Anesth Analg 2002;95:645-649
© 2002 International Anesthesia Research Society
TECHNOLOGY, COMPUTING, AND SIMULATION
Clinical Application of Acoustic Reflectometry in Predicting the Difficult Airway
E. Andrew Ochroch, MD*, and
David M. Eckmann, PhD MD*
*Department of Anesthesiology, University of Pennsylvania Health System; and Institute of Medicine and Engineering, University of Pennsylvania, Philadelphia, PA
Address correspondence to E. Andrew Ochroch MD, 416C Ravdin Ct., 3400 Spruce St., Philadelphia, PA 19104. Address e-mail to Ochrocha{at}uphs.upenn.edu
 |
Abstract
|
|---|
Acoustic reflectometry, a noninvasive test that produces a length versus cross-sectional area map of the airway, has been used to identify difficult-to-tracheally intubate patients in a small retrospective case-control study. A critical airway volume of 40.2 mL separated those patients whose tracheas were impossible to intubate from those who were easily intubated. To determine if this technology was applicable for prospectively predicting difficult intubation and difficult ventilation in routine clinical practice, we performed a double-blinded, prospective cohort study. Our a priori hypothesis was that small airway volumes in adults (<40.2 mL) would predict absolute inability to intubate. We conclude that by use of acoustic reflectometry, there was no relationship between inability to intubate, poor glottic visualization, and multiple laryngoscopies with airway volume.
IMPLICATIONS: Acoustic reflectometry, a noninvasive test that uses sound to produce a length versus cross-sectional area map of the airway, was not able to predict inability to intubate, poor glottic visualization, and multiple laryngoscopies.
|
Introduction
|
|---|
The identification and management of a difficult airway remains a significant challenge for the anesthesiologist. The ability to continuously oxygenate and ventilate patients is critical to their safety. Reported rates of difficult intubation range from 12.5%, defined as multiple laryngoscopies with poor views (1), to 1% in a general surgical population, defined as a Cormack and Lehane (CL) Class IV view (2) (no laryngeal structures or epiglottis viewed), to 1 in 750 in the obstetric population, defined as absolute failed intubation (3).
Difficult mask ventilation (DMV) is less well categorized than difficult intubation and has only been evaluated in a small number of clinical studies. Reported rates range from 0.07% when DMV is defined as "inability to obtain chest excursion sufficient to maintain a clinically acceptable capnogram waveform despite optimal head and neck positioning and use of muscle paralysis, use of an oral airway, and optimal application of a face mask by anesthesia personnel" (2) to 5% when defined as the "inability of an unassisted anesthesiologist to maintain the measured oxygen saturation as measured by pulse oximetry >92%, or to prevent or reverse signs of inadequate ventilation during positive-pressure mask ventilation under general anesthesia (4)." Other studies using similarly subjective ratings of DMV have shown rates of 0.9% (5) and 1.4% (6).
Advances in planning for and dealing with the patient who unexpectedly cannot be oxygenated, ventilated, or intubated include the development of specific algorithms for action (7) and the introduction of numerous devices to facilitate the establishment of a patent artificial airway. The ability to predict difficult intubation and difficult ventilation has not improved. The physiognomy based prediction systems have specificities in the 93%–97% range and sensitivities in the 80%–90% range (2,5,8). The high false negative and false positive rates make them clinically inapplicable for predicting difficult intubation. To illustrate, if the sensitivity of a test to predict difficult intubation is 90%, its specificity is 95%, and the prevalence of difficult intubation is 1%, the positive predictive value (the rate of a positive test correctly predicting a positive result) is only 15.4% (9). The only system for predicting DMV (4) has similar drawbacks.
Acoustic reflectometry (AR), a noninvasive test that produces a length versus cross-sectional area map of the airway (see Methods), has been used previously to identify difficult-to-intubate patients in a small retrospective case-control study. A critical airway volume of 40.2 mL separated those patients who had a clinical history of being impossible to intubate from a cohort of patients matched for age, sex, and weight who had previously been intubated (10). To determine if this technology can be applied prospectively to predict difficult intubation and difficult ventilation in routine clinical practice, we performed a double-blinded, prospective cohort study. Our a priori hypothesis was that small airway volumes in adults (<40.2 mL) are associated with a diminished ability to view the glottic opening and will therefore predict absolute inability to intubate. Our secondary aim was to determine if small airway volume predicted difficulty of ventilation via a mask.
|
Methods
|
|---|
With IRB approval, we collected data on patients during the preoperative visit (airway volume measurement) and during their anesthesia (mask ventilation and intubation data). Patients scheduled for elective surgery were evaluated in an anesthesia evaluation clinic. Inclusion criteria were any patient who the anesthesia staff determined would have a general anesthetic and endotracheal intubation. Exclusion criteria were anyone not having general anesthesia with an endotracheal tube (ETT) and those patients who would have a fiberoptic intubation.
