Radiol Phys Technol DOI 10.1007/s12194-014-0264-3
Method for detection of aspiration based on B-mode video ultrasonography Yuka Miura • Gojiro Nakagami • Koichi Yabunaka Haruka Tohara • Ryoko Murayama • Hiroshi Noguchi • Taketoshi Mori • Hiromi Sanada
•
Received: 14 January 2014 / Revised: 20 March 2014 / Accepted: 24 March 2014 Ó Japanese Society of Radiological Technology and Japan Society of Medical Physics 2014
Abstract In this study, we aimed to develop a new method for detection of aspiration based on B-mode video ultrasonography and to evaluate its performance. To detect aspirated boluses by B-mode video ultrasonography in patients with dysphagia, we placed a linear array transducer above the thyroid cartilage and observed the area around the vocal folds. Forty-two ultrasonographic measurements were obtained from 17 patients with dysphagia who also underwent videofluoroscopy or videoendoscopy measurements at the same time. Aspirated boluses were observed in B-mode Y. Miura G. Nakagami H. Sanada (&) Department of Gerontological Nursing/Wound Care Management, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan e-mail: [email protected] Y. Miura Japan Society for the Promotion of Science, 5-3-1 Kojimachi, Chiyoda-ku, Tokyo 102-0083, Japan K. Yabunaka Department of Ultrasound, Katsuragi Hospital, 2-33-1 Habu-cho, Kishiwada, Osaka 596-0825, Japan H. Tohara Gerodontology and Oral Rehabilitation, Department of Gerontology and Gerodontology, Graduate School of Medical and Dental Sciences, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8549, Japan R. Murayama Department of Advanced Nursing Technology, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan H. Noguchi T. Mori Department of Life Support Technology (Molten), Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
video ultrasonographic images as hyperechoic, long, narrow objects that passed through the vocal folds beneath the anterior wall of the trachea, with movement different from that of the surrounding structure. The sensitivity of aspiration detection was 0.64, and the specificity was 0.84. This newly developed detection method will enable patients with dysphagia to receive appropriate daily swallowing care. Keywords Aspiration pneumonia Deglutition disorders Sensitivity and specificity Ultrasound
1 Introduction Pneumonia is the third leading cause of death in Japan, and aspiration pneumonia accounts for 90 % of cases, which predominantly occurs in elderly people [1]. Aspiration is defined as passage of material below the vocal folds [2]. With repeated aspiration, bacteria within food or saliva are often introduced into the lungs, where they can cause aspiration pneumonia [3]. Most aspiration pneumonia is caused by silent aspiration without producing a reflexive cough [4]. Therefore, a timely detection method for aspiration, including silent aspiration, is needed for preventing pneumonia in elderly people. Convenient screening tests for dysphagia are often used at the bedside, such as the repetitive saliva swallowing test [5], water swallowing test [6], and food test [7]. However, these tests use clinical signs such as coughing as an indicator of the presence of aspiration, and they cannot detect silent aspiration. A cough test, which uses an induced cough reflex by citric acid particle inhalation as a surrogate indicator of silent aspiration, has been proposed as a useful screening test for silent aspiration [8]. Although this test shows high sensitivity and specificity, clinicians cannot
Y. Miura et al.
directly observe silent aspiration. In other words, clinicians cannot know the appropriate viscosity of the bolus and the effective swallowing posture that prevent aspiration. Apart from these screening tests, the development of a method of direct detection of aspiration, including silent aspiration, would also be effective for daily swallowing care. Patients would be able to receive appropriate care according to the detection of aspiration. Videofluoroscopy (VF) and videoendoscopy (VE) are considered to be the most appropriate detection methods for silent aspiration because we can directly observe the passage of a bolus into the trachea [9, 10]. However, these procedures for detection of silent aspiration in patients with dysphagia are sometimes not common in clinical settings because they are invasive and complicated. Therefore, a new noninvasive method for detection of aspiration in a patient’s daily feeding is required. B-mode video ultrasonography (BV-US) appears to be an ideal method for detection of silent aspiration because it is noninvasive and performed in real time. However, passage of a bolus into the trachea has not been detected by BV-US. The trachea contains air, which causes ultrasound attenuation [11]; thus, most researchers believe that it is difficult to detect material in the trachea by BV-US. There are a few reports on US measurement of hyoid bone or geniohyoid muscle movements during swallowing [12–14]. It is difficult to modify food viscosity or change the swallowing posture based on these movements; hence, a new technique for detecting material in the trachea by US is needed. Several reports have included US images of the bolus in the oral cavity during the different phases of oral and pharyngeal swallowing [15–17]. There are some amounts of air present in the oral cavity and the pharynx; therefore, we focused on the region around the vocal folds when imaging aspirated boluses because ultrasound attenuation by air would be minimal in this location. US observation of the vocal folds indicated that the trachea was sufficiently narrow at the vocal folds for visualization of a bolus in the trachea [18]. We hypothesized that an aspirated bolus in the trachea would be observed as a hyperechoic object-like bolus in the oral cavity. In this study, we aimed to develop a new method for detection of aspiration based on BV-US and to evaluate its performance in a clinical setting by comparing the results with those of VF or VE examination.
