterior views. Each patient took 2 doses of 3 standardized formulas (5mL each of thin, thick, and paste) of barium sulfate (E-Z- HD). b The thin barium sulfate (suspension of 340g of E-Z-HD and 65mL of water) is standard for routine gastrointestinal examinations. Thick barium was prepared by adding an extra 7.5mL E-Z HD powder to 15mL standard thin barium preparation, and the paste barium was prepared by adding an extra 12mL E-Z-HD powder to 15mL standard thin barium preparation. An experienced radiologist with no information about the pulse oximetry results interpreted the VFSS results. The main focus of interest was the existence of penetration (a bolus or portion of bolus entering the laryngeal vestibule but not passing through the vocal cords) or aspiration (a bolus or portion of bolus passing through the vocal cords and entering the subglottic space). The precise time of test barium swallowing and occurrence of penetration or aspiration was read and recorded for comparing them with the episodes of desaturation recorded by pulse oximetry. A third researcher compared the pulse oximetry and VFSS results. Statistical Analysis The relationship between desaturation and aspiration was examined using the chi-square test. RESULTS On the pretest, fluctuation of SpO 2 was 3% or less in healthy young adults. During a 10-minute rest, 4 of the 40 healthy subjects displayed no change in SpO 2, 10 had a transient desaturation of up to 1%, 18 had transient reduced SpO 2 to 2%, and 8 had a 3% fluctuation. While drinking 20mL water, none of the healthy subjects displayed a decrease of SpO 2 exceeding 2%. Moreover, 15 of these subjects displayed no change of SpO 2, 24 had a temporal 1% reduction of SpO 2, and 1 had a transient 2% reduction of SpO 2. This indicated that the fluctuation of SpO 2 was below 3% in healthy young adults when drinking 20mL water. Twenty-three of the 60 patients were defined as aspirators on VFSS examination that is, these 23 patients had least 1 aspiration event on VFSS examination. Fourteen of the 23 aspirators were classified with silent aspiration (aspiration without cough response). No significant difference was observed in age and baseline SpO 2 between the aspirator and nonaspirator groups. The mean age of the aspirator group was 62.4 14.9 years, whereas that of the nonaspirator group was 59.8 18.4 years (P.54). The baseline mean SpO 2 in the aspirator group was 96.8% 1.0%, whereas that in the nonaspirator group was 96.7% 1.0% (P.77). The fluctuation of SpO 2 in patients before VFSS was below 2%. Before VFSS, the SpO 2 of most patients exceeded 95%, and only 2 patients had transient desaturation with SpO 2 below 95%. Comparing the result of pulse oximetry with the VFSS findings revealed that 9 patients (39.1%) among 23 aspirators displayed significant oxygen desaturation (SpO 2 fall exceeding 3%). Meanwhile, of the 37 patients without aspiration on VFSS examination, 15 (40.5%) had an SpO 2 fall exceeding 3%. Restated, the sensitivity of pulse oximetry for detecting aspiration was 39.1% and the specificity was 59.4%. In the entire study sample, 31 patients (51.7%) were accurately identified, using pulse oximetry, as either aspirators or nonaspirators. The positive predictive rate of pulse oximetry in detecting aspiration was 37.5%; the negative predictive rate was 61.1%. No significant relationship existed between reduction in SpO 2 and aspiration on VFSS using the chi-square test (P.87; table 2). Moreover, 39 episodes of desaturation, defined by pulse oximetry, occurred in 28 patients; and 37 events of aspiration Table 2: Relationship Between SpO 2 Desaturation and Aspiration SpO 2 Aspirator* Nonaspirator Desaturator 9 15 Nondesaturator 14 22 NOTE. P.87 ( 2 test). *Subjects who had at least 1 event of aspiration on VFSS examination. Subjects who had SpO 2 decrement greater than 3% by pulse oximetry on VFSS examination.