A single technician performed all AR tests. With the neck held both neutral and then extended on a foam intubation pillow, airway curves were recorded in the standard intubating position during the preoperative evaluation. Post hoc analysis of the curves (as described below) produced airway volumes.
The attending anesthesiologists in the operating room made all clinical airway management decisions. They were blinded to the AR data. Intraoperative data were collected onto a scannable form generated with Teleform® software (version 5.2, Cardiff Software, Vista, CA). Data were collected on both mask ventilation and intubation. Mask ventilation was graded using a modification of the scoring system proposed by Benumof (7). The following scores were assigned: 0 = no mask used, 1 = natural airway, 2 = chin lift used, 3 = oral and/or nasal airway, 4 = two hands on the mask plus oral and/or nasal airway, 5 = two anesthesiologists working together, and 6 = impossible to exchange gas via a mask airway.
Data collected included the success or failure of direct laryngoscopy (yes/no), the number of attempted laryngoscopies (0, 1, 2, 3, 4, 5, or >5 attempts), the experience of the laryngoscopist (certified registered nurse anesthetist, CA1, CA2, attending), the CL classification of the best glottic view achieved (1, 2, 3, 4), whether an alternative device (laryngeal mask airway [LMA], Laryngeal Mask Company, Henley-on-Thames, UK; light wand, or fiberoptic bronchoscopy [FOB]) was used, and was its use planned, unplanned, or urgent/emergent. These data were scanned into a Microsoft Access® (Bellevue, WA) database using the Teleform® software scanner interface.
AR uses sound waves propagated down a wave tube and into the patient’s airway. The reflectometer impulse, 2 ms in duration, is characterized by a flat spectral range from 0–5000 kHz (low-pass filter) and is repeated at the rate of 5 pulses per second (11). Two microphones in the wave tube recorded the reflected sound (12,13). The cross-sectional area of the airway with which the sound was interacting determined the amplitude and phase of the reflected sound. The timing of the sound’s return to the wave tube is a function of the total distance traveled (11–13). Signal analysis is used to generate a plot of cumulative airway volume versus distance along the airway axis (10). For this study, plots were broken down into separate oral, pharyngeal, and total volumes for neutral and extended neck positions using proprietary software from E. Benson Hood Laboratories (Pembroke, MA). The data were exported to a Microsoft Access® database.
The separate databases of airway management data and AR evaluation data were merged and then analyzed using Systat 10 (SPSS Science, Chicago, IL). A two-tailed two-sample t-test was used to judge the relationship between absolute ability (yes/no) to place an ETT by direct laryngoscopy and means of oral airway volumes (neutral and extended), pharyngeal volumes (neutral and extended), total airway volumes (neutral and extended), and absolute and percentage change in airway volumes between neutral and extended. Although normality of airway volumes could not be assumed, the large n, applicability of the central limit theorem, and robustness of the t statistic made this choice appropriate (14). Significance was assumed if P < 0.05 and the 95% confidence interval (CI) did not overlap the 95% CI of the null hypothesis. Because the normality of distribution of airway volumes could not be assumed, the Spearman correlation (r) was used to judge the correlation between number of laryngoscopies, CL scale, and means of oral airway volumes (neutral and extended), pharyngeal volumes (neutral and extended), total airway volumes (neutral and extended), and absolute and percentage change in airway volumes between neutral and extended. Tests of significance were applied if the r > 0.7 or r < -0.7 (i.e., there was a strong correlation) and was assumed if P < 0.05 and the 95% CI did not overlap the 95% CI of the null hypothesis.
To investigate the validity of the preexisting rule (10), patients were separated into two groups having total airway volumes less than 40.2 mL or total airway volumes more than or equal to 40.2 mL. A 
2 analysis was used to judge the relationship between these two groups and the absolute inability to intubate patients. Group status was correlated with surrogate markers of difficulty of intubation using the Spearman correlation as described above.