2 Methods 2.1 Patients and setting This was a cross-sectional observational study conducted from June to November 2012. All data were collected at the dysphagia outpatient clinic of a general hospital in Chiba
prefecture, Japan. Individuals[60 years old who underwent VF or VE examinations were included. Patients who could not undergo BV-US examination at the same time with VF or VE examination were excluded. Age, sex, and disease history were collected from the medical records. Items related to examinations, including type of test food, patient’s posture, aspiration, and cough reflex, were recorded at each examination. The study protocol was approved by the Ethics Committee of the Graduate School of Medicine, The University of Tokyo (#3260). Written informed consent was obtained from all patients or their proxies. 2.2 BV-US examination BV-US examination was performed at the same time as VF or VE examinations by the same operator for all patients. An examination consisted of more than one measurement, and the test food for each measurement was modified during the examination. The test foods consisted of three types as follows: thin liquid, thick liquid, and solid food. The examples of thin liquids are water, milk, and soup. They often cause aspiration, and it is more difficult for patients with dysphagia to swallow such thin liquids compared with thick liquids. Thick liquids represent an increased consistency to allow patients with dysphagia to swallow easily. They are less easy to pour and drizzle from a cup or bowl like honey. Solid foods hold their shape. They cannot be poured and are generally consumed with a spoon. They are easier to swallow than liquids. The solid food was served by an experienced speech-languagehearing therapist (ST) with a spoon to adjust the amount to the patient’s swallowing ability. Thin and thick liquids were also served by a ST with a cup or a syringe. We did not take into account the amount of the test food to analyze the data because the ability of visualization of aspirated boluses in BV-US movies did not appear to be based on their initial sizes, but on their swallowed sizes. The amounts of the aspirated boluses were not different between each VF and VE image. Bedside US (M-Turbo; Sonosite, Bothwell, WA, USA) was performed with use of a 6–15 MHz (HFL509) linear array transducer. The head and neck positions of the patients were not fixed, to facilitate swallowing in their most comfortable posture. The transducer for the longitudinal scan was placed above the thyroid cartilage (Fig. 1a). The patient was asked to speak for detection of vibrations of the vocal folds on the display screen for determining the landmark during swallowing (Fig. 1b). Before each measurement, the transducer was placed at the vocal folds shown in the middle of the display screen, with the cranial side on the left. The images were captured on a hard disk. The operator held the transducer to keep the initial location during swallowing and ensured that the swallowing movements were not
Method for detection of aspiration Fig. 1 a Setting of the transducer during scanning of swallowing in the sagittal plane. b Ultrasonographic image of the vocal folds. The cranial side is the left side, and the skin surface is the upper side
Fig. 2 a Image of an aspirated bolus by ultrasonographic (US) examination. b An image of an aspirated bolus by videofluoroscopic examination that was obtained simultaneously with US examination
disturbed. An operator obtained BV-US movies of swallowing within 6–30 s at a rate of 30 frames per second. We decided to use this frame rate to correspond to the VF and VE examination. Image acquisition settings for the ultrasound machine, such as focus, frequency, and zoom, were maintained for all measurements for standardization of the image quality. We had to focus on a trachea that was located above the esophagus to detect aspirated boluses. The esophagus was always visualized within 4 cm from the contact surface with a probe in BV-US images. Therefore, we decided on a scanning depth of 4 cm from the skin in order to visualize all structures of the trachea. The echo gain and dynamic range were tuned to a proper level for each measurement. The US operator evaluated the aspiration without referring to the VF or VE results to avoid any bias. Aspiration on BV-US was interpreted as passage of a hyperechoic object through the vocal folds beneath the tracheal anterior wall, with the movement different from that of the surrounding structure (Fig. 2a). Twenty BV-US measurements were obtained at the same time as VF, and the remaining 31 BV-US measurements were obtained with VE. Two patients (six measurements) were excluded because a tracheostomy hole or a labored cough caused difficulty in placement of the transducer at an appropriate position during the examination. In addition, three measurements were excluded after a discussion with an experienced sonographer because of high attenuation caused by
air between the transducer and skin. Finally, 42 measurements from 17 patients were analyzed. 2.3 VF and VE examinations All patients received VF (DI Station ADR-1000A; Toshiba Medical Systems, Tokyo, Japan) or VE (FNL-10RBS; PENTAX MEDICAL, Ontario, Canada) assessment of swallowing simultaneously with BV-US by experienced dentists (Fig. 2b). We used barium sulfate for VF examinations as a contrast medium, and we used a coloring agent for VE examinations. Both additives are commonly used in these examinations, and both methods are widely used as gold standards in clinical settings. In particular, previous studies revealed that there was a good agreement between these two methods in terms of detection of aspiration [9, 19]. The dentists decided on the examination protocol and the number of measurements that should be performed. They changed the test food viscosity and measurements if necessary. An experienced dentist evaluated aspiration in each VF and VE measurement. 2.4 Data analysis For comparing demographic data, patients who aspirated at least once during the examination were classified into the ‘‘aspiration’’ group, whereas others were classified into the
Y. Miura et al.
‘‘without-aspiration’’ group. A t test or Fisher’s exact test was used for comparison of the patient characteristics between the ‘‘aspiration’’ and ‘‘without-aspiration’’ groups. We used the results of the VF or VE examinations as the reference to calculate the sensitivity and specificity of aspiration detection for each BV-US measurement. It was important to evaluate aspiration for each measurement rather than for each patient because the operators changed the test food for the patients between measurements. Therefore, the performance of aspiration detection by BVUS on a measurement basis was evaluated. Two separate measurements were obtained from all patients at an interval of 1 month. The repeatability of the aspiration detection method was assessed by calculation of the kappa coefficient for the two measurements for each patient. The statistical significance level was set at p \ 0.05. Statistical Analysis System software version 9.2 (SAS Institute, Cary, NC, USA) was used for all statistical analyses.