732 PULSE OXIMETRY DETECTS ASPIRATION, Wang Table 3: Relationship Between SpO 2 Desaturation and Aspiration VFSS in Stroke Patients SpO 2 Aspirator* Nonaspirator Desaturator 7 5 Nondesaturator 5 10 NOTE. P.36 ( 2 test). *Stroke patients who had at least 1 event of aspiration on VFSS examination. Stroke patients who had SpO 2 decrement greater than 3% by pulse oximetry on VFSS examination. occurred in the 23 aspirators on VFSS. Of the 37 aspiration events, only 11 events (29.7%) were combined with a desaturation episode (desaturation was followed by aspiration within 2min). Of the 39 desaturation episodes, 29 occurred during VFSS examination, 4 occurred before VFSS examination, and 6 occurred after VFSS examination. The duration of desaturation persisted from 12 seconds to as long as 8 minutes. Only 12 of the 39 episodes of desaturation were found to be combined with an aspiration event on VFSS. Twenty-seven of the 60 patients were stroke patients. Twelve of these 27 patients were identified as aspirators on VFSS, and only 7 of these 12 aspirators had significant reduction in SpO 2 (table 3). The sensitivity of pulse oximetry in detecting aspiration in stroke patients thus was 58.3%, and the specificity was 66.7%. The reduction in SpO 2 did not correlate with aspiration on VFSS in stroke patients using the chi-square test (P.36). DISCUSSION Our work objectively assessed the correlation between SpO 2 decline on pulse oximetry and aspiration on VFSS, using a prospective double-blind design. We found no significant correlation between SpO 2 decline on pulse oximetry and aspiration on VFSS using the present setting. SpO 2 reduction exceeding 3% cannot predict the occurrence of aspiration on VFSS and is only 51.7% accurate. These findings contradict most other investigations. 14,19,20,23 Rogers et al 14 described 2 adult cerebral palsy (CP) patients and 1 multiple sclerosis (MS) patient with obvious hypoxemia during regular meals, and in whom hypoxemia occurred only when taking specific foods. Accordingly, Rogers proposed that hypoxemia during feeding may result from mis-swallowing of food into the airway. However, Rogers described only 3 cases and provided no direct evidence or observations to support the cause-result relationship between aspiration and hypoxemia. One of the 3 patients even displayed no aspiration on VFSS. Additionally, either weakness of respiratory muscle or incoordination of respiration may exist in patients with MS or CP, and probably causes hypoxemia without aspiration. Although Rogers work strongly implies that pulse oximetry may be a good method for detecting aspiration, we admit that our study does not enable any direct association between aspiration and hypoxemia. Zaidi et al 19 analyzed the degree of SpO 2 decrement among 3 different risky aspiration groups of stroke patients. The patients were classified as definite aspirators, possible aspirators, and nonaspirators, according to evaluation by a speech pathologist. Zaidi showed that the mean decrease of SpO 2 (1.4%) in the nonaspirator group was significantly less than that in the aspirator group (4.6%) when the patients drank 10mL water. However, the presence of silent aspiration led to questioning of the accuracy of defining aspiration by clinical assessment, which occurred in approximately 40% of dysphagic stroke patients. 24 Some investigators 2,5 have stated that the combination of multiple clinical evaluation tools can increase the accuracy of clinical evaluation in detecting aspiration; however, clinical evaluation is still not accepted as a standard method for defining aspiration. Moreover, Zaidi s work indicates that the aspirator tended to be associated with decreased SpO 2 and does not show correlation between episodes of desaturation and aspiration, because the study only compared the mean SpO 2 in different groups and individual comparisons were not made. Whether the desaturation episode indicates the occurrence of aspiration requires further study. Sherman et al 23 further classified 46 patients with dysphagia into 4 groups namely, aspirated, penetrated but not cleared, penetrated and cleared, and no penetration by VFSS. Sherman demonstrated that patients displaying aspiration or penetration without clearing had a more significant decrease in SpO 2 than patients without aspiration or with penetration but clearing. The result implied that decreased SpO 2 is associated with aspiration and suggested a direct relationship between the amount of decrease in SpO 2 and aspiration severity. However, this investigation still failed to demonstrate the relationship between each episode of desaturation and the event of aspiration. Collins and Bakheit 20 simultaneously monitored SpO 2 in 54 dysphagic stroke patients undergoing VFSS. They compared the level of SpO 2 with patient swallowing condition. The comparison revealed that 73% of patients with aspiration on VFSS were identified by pulse oximetry. Pulse oximetry accurately identified 81.5% of stroke patients with dysphagia, whether or not they had aspiration on VFSS. Collins and Bakheit 20 also concluded that pulse oximetry was a reliable method of diagnosis of aspiration in most dysphagic patients. Their result differed from ours, although both works shared almost the same design. The only procedural differences between the 2 studies involved the amount of test meal on VFSS. Collins and Bakheit 20 used a 150-mL liquid, 3-oz (84-g) mousse, and one half of a 2-in (5.1-cm) barium-impregnated shortbread biscuit as the barium meal. The quantity of test barium meal was much larger than that in a regular modified barium swallowing examination. 