Reliability of acoustic data acquisition was investigated using 53 patients in whom repeated measurements were made on separate clinic visits. Spearman correlation was performed to analyze test-retest reliability using the same methodology detailed above.
|
Results
|
|---|
Complete data sets were recorded for 1748 patients (Table 1). Of these, 141 patients were not successfully intubated. The mean number of intubation attempts was 2.00 (SD, 1.6). The remaining 1607 patients were successfully intubated. The mean number of intubation attempts was 1.15 (SD, 0.48). A total of 49 patients had three or more laryngoscopies. All 49 of these patients were easily ventilated via a mask, and 41 of these patients were eventually intubated successfully.
The ability for airway volumes to predict difficulty intubating was analyzed according to published definitions of difficulty of intubation. The absolute inability to place an ETT did not depend on mean total airway volume neutral (P = 0.386; 95% CI, -6.392–15.947), mean total airway volume extended (P = 0.786; 95% CI, -8.955–11.701), mean pharyngeal volume neutral (P = 0.439; 95% CI, -6.692–14.92), mean pharyngeal volume extended (P = 0.671; 95% CI, -7.765–11.843), the difference between the mean total airway volume extended and the mean total airway volume neutral (P = 0.099; 95% CI, -7.502–0.692), and the difference between the mean pharyngeal volume extended and the mean pharyngeal volume neutral (P = 0.213; 95% CI, -5.418–1.269) (Table 1).
Defining difficult airway as requiring three or more laryngoscopies, no correlation with airway volumes was found (Table 2). The number of laryngoscopies did not correlate with total airway volume neutral (r = 0.009), total airway volume extended (r = -0.074), pharyngeal volume neutral (r = -0.063), pharyngeal volume extended (r = -0.084), the difference between the mean total airway volume extended and the mean total airway volume neutral (r = -0.052), and the difference between the mean pharyngeal volume extended and the mean pharyngeal volume neutral (r = 0.045). Overall, the number of laryngoscopies did not correlate with airway volumes.
View this table:
[in this window]
[in a new window]
|
Table 2. Table 2. Number of Patients Whose Tracheas were Intubated or Not Intubated Correlated by Indicators of Difficult Intubation
|
|
Defining difficult airway as a CL grade of 3 or 4, similar results were found. The CL grade did not correlate with total airway volume neutral (r = -0.009), total airway volume extended (r = -0.048), pharyngeal volume neutral (r = -0.008), pharyngeal volume extended (r = -0.034), the difference between the mean total airway volume extended and the mean total airway volume neutral (r = -0.074), and the difference between the mean pharyngeal volume extended and the mean pharyngeal volume neutral (r = -0.045). Overall, airway volumes did not correlate with the view achieved of the glottic aperture.
Applying the preexisting rule of 40.2 mL of airway volume being a defining line between intubatable and unintubatable patients (10), the results did not achieve significance in distinguishing between patients who could and could not be intubated (P = 0.788). There were 152 of 199 patients with total airway volumes <42 mL who were successfully intubated. Applying the rule did not achieve significance in distinguishing between patients who required three or more laryngoscopies and those who required only one or two (P = 0.427) and did not achieve significance in distinguishing between patients in whom a CL scale of 3 or 4 vs 1 or 2 was achieved (P = 0.553).
Using the absolute ability to intubate as a true positive result and an airway volume of <40.2 mL as a positive test, we calculate the sensitivity and specificity of AR to be 62% and 77%, respectively. On secondary analysis using total airway volumes of 20, 30, 40, 50, 60, and 70 mL as cutoff points to generate the receiver operating characteristic curve, the calculated area under the receiver operating characteristic curve is 0.688. Area values this small indicate that the test has poor distinguishing characteristics (15).
One patient could not be ventilated via a mask (ventilation score = 6) (Table 3). The three patients with clinically significant indicators of two anesthesiologists working together to ventilate via a mask the patient (ventilation score = 5) or inability to ventilate via a mask (ventilation score = 6) were all easily intubated (Table 3). The absolute inability to ventilate via a mask (ventilation score = 6) was highly correlated with reduced pharyngeal volume, but this is a single result and an extremely unstable statistic (one of 1748 total patients). Combining the three patients with ventilation scores of 5 and 6 and comparing their mean airway volumes with all other scores failed to produce a significant result, and the small number of events makes this a highly unstable statistic.