Table 1 Demographic data Aspiration N=8
Without aspiration N=9
p value
Age (years), mean ± SD
71 ± 9.2
69 ± 6.2
0.583a
Male, n (%)
8 (100)
8 (88.9)
0.307b
Stroke
5 (62.5)
4 (44.4)
0.637b
Pneumonia
0 (0.0)
2 (22.2)
0.471b
Head and neck cancer
0 (0.0)
2 (22.2)
0.471b
Others Silent aspiration
3 (37.5) 6 (75.0)
1 (11.1) –
0.294a –
Disease, n (%)
a
t test
b
Fisher’s exact test
Table 2 Results of the aspiration detection method based on BV-US N = 42 (measurements)
3 Results This study included 17 patients (16 males), with an average age of 70 ± 7.6 years. There were no differences in the patient characteristics between the aspiration and withoutaspiration groups. All of the 8 subjects in the aspiration group were males. They had the following diseases: Five had a stroke, two had Parkinson’s disease, and one had amyotrophic lateral sclerosis. Six patients had silent aspiration at least once during the examination (Table 1). Eleven BV-US measurements were diagnosed as aspiration with reference to the VF/VE measurements. Among the seven measurements that were detected correctly as aspiration by BV-US measurements, five were obtained from stroke patients and one each was obtained from a patient with Parkinson’s disease and a patient with amyotrophic lateral sclerosis. Among the four measurements that were not detected correctly as aspiration by BV-US measurements, two were obtained from stroke patients and one each was obtained from a patient with Parkinson’s disease and a patient with amyotrophic lateral sclerosis. All of the measurements that were not detected as aspiration by BVUS measurements were obtained with thin-liquid swallowing. Among the seven measurements that were correctly detected by BV-US measurements as aspiration, two were obtained from solid food, two were obtained from thick liquid, and three were obtained from thin liquid. Among the seven measurements that were detected correctly by US measurements as aspiration, three measurements were obtained by VF examination and four were obtained by VE examination. Table 2 shows the results of the detection of the aspiration based on BV-US measurements. The sensitivity and specificity of the aspiration
BV-US
Aspirated Not aspirated
VF/VE Aspirated
Not aspirated
7 4
5 26
VF Videofluoroscopy, VE videoendoscopy, BV-US B-mode video ultrasonography
detection method based on BV-US were 0.64 and 0.84, respectively. The kappa coefficient was 0.66.
4 Discussion This is the first study to show that an aspirated bolus in the trachea was observed as a hyperechoic, long, narrow object moving along the trachea wall by BV-US examination. In addition, this is the first study to use comparisons with reference test findings to confirm the usefulness of BV-US examination findings. We focused directly on the vocal folds to observe aspirated boluses. There were no marked differences in the location of the vocal folds and the characteristics of the surrounding soft tissues among the patients who had various diseases, which allowed us to detect aspiration in BVUS images from these patients. A previous method that measured surrounding muscle movement by B-mode US showed that the extent of the differences varied widely between different diseases [20]. Moreover, a sex difference has been reported in the movement of the surrounding muscle [13]. Thus, it is difficult to detect abnormal swallowing in a variety of patients by movement of the surrounding structures. In male subjects, it appears to be more difficult to observe material in the trachea because males have a larger laryngeal prominence than female subjects,
Method for detection of aspiration
which causes attenuation of US. In this study, all of the aspirated patients were male, and the aspirated boluses were detected successfully in the BV-US images. Therefore, it appears that this aspiration detection method can be applied to both males and females. In previous studies, the sensitivity and specificity of a repetitive saliva swallowing test were 0.28 and 0.76, respectively [5]; the sensitivity and specificity of a 100-ml water swallowing test using choking or a wet-hoarse voice as the sole factor for predicting the presence of aspiration were 0.48 and 0.92, respectively; [6], and the sensitivity and specificity of a food test were 0.72 and 0.62, respectively [7]. In this study, the BV-US-based aspiration detection method gave a sensitivity of 0.64 and a specificity of 0.84 when we interpreted a hyperechoic, long, narrow object moving along the tracheal wall in the BV-US images as an indicator of the presence of aspiration. We believe that the good results for sensitivity and specificity were achieved by direct observation of the bolus in the trachea. We correctly detected aspirated boluses in four of the seven images that were obtained from silent aspiration cases. We consider that the scanning methods we used were appropriate, but the reduction in the sensitivity occurred by the aspiration of thin liquids. When we separately ran the analysis for thin liquids, the sensitivity and specificity were 0.43 and 0.81, respectively. On the other hand, the sensitivity and specificity for detecting aspirated thick liquid or solid food by BV-US were 1.00 and 0.81, respectively. These results indicate that this scanning method is optimal at least for swallowing of thick liquid and solid food. Four images were not detected as aspiration by the BV-US method. The reason is related to the speed of the bolus when it passes through the larynx, which differs according to the grade of bolus viscosity [21]. All of the measurements in which we could not detect aspiration from the BV-US images were caused by swallowing of thin liquid. It was difficult to detect such boluses from the BV-US measurements because the duration of the thin-liquid passage through the BV-US scan area was relatively short. The objects that we could detect as aspirated boluses in the BVUS measurements had characteristic movements and shapes. Therefore, if we can identify quantitative characteristics of the object, we will be able to develop some image-processing method that helps our evaluation in the BV-US measurements. In future studies, we hope to develop an image-processing method that can improve the sensitivity and specificity of the BV-US measurements. The results of this study suggest that the newly developed noninvasive detection method would help clinicians provide daily appropriate swallowing care. BV-US measurements performed simultaneously with VE measurements could detect bolus aspiration. The
BV-US performance of aspirated bolus detection was not different between the measurements obtained by VF and those obtained by VE. Thus, special test foods such as barium liquid are not required for the BV-US-based method. Patients can undergo examination for detection of aspiration in their hospital room with their usual feeding. Consequently, clinicians will be able to detect aspirations that have been overlooked by lack of timely detection methods and silent aspirations that have been overlooked by bedside screening tests. The limitation of this study was that the performance of evaluation was based on BV-US measurements obtained by one operator. The BV-US operation technique has an influence on the evaluation of measurements. Future studies are required for establishment of the necessary training in the BV-US operation technique that can assure reproducible detection of aspiration among various clinicians. 5 Conclusion In this study, a new method based on BV-US for detection of aspiration, including silent aspiration, was developed. The sensitivity and specificity of the aspiration detection method were 0.64 and 0.84, respectively, when we interpreted a hyperechoic long, narrow object moving along the tracheal wall in the images as an indicator of an aspirated bolus. Acknowledgments This study was funded by a grant-in-aid for Challenging Exploratory Research from the Japan Society for the Promotion of Science to Hiromi Sanada (Grant No. 23659999). We thank the director of the Katsuragi Hospital, Dr. Mutsumi Ohue, who provided advice and assistance during the study. Furthermore, we would like to express our sincere thanks to all of the dysphagia patients who participated in this study, and to the clinical staff who supported the data collection. Conflict of interest The authors declare that they had no conflicts of interest related to this study.