3 Generally, 10 to 15mL of barium is recommend for VFSS, to avoid aspiration pneumonia caused by aspiration of test meal. Our work thus used a 5-mL thin, thick, and paste barium twice and stopped the examination if major aspiration occurred. The amount of aspiration in our work probably was inadequate for causing reduced SpO 2, which may partially explain the different results obtained in our work from that of Collins and Bakheit. 20 However, in our experience, most patients referred for VFSS were unable to swallow the large bolus volumes used in the study by Collins and Bakheit. 20 Although they did not mention the severity of dysphagia in their patients, we assumed that the patients in their investigation had relatively mild dysphagia, because they could swallow large volumes. The results of their work may be suitable only for people with mild dysphagia. Sellars et al 22 performed VFSS and pulse oximetry simultaneously in 6 dysphagia patients and obtained similar results to ours; that is, no correlation was found between decreased SpO 2 and aspiration detected on VFSS. The reason for reduced SpO 2 in patients with dysphagia on meal remains unclear. Researchers who believe that aspiration causes decreased SpO 2 have assumed that on aspiration, the ingested material would occlude the airway and reduce lung tissue airflow. 17 Additionally, it has been assumed that the ingested material would stimulate laryngeal chemore-

PULSE OXIMETRY DETECTS ASPIRATION, Wang 733 ceptor, causing bronchospasm. 17 Either reaction would impair ventilation-perfusion impairment and cause further decline of SpO 2. Furthermore, the ingested material stimulated the center chemoreceptor, induced respiratory depression, and caused further decrease in SpO 2. The question is how much of the ingested material was required to cause the above reaction and induce significant reduction in SpO 2. Some studies used 10 to 15mL of barium as a test meal, and the desaturation was still observed with aspiration of as little as 1 to 2mL. Whether as little as 1mL of aspiration will cause desaturation is debatable. Teramoto et al 21 have argued that the presence of more than 50mL of water in 1 lobe of the lung for bronchoalveolar lavage does not always cause significant desaturation. Moreover, Teramoto suggested that the occurrence of desaturation in dysphagic patients during feeding might result from poor coordination between swallowing and breathing rather than from aspiration. According to Teramoto s findings, the quantity of aspiration in routine VFSS, as in the work of Sellars et al 22 and in our own work, should not cause significant reduction in SpO 2. This phenomenon may explain why we found no correlation between decreased SpO 2 on pulse oximetry and aspiration on VFSS. Collins and Bakheit 20 used up to 150mL of test meal; when aspiration occurred, the amount of that aspiration would significantly exceed that in previous reports. The large quantity of aspiration might explain a high correlation between aspiration and reduction in SpO 2 in the work of Collins and Bakheit. 20 The other cause of reduced SpO 2 on eating in patients with dysphagia was poor coordination of breathing during swallowing. 21 Clark 25 found that healthy subjects hold their breath for 0.5 to 0.25 seconds during swallowing when the larynx is elevated. Clark called this phenomenon deglutition apnea. Selley et al 26 further described the breathing pattern in normal swallowing. Selley noted that healthy subjects inhaled first; then a small expiration ensued before swallowing; swallowing then occurred during deglutition apnea; and finally exhaling occurred. Deglutition apnea may involve a delay of up to 1 second in the aged. 27 Moreover, the delay may be further increased to 3.4 seconds in dysphagic stroke patients. 27 Apnea of up to 3.5 seconds on swallowing may cause reduced SpO 2 even without aspiration. Therefore, the decreased SpO 2 was not necessarily caused by aspiration. To clarify the relationship between aspiration and change of SpO 2 requires simultaneously monitoring patients using pulse oximetry, VFSS with different amounts of test barium, and respiratory inductively plethymography. Respiratory plethymography can show whether apnea has occurred when desaturation is detected on pulse oximetry. Combined with VFSS, the cause of reduction in SpO 2 during swallowing can be clarified. Generally, pulse oximetry is a reliable method for measuring arterial oxygen saturation; however, arterial oxygen saturation is not as reliable as blood gas for detecting arterial oxygen tension, owing to its sigmoid-shaped oxygen dissociation curve. Our