The reliability of acoustic data acquisition was investigated by using the 53 patients in who repeated measurements (repeat visit to the anesthesia evaluation clinic) were acquired. Test-retest reliability was very good (r = 0.781; P < 0.02).
|
Discussion
|
|---|
In this study, 141 of 1748 patients (8.3%) could not be intubated with conventional direct laryngoscopy. This rate is much more frequent than has been reported in most other studies (1–3). We did not corroborate the findings of a previous case-matched study (10) that difficult intubation could be expected if AR airway volume was <40.2 mL. This is perhaps because of the change in airway management in the last decade and is discussed below (see Limitations).
Complete inability to ventilate via a mask was predicted accurately by AR. However, this is a very rare event (<0.06%). Because we only recorded one such event, the clinical importance of this statistically significant finding remains in doubt. With further study, AR may be useful in preventing the can’t ventilate/can’t intubate situation by prompting use of nonconventional ventilation or intubation from the very beginning of a general anesthetic.
Our study showed that airway volume measurement by AR is a reliable measure in clinical practice and is supported by previous work (16). AR has previously been shown to match airway volume measurements from computed tomography (CT) scans (17). Unfortunately, we did not find any clinical use of AR to predict the absolute inability to intubate a patient.
Limitations
Initial reading may suggest that a significant limitation of this study is the small study size versus the published incidences of difficult airways in the nongravid population. Our study was stopped because an interim analysis was performed to determine if the airway information was so predictive that it should be made available to the attending anesthesiologist. Surprisingly, the trend of the airway data was for the unintubatable patients to have slightly larger airway volumes. A power analysis reveals that over 22,000 patients would be required to find a difference (if there was one) with an 
of 0.05 and a power of 80% (ß = 0.2), assuming our mean and SD of total airway volumes hold true in a larger patient sample. Consequently, the study was stopped because we would never be able to prove our a priori hypothesis to be true.
The main issue with evaluating this study lies in the change in airway management in the last decade. The mean number of intubation attempts in our unintubatable group was two (SD, 1.6). The clinical practice of performing repeated laryngoscopies has been replaced with the rapid use of alternative means of securing the airway. In our study, 55% of patients had an LMA placed after failure of laryngoscopy, and half of these patients (28%) then had their ETT placed by FOB via the LMA. Forty-two percent of patients who were unintubatable had their ETT placed by FOB alone, and 3% were intubated using other means (e.g., light wand, Bullard scope, or retrograde wire). One patient had an emergency tracheostomy to establish a surgical airway. Consequently, our rates of could not intubate are not comparable to studies like those of el-Ganzouri et al. (2) because the definition of could not intubate has so drastically changed.
In analyzing our data, our main outcome was to determine if total airway volume correlated with absolute inability to intubate. All of our other data are presented as secondary analysis and are only of interest for future hypothesis generation. Consequently, no true relationship between surrogate markers of difficult airway (CL scale and number of laryngoscopies) and airway volumes has been determined.
|
Acknowledgments
|
|---|
Supported, in part, by National Institutes of Health grant number R44 HL55139.
|
References
|
|---|
- Mallampati SR, Gatt SP, Gugino LD, et al. A clinical sign to predict difficult tracheal intubation: a prospective study. Can Anaesth Soc J 1985; 32: 429–34.[ISI][Medline]
- el-Ganzouri AR, McCarthy RJ, Tuman KJ, et al. Preoperative airway assessment: predictive value of a multivariate risk index. Anesth Analg 1996; 82: 1197–204.[Abstract]
- Rocke DA, Murray WB, Rout CC, Gouws E. Relative risk analysis of factors associated with difficult intubation in obstetric anesthesia. Anesthesiology 1992; 77: 67–77.[ISI][Medline]
- Langeron O, Masso E, Huraux C, et al. Prediction of difficult mask ventilation. Anesthesiology 2000; 92: 1229–36.[ISI][Medline]
- Rose DK, Cohen MM. The airway: problems and predictions in 18,500 patients. Can J Anaesth 1994; 41: 372–83.[Abstract/Free Full Text]
- Asai T, Koga K, Vaughan RS. Respiratory complications associated with tracheal intubation and extubation. Br J Anaesth 1998; 80: 767–75.[Abstract/Free Full Text]
- Benumof JL. Management of the difficult adult airway: with special emphasis on awake tracheal intubation. Anesthesiology 1991; 75: 1087–110.[ISI][Medline]
- Willson ME, Speigelhalter D, Robertson J, Lesser P. Predicting difficult intubation. Br J Anaesth 1988; 61: 211–6.[Abstract/Free Full Text]
- Hennekens CH, Buring JE. Epidemiology in medicine. In: Mayrent SL, ed. Boston, MA/Toronto, Canada