References 1. Ministry of Health Labour and Welfare. Vital statistics. http:// www.mhlw.go.jp/toukei/saikin/hw/jinkou/geppo/nengai11/index. html. Accessed 10 Jan 2013. 2. Rosenbek JC, Robbins JA, Roecker EB, Coyle JL, Wood JL. A penetration-aspiration scale. Dysphagia. 1996;11:93–8. 3. Ramsey D, Smithard D, Kalra L. Silent aspiration: what do we know? Dysphagia. 2005;20:218–25. 4. Kikuchi R, Watabe N, Konno T, Mishina N, Sekizawa K, Sasaki H. High incidence of silent aspiration in elderly patients with community-acquired pneumonia. Am J Respir Crit Care Med. 1994;150:251–3. 5. Nishiwaki K, Tsuji T, Liu M, Hase K, Tanaka N, Fujiwara T. Identification of a simple screening tool for dysphagia in patients with stroke using factor analysis of multiple dysphagia variables. J Rehabil Med. 2005;37:247–51.
Y. Miura et al. 6. Wu MC, Chang YC, Wang TG, Lin LC. Evaluating swallowing dysfunction using a 100-ml water swallowing test. Dysphagia. 2004;19:43–7. 7. Tohara H, Saitoh E, Mays KA, Kuhlemeier K, Palmer JB. Three tests for predicting aspiration without videofluorography. Dysphagia. 2003;18:126–34. 8. Sato M, Tohara H, Iida T, Wada S, Inoue M, Ueda K. Simplified cough test for screening silent aspiration. Arch Phys Med Rehabil. 2012;93:1982–6. 9. Wu CH, Hsiao TY, Chen JC, Chang YC, Lee SY. Evaluation of swallowing safety with fiberoptic endoscope: comparison with videofluoroscopic technique. Laryngoscope. 1997;107:396–401. 10. Smith CH, Logemann JA, Colangelo LA, Rademaker AW, Pauloski BR. Incidence and patient characteristics associated with silent aspiration in the acute care setting. Dysphagia. 1999;14:1–7. 11. Merritt CRB. Physics of ultrasound. In: Rumack CM, Wilson SR, Charboneau JW, Levine D, editors. Diagnostic ultrasound. Philadelphia: Elsevier mosby; 2011. p. 2–33. 12. Huang YL, Hsieh SF, Chang YC, Chen HC, Wang TG. Ultrasonographic evaluation of hyoid-larynx approximation in dysphagic stroke patients. Ultrasound Med Biol. 2009;35:1103–8. 13. Yabunaka K, Konishi H, Nakagami G, Sanada H, Iizaka S, Sanada S, Ohue M. Ultrasonographic evaluation of geniohyoid muscle movement during swallowing: a study on healthy adults of various ages. Radiol Phys Technol. 2012;5:34–9. 14. Yabunaka K, Sanada H, Sanada S, Konishi H, Hashimoto T, Yatake H, Yamamoto K, Katsuda T, Ohue M. Sonographic
15.
16.
17.
18.
19.
20.
21.
assessment of hyoid bone movement during swallowing: a study of normal adults with advancing age. Radiol Phys Technol. 2011;4:73–7. Casas MJ, Seo AH, Kenny DJ. Sonographic examination of the oral phase of swallowing: bolus image enhancement. J Clin Ultrasound. 2002;30:83–7. Geddes DT, Chadwick LM, Kent JC, Garbin CP, Hartmann PE. Ultrasound imaging of infant swallowing during breast-feeding. Dysphagia. 2010;25:183–91. Tamburrini S, Solazzo A, Sagnelli A, Del Vecchio L, Reginelli A, Monsorro M, Grassi R. Amyotrophic lateral sclerosis: sonographic evaluation of dysphagia. Radiol Med. 2010;115:784–93. Hu Q, Zhu SY, Luo F, Gao Y, Yang XY. High-frequency sonographic measurements of true and false vocal cords. J Ultrasound Med. 2010;29:1023–30. Langmore SE, Schatz K, Olson N. Endoscopic and videofluoroscopic evaluations of swallowing and aspiration. Ann Otol Rhinol Laryngol. 1991;100:678–81. Kuhl V, Eicke BM, Dieterich M, Urban PP. Sonographic analysis of laryngeal elevation during swallowing. J Neurol. 2003;250:333–7. Choi KH, Ryu JS, Kim MY, Kang JY, Yoo SD. Kinematic analysis of dysphagia: significant parameters of aspiration related to bolus viscosity. Dysphagia. 2011;26:392–8.