investigation selected pulse oximetry to define arterial oxygen tension, because it is noninvasive, widely available, and more practical clinically. Patients with respiratory problems may have low SpO 2. For patients with SpO 2 in the low 90s, SpO 2 falls more when aspiration occurs than it does in patients with SpO 2 in the high 90s, owing to the sigmoid-shaped dissociation curve of oxygen saturation. For these reasons, our work excluded the patients with respiratory disease and those with a history of smoking. The patients in our study had a baseline SpO 2 exceeding 95% and had the lowest rates of error from the sigmoid-shaped dissociation curve of oxygen saturation. CONCLUSIONS Our investigation found that the regular amount of test barium on VFSS did not seem to cause desaturation even in the event of aspiration. Decreased SpO 2 exceeding 3% does not indicate aspiration on VFSS. Therefore, the application of pulse oximetry to detect aspiration during regular meals requires further investigation. References 1. Logemann JA. Evaluation and treatment of swallowing disorder. San Diego: College-Hill Pr; 1983. 2. Splaingard ML, Hutchins B, Sulton LD, Chaudhuri G. Aspiration in rehabilitation patients: videofluoroscopy versus bedside clinical assessment. Arch Phys Med Rehabil 1988;69:637-40. 3. Logemann JA. Role of modified barium swallow in management of patients with dysphagia. Otol Head Neck Surg 1997; 116:335-8. 4. Wright FR. A swallow assessment clinic-why and how. Radiology 1991;8:55-7. 5. Lin YN, Wang TG, Chang YC, Hsieh FM, Lien IN. Validation of clinical swallowing evaluation in stroke patients. J Rehabil Med Assoc ROC 1998;26:175-80. 6. Horner J, Massey EW, Riski JE, Lathrop DL, Chase KN. Aspiration following stroke: clinical correlates and outcome. Neurology 1998;38:1359-62. 7. DePippo KL, Holas MA, Reding MJ. Validation of the 3-oz water swallow test for aspiration following stroke. Arch Neurol 1992; 49:1259-61. 8. Garon BR, Engle M, Ormiston C. Reliability of the 3-oz water swallow test utilizing cough reflex as sole indicator of aspiration. J Neurol Rehabil 1995;9:139-43. 9. Tohara H, Saitoh E, Mays K, Kuhlemeier K, Palmer JB. Three tests for predicting aspiration without videofluorography. Dysphagia 2003;18:126-34. 10. Teramoto S, Fukuchi Y. Detection of aspiration and swallowing disorder in older stroke patients: simple swallowing provocation test versus water swallowing test. Arch Phys Med Rehabil 2000; 81:1517-9. 11. Addington WR, Stephens RE, Gilliland K, Rodriguez M. Assessing the laryngeal cough reflex and the risk of developing pneumonia after stroke. Arch Phys Med Rehabil 1999;80:150-4. 12. Brown SE, Cascrari RJ, Light RW. Arterial oxygen saturation during meals in patients with severe obstructive pulmonary disease. South Med J 1983;76:194-8. 13. Garg M, Kuryner SI, Bantista DB, Keens TG. Clinical unspecific hypoxemia during sleep and feeding in infants with bronchopulmonary dysphagia. Pediatrics 1988;81:184-8. 14. Rogers B, Msall M, Shucard D. Hypoxemia during oral feeding in adult with dysphagia and severe neurological disabilities. Dysphagia 1993;8:43-8. 15. Bowers WA, Corke BC, Hulka J. Pulse oximetry: a review of the theory, accuracy, and clinical application. Obstet Gynecol 1989; 74:541-6. 16. Tremper KK, Barker SJ. Pulse oximetry. Anesthesiology 1989; 70:98-108. 17. Colebatch BJ, Halmagyi DF. Reflex airway reaction to fluid aspiration. J Appl Physiol 1962;17:787-94. 18. Rodriguez-Roisin R, Ferrer A, Navajas D, Agusti AG, Wagner PD, Roca J. Ventilation-perfusion mismatch after methacholine challenge in patients with mild bronchial asthma. Am Rev Respir Dis 1991;144:88-94.

734 PULSE OXIMETRY DETECTS ASPIRATION, Wang 19. Zaidi NH, Smith HA, King SC, Park C, O Neill PA, Connolly MJ. Oxygen desaturation on swallowing as a potential marker of aspiration in acute stroke. Age Ageing 1995;24:267-70. 20. Collins MJ, Bakheit AM. Does pulse oximetry reliably detect aspiration in dysphagic stroke patients? Stroke 1997;28: 1773-5. 21. Teramoto S, Fukuchi Y, Ouchi Y. Oxygen desaturation on swallowing in patients with stroke: what does it mean? [letter]. Age Ageing 1996;25:333-4. 22. Sellars C, Dunnet C, Carter R. A preliminary comparison of videofluoroscopy of swallow and pulse oximetry in the identification of aspiration in dysphagic patients. Dysphagia 1998;13:82-6. 23. Sherman B, Nisenboum JM, Jesberger BL, Morrow CA, Jesberger JA. Assessment of dysphagia with the use of pulse oximetry. Dysphagia 1999;14:152-6. 24. Horner J, Massey EW. Silent aspiration following stroke. Neurology 1988;38:317-9. 25. Clark GA. Deglutition apnoea. J Physiol 1920;54:59. 26. Selley WG, Flack FC, Ellis RE, Brooks WA. Respiratory patterns associated with swallowing: part 1. The normal adult pattern and changes with age. Age Ageing 1989;18:168-72. 27. Selley WG, Flack FC, Ellis RE, Brooks WA. Respiratory patterns associated with swallowing: part 2. Neurologically impaired dysphagic patients. Age Ageing 1989;18:173-6. Suppliers a. Capnography; Novometrix Medical System Inc,1 Barnes Industrial Pk Rd, Wallingford, CT 06492. b. E-Z-EM Inc, 1111 Marcus Ave, Ste LL-26, Lake Success, NY 11042.