Method for detection of aspiration based on B-mode video ultrasonography Yuka Miura • Gojiro Nakagami • Koichi Yabunaka Haruka Tohara • Ryoko Murayama • Hiroshi Noguchi • Taketoshi Mori • Hiromi Sanada
•
Received: 14 January 2014 / Revised: 20 March 2014 / Accepted: 24 March 2014 Ó Japanese Society of Radiological Technology and Japan Society of Medical Physics 2014
Abstract In this study, we aimed to develop a new method for detection of aspiration based on B-mode video ultrasonography and to evaluate its performance. To detect aspirated boluses by B-mode video ultrasonography in patients with dysphagia, we placed a linear array transducer above the thyroid cartilage and observed the area around the vocal folds. Forty-two ultrasonographic measurements were obtained from 17 patients with dysphagia who also underwent videofluoroscopy or videoendoscopy measurements at the same time. Aspirated boluses were observed in B-mode Y. Miura G. Nakagami H. Sanada (&) Department of Gerontological Nursing/Wound Care Management, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan e-mail: [email protected] Y. Miura Japan Society for the Promotion of Science, 5-3-1 Kojimachi, Chiyoda-ku, Tokyo 102-0083, Japan K. Yabunaka Department of Ultrasound, Katsuragi Hospital, 2-33-1 Habu-cho, Kishiwada, Osaka 596-0825, Japan H. Tohara Gerodontology and Oral Rehabilitation, Department of Gerontology and Gerodontology, Graduate School of Medical and Dental Sciences, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8549, Japan R. Murayama Department of Advanced Nursing Technology, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan H. Noguchi T. Mori Department of Life Support Technology (Molten), Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
video ultrasonographic images as hyperechoic, long, narrow objects that passed through the vocal folds beneath the anterior wall of the trachea, with movement different from that of the surrounding structure. The sensitivity of aspiration detection was 0.64, and the specificity was 0.84. This newly developed detection method will enable patients with dysphagia to receive appropriate daily swallowing care. Keywords Aspiration pneumonia Deglutition disorders Sensitivity and specificity Ultrasound
1 Introduction Pneumonia is the third leading cause of death in Japan, and aspiration pneumonia accounts for 90 % of cases, which predominantly occurs in elderly people [1]. Aspiration is defined as passage of material below the vocal folds [2]. With repeated aspiration, bacteria within food or saliva are often introduced into the lungs, where they can cause aspiration pneumonia [3]. Most aspiration pneumonia is caused by silent aspiration without producing a reflexive cough [4]. Therefore, a timely detection method for aspiration, including silent aspiration, is needed for preventing pneumonia in elderly people. Convenient screening tests for dysphagia are often used at the bedside, such as the repetitive saliva swallowing test [5], water swallowing test [6], and food test [7]. However, these tests use clinical signs such as coughing as an indicator of the presence of aspiration, and they cannot detect silent aspiration. A cough test, which uses an induced cough reflex by citric acid particle inhalation as a surrogate indicator of silent aspiration, has been proposed as a useful screening test for silent aspiration [8]. Although this test shows high sensitivity and specificity, clinicians cannot
Y. Miura et al.
directly observe silent aspiration. In other words, clinicians cannot know the appropriate viscosity of the bolus and the effective swallowing posture that prevent aspiration. Apart from these screening tests, the development of a method of direct detection of aspiration, including silent aspiration, would also be effective for daily swallowing care. Patients would be able to receive appropriate care according to the detection of aspiration. Videofluoroscopy (VF) and videoendoscopy (VE) are considered to be the most appropriate detection methods for silent aspiration because we can directly observe the passage of a bolus into the trachea [9, 10]. However, these procedures for detection of silent aspiration in patients with dysphagia are sometimes not common in clinical settings because they are invasive and complicated. Therefore, a new noninvasive method for detection of aspiration in a patient’s daily feeding is required. B-mode video ultrasonography (BV-US) appears to be an ideal method for detection of silent aspiration because it is noninvasive and performed in real time. However, passage of a bolus into the trachea has not been detected by BV-US. The trachea contains air, which causes ultrasound attenuation [11]; thus, most researchers believe that it is difficult to detect material in the trachea by BV-US. There are a few reports on US measurement of hyoid bone or geniohyoid muscle movements during swallowing [12–14]. It is difficult to modify food viscosity or change the swallowing posture based on these movements; hence, a new technique for detecting material in the trachea by US is needed. Several reports have included US images of the bolus in the oral cavity during the different phases of oral and pharyngeal swallowing [15–17]. There are some amounts of air present in the oral cavity and the pharynx; therefore, we focused on the region around the vocal folds when imaging aspirated boluses because ultrasound attenuation by air would be minimal in this location. US observation of the vocal folds indicated that the trachea was sufficiently narrow at the vocal folds for visualization of a bolus in the trachea [18]. We hypothesized that an aspirated bolus in the trachea would be observed as a hyperechoic object-like bolus in the oral cavity. In this study, we aimed to develop a new method for detection of aspiration based on BV-US and to evaluate its performance in a clinical setting by comparing the results with those of VF or VE examination.
2 Methods 2.1 Patients and setting This was a cross-sectional observational study conducted from June to November 2012. All data were collected at the dysphagia outpatient clinic of a general hospital in Chiba
prefecture, Japan. Individuals[60 years old who underwent VF or VE examinations were included. Patients who could not undergo BV-US examination at the same time with VF or VE examination were excluded. Age, sex, and disease history were collected from the medical records. Items related to examinations, including type of test food, patient’s posture, aspiration, and cough reflex, were recorded at each examination. The study protocol was approved by the Ethics Committee of the Graduate School of Medicine, The University of Tokyo (#3260). Written informed consent was obtained from all patients or their proxies. 2.2 BV-US examination BV-US examination was performed at the same time as VF or VE examinations by the same operator for all patients. An examination consisted of more than one measurement, and the test food for each measurement was modified during the examination. The test foods consisted of three types as follows: thin liquid, thick liquid, and solid food. The examples of thin liquids are water, milk, and soup. They often cause aspiration, and it is more difficult for patients with dysphagia to swallow such thin liquids compared with thick liquids. Thick liquids represent an increased consistency to allow patients with dysphagia to swallow easily. They are less easy to pour and drizzle from a cup or bowl like honey. Solid foods hold their shape. They cannot be poured and are generally consumed with a spoon. They are easier to swallow than liquids. The solid food was served by an experienced speech-languagehearing therapist (ST) with a spoon to adjust the amount to the patient’s swallowing ability. Thin and thick liquids were also served by a ST with a cup or a syringe. We did not take into account the amount of the test food to analyze the data because the ability of visualization of aspirated boluses in BV-US movies did not appear to be based on their initial sizes, but on their swallowed sizes. The amounts of the aspirated boluses were not different between each VF and VE image. Bedside US (M-Turbo; Sonosite, Bothwell, WA, USA) was performed with use of a 6–15 MHz (HFL509) linear array transducer. The head and neck positions of the patients were not fixed, to facilitate swallowing in their most comfortable posture. The transducer for the longitudinal scan was placed above the thyroid cartilage (Fig. 1a). The patient was asked to speak for detection of vibrations of the vocal folds on the display screen for determining the landmark during swallowing (Fig. 1b). Before each measurement, the transducer was placed at the vocal folds shown in the middle of the display screen, with the cranial side on the left. The images were captured on a hard disk. The operator held the transducer to keep the initial location during swallowing and ensured that the swallowing movements were not
Method for detection of aspiration Fig. 1 a Setting of the transducer during scanning of swallowing in the sagittal plane. b Ultrasonographic image of the vocal folds. The cranial side is the left side, and the skin surface is the upper side
Fig. 2 a Image of an aspirated bolus by ultrasonographic (US) examination. b An image of an aspirated bolus by videofluoroscopic examination that was obtained simultaneously with US examination
disturbed. An operator obtained BV-US movies of swallowing within 6–30 s at a rate of 30 frames per second. We decided to use this frame rate to correspond to the VF and VE examination. Image acquisition settings for the ultrasound machine, such as focus, frequency, and zoom, were maintained for all measurements for standardization of the image quality. We had to focus on a trachea that was located above the esophagus to detect aspirated boluses. The esophagus was always visualized within 4 cm from the contact surface with a probe in BV-US images. Therefore, we decided on a scanning depth of 4 cm from the skin in order to visualize all structures of the trachea. The echo gain and dynamic range were tuned to a proper level for each measurement. The US operator evaluated the aspiration without referring to the VF or VE results to avoid any bias. Aspiration on BV-US was interpreted as passage of a hyperechoic object through the vocal folds beneath the tracheal anterior wall, with the movement different from that of the surrounding structure (Fig. 2a). Twenty BV-US measurements were obtained at the same time as VF, and the remaining 31 BV-US measurements were obtained with VE. Two patients (six measurements) were excluded because a tracheostomy hole or a labored cough caused difficulty in placement of the transducer at an appropriate position during the examination. In addition, three measurements were excluded after a discussion with an experienced sonographer because of high attenuation caused by
air between the transducer and skin. Finally, 42 measurements from 17 patients were analyzed. 2.3 VF and VE examinations All patients received VF (DI Station ADR-1000A; Toshiba Medical Systems, Tokyo, Japan) or VE (FNL-10RBS; PENTAX MEDICAL, Ontario, Canada) assessment of swallowing simultaneously with BV-US by experienced dentists (Fig. 2b). We used barium sulfate for VF examinations as a contrast medium, and we used a coloring agent for VE examinations. Both additives are commonly used in these examinations, and both methods are widely used as gold standards in clinical settings. In particular, previous studies revealed that there was a good agreement between these two methods in terms of detection of aspiration [9, 19]. The dentists decided on the examination protocol and the number of measurements that should be performed. They changed the test food viscosity and measurements if necessary. An experienced dentist evaluated aspiration in each VF and VE measurement. 2.4 Data analysis For comparing demographic data, patients who aspirated at least once during the examination were classified into the ‘‘aspiration’’ group, whereas others were classified into the
Y. Miura et al.
‘‘without-aspiration’’ group. A t test or Fisher’s exact test was used for comparison of the patient characteristics between the ‘‘aspiration’’ and ‘‘without-aspiration’’ groups. We used the results of the VF or VE examinations as the reference to calculate the sensitivity and specificity of aspiration detection for each BV-US measurement. It was important to evaluate aspiration for each measurement rather than for each patient because the operators changed the test food for the patients between measurements. Therefore, the performance of aspiration detection by BVUS on a measurement basis was evaluated. Two separate measurements were obtained from all patients at an interval of 1 month. The repeatability of the aspiration detection method was assessed by calculation of the kappa coefficient for the two measurements for each patient. The statistical significance level was set at p \ 0.05. Statistical Analysis System software version 9.2 (SAS Institute, Cary, NC, USA) was used for all statistical analyses.
Table 1 Demographic data Aspiration N=8
Without aspiration N=9
p value
Age (years), mean ± SD
71 ± 9.2
69 ± 6.2
0.583a
Male, n (%)
8 (100)
8 (88.9)
0.307b
Stroke
5 (62.5)
4 (44.4)
0.637b
Pneumonia
0 (0.0)
2 (22.2)
0.471b
Head and neck cancer
0 (0.0)
2 (22.2)
0.471b
Others Silent aspiration
3 (37.5) 6 (75.0)
1 (11.1) –
0.294a –
Disease, n (%)
a
t test
b
Fisher’s exact test
Table 2 Results of the aspiration detection method based on BV-US N = 42 (measurements)
3 Results This study included 17 patients (16 males), with an average age of 70 ± 7.6 years. There were no differences in the patient characteristics between the aspiration and withoutaspiration groups. All of the 8 subjects in the aspiration group were males. They had the following diseases: Five had a stroke, two had Parkinson’s disease, and one had amyotrophic lateral sclerosis. Six patients had silent aspiration at least once during the examination (Table 1). Eleven BV-US measurements were diagnosed as aspiration with reference to the VF/VE measurements. Among the seven measurements that were detected correctly as aspiration by BV-US measurements, five were obtained from stroke patients and one each was obtained from a patient with Parkinson’s disease and a patient with amyotrophic lateral sclerosis. Among the four measurements that were not detected correctly as aspiration by BV-US measurements, two were obtained from stroke patients and one each was obtained from a patient with Parkinson’s disease and a patient with amyotrophic lateral sclerosis. All of the measurements that were not detected as aspiration by BVUS measurements were obtained with thin-liquid swallowing. Among the seven measurements that were correctly detected by BV-US measurements as aspiration, two were obtained from solid food, two were obtained from thick liquid, and three were obtained from thin liquid. Among the seven measurements that were detected correctly by US measurements as aspiration, three measurements were obtained by VF examination and four were obtained by VE examination. Table 2 shows the results of the detection of the aspiration based on BV-US measurements. The sensitivity and specificity of the aspiration
BV-US
Aspirated Not aspirated
VF/VE Aspirated
Not aspirated
7 4
5 26
VF Videofluoroscopy, VE videoendoscopy, BV-US B-mode video ultrasonography
detection method based on BV-US were 0.64 and 0.84, respectively. The kappa coefficient was 0.66.
4 Discussion This is the first study to show that an aspirated bolus in the trachea was observed as a hyperechoic, long, narrow object moving along the trachea wall by BV-US examination. In addition, this is the first study to use comparisons with reference test findings to confirm the usefulness of BV-US examination findings. We focused directly on the vocal folds to observe aspirated boluses. There were no marked differences in the location of the vocal folds and the characteristics of the surrounding soft tissues among the patients who had various diseases, which allowed us to detect aspiration in BVUS images from these patients. A previous method that measured surrounding muscle movement by B-mode US showed that the extent of the differences varied widely between different diseases [20]. Moreover, a sex difference has been reported in the movement of the surrounding muscle [13]. Thus, it is difficult to detect abnormal swallowing in a variety of patients by movement of the surrounding structures. In male subjects, it appears to be more difficult to observe material in the trachea because males have a larger laryngeal prominence than female subjects,
Method for detection of aspiration
which causes attenuation of US. In this study, all of the aspirated patients were male, and the aspirated boluses were detected successfully in the BV-US images. Therefore, it appears that this aspiration detection method can be applied to both males and females. In previous studies, the sensitivity and specificity of a repetitive saliva swallowing test were 0.28 and 0.76, respectively [5]; the sensitivity and specificity of a 100-ml water swallowing test using choking or a wet-hoarse voice as the sole factor for predicting the presence of aspiration were 0.48 and 0.92, respectively; [6], and the sensitivity and specificity of a food test were 0.72 and 0.62, respectively [7]. In this study, the BV-US-based aspiration detection method gave a sensitivity of 0.64 and a specificity of 0.84 when we interpreted a hyperechoic, long, narrow object moving along the tracheal wall in the BV-US images as an indicator of the presence of aspiration. We believe that the good results for sensitivity and specificity were achieved by direct observation of the bolus in the trachea. We correctly detected aspirated boluses in four of the seven images that were obtained from silent aspiration cases. We consider that the scanning methods we used were appropriate, but the reduction in the sensitivity occurred by the aspiration of thin liquids. When we separately ran the analysis for thin liquids, the sensitivity and specificity were 0.43 and 0.81, respectively. On the other hand, the sensitivity and specificity for detecting aspirated thick liquid or solid food by BV-US were 1.00 and 0.81, respectively. These results indicate that this scanning method is optimal at least for swallowing of thick liquid and solid food. Four images were not detected as aspiration by the BV-US method. The reason is related to the speed of the bolus when it passes through the larynx, which differs according to the grade of bolus viscosity [21]. All of the measurements in which we could not detect aspiration from the BV-US images were caused by swallowing of thin liquid. It was difficult to detect such boluses from the BV-US measurements because the duration of the thin-liquid passage through the BV-US scan area was relatively short. The objects that we could detect as aspirated boluses in the BVUS measurements had characteristic movements and shapes. Therefore, if we can identify quantitative characteristics of the object, we will be able to develop some image-processing method that helps our evaluation in the BV-US measurements. In future studies, we hope to develop an image-processing method that can improve the sensitivity and specificity of the BV-US measurements. The results of this study suggest that the newly developed noninvasive detection method would help clinicians provide daily appropriate swallowing care. BV-US measurements performed simultaneously with VE measurements could detect bolus aspiration. The
BV-US performance of aspirated bolus detection was not different between the measurements obtained by VF and those obtained by VE. Thus, special test foods such as barium liquid are not required for the BV-US-based method. Patients can undergo examination for detection of aspiration in their hospital room with their usual feeding. Consequently, clinicians will be able to detect aspirations that have been overlooked by lack of timely detection methods and silent aspirations that have been overlooked by bedside screening tests. The limitation of this study was that the performance of evaluation was based on BV-US measurements obtained by one operator. The BV-US operation technique has an influence on the evaluation of measurements. Future studies are required for establishment of the necessary training in the BV-US operation technique that can assure reproducible detection of aspiration among various clinicians. 5 Conclusion In this study, a new method based on BV-US for detection of aspiration, including silent aspiration, was developed. The sensitivity and specificity of the aspiration detection method were 0.64 and 0.84, respectively, when we interpreted a hyperechoic long, narrow object moving along the tracheal wall in the images as an indicator of an aspirated bolus. Acknowledgments This study was funded by a grant-in-aid for Challenging Exploratory Research from the Japan Society for the Promotion of Science to Hiromi Sanada (Grant No. 23659999). We thank the director of the Katsuragi Hospital, Dr. Mutsumi Ohue, who provided advice and assistance during the study. Furthermore, we would like to express our sincere thanks to all of the dysphagia patients who participated in this study, and to the clinical staff who supported the data collection. Conflict of interest The authors declare that they had no conflicts of interest related to this study.
References 1. Ministry of Health Labour and Welfare. Vital statistics. http:// www.mhlw.go.jp/toukei/saikin/hw/jinkou/geppo/nengai11/index. html. Accessed 10 Jan 2013. 2. Rosenbek JC, Robbins JA, Roecker EB, Coyle JL, Wood JL. A penetration-aspiration scale. Dysphagia. 1996;11:93–8. 3. Ramsey D, Smithard D, Kalra L. Silent aspiration: what do we know? Dysphagia. 2005;20:218–25. 4. Kikuchi R, Watabe N, Konno T, Mishina N, Sekizawa K, Sasaki H. High incidence of silent aspiration in elderly patients with community-acquired pneumonia. Am J Respir Crit Care Med. 1994;150:251–3. 5. Nishiwaki K, Tsuji T, Liu M, Hase K, Tanaka N, Fujiwara T. Identification of a simple screening tool for dysphagia in patients with stroke using factor analysis of multiple dysphagia variables. J Rehabil Med. 2005;37:247–51.
Y. Miura et al. 6. Wu MC, Chang YC, Wang TG, Lin LC. Evaluating swallowing dysfunction using a 100-ml water swallowing test. Dysphagia. 2004;19:43–7. 7. Tohara H, Saitoh E, Mays KA, Kuhlemeier K, Palmer JB. Three tests for predicting aspiration without videofluorography. Dysphagia. 2003;18:126–34. 8. Sato M, Tohara H, Iida T, Wada S, Inoue M, Ueda K. Simplified cough test for screening silent aspiration. Arch Phys Med Rehabil. 2012;93:1982–6. 9. Wu CH, Hsiao TY, Chen JC, Chang YC, Lee SY. Evaluation of swallowing safety with fiberoptic endoscope: comparison with videofluoroscopic technique. Laryngoscope. 1997;107:396–401. 10. Smith CH, Logemann JA, Colangelo LA, Rademaker AW, Pauloski BR. Incidence and patient characteristics associated with silent aspiration in the acute care setting. Dysphagia. 1999;14:1–7. 11. Merritt CRB. Physics of ultrasound. In: Rumack CM, Wilson SR, Charboneau JW, Levine D, editors. Diagnostic ultrasound. Philadelphia: Elsevier mosby; 2011. p. 2–33. 12. Huang YL, Hsieh SF, Chang YC, Chen HC, Wang TG. Ultrasonographic evaluation of hyoid-larynx approximation in dysphagic stroke patients. Ultrasound Med Biol. 2009;35:1103–8. 13. Yabunaka K, Konishi H, Nakagami G, Sanada H, Iizaka S, Sanada S, Ohue M. Ultrasonographic evaluation of geniohyoid muscle movement during swallowing: a study on healthy adults of various ages. Radiol Phys Technol. 2012;5:34–9. 14. Yabunaka K, Sanada H, Sanada S, Konishi H, Hashimoto T, Yatake H, Yamamoto K, Katsuda T, Ohue M. Sonographic
15.
16.
17.
18.
19.
20.
21.
assessment of hyoid bone movement during swallowing: a study of normal adults with advancing age. Radiol Phys Technol. 2011;4:73–7. Casas MJ, Seo AH, Kenny DJ. Sonographic examination of the oral phase of swallowing: bolus image enhancement. J Clin Ultrasound. 2002;30:83–7. Geddes DT, Chadwick LM, Kent JC, Garbin CP, Hartmann PE. Ultrasound imaging of infant swallowing during breast-feeding. Dysphagia. 2010;25:183–91. Tamburrini S, Solazzo A, Sagnelli A, Del Vecchio L, Reginelli A, Monsorro M, Grassi R. Amyotrophic lateral sclerosis: sonographic evaluation of dysphagia. Radiol Med. 2010;115:784–93. Hu Q, Zhu SY, Luo F, Gao Y, Yang XY. High-frequency sonographic measurements of true and false vocal cords. J Ultrasound Med. 2010;29:1023–30. Langmore SE, Schatz K, Olson N. Endoscopic and videofluoroscopic evaluations of swallowing and aspiration. Ann Otol Rhinol Laryngol. 1991;100:678–81. Kuhl V, Eicke BM, Dieterich M, Urban PP. Sonographic analysis of laryngeal elevation during swallowing. J Neurol. 2003;250:333–7. Choi KH, Ryu JS, Kim MY, Kang JY, Yoo SD. Kinematic analysis of dysphagia: significant parameters of aspiration related to bolus viscosity. Dysphagia. 2011;26:392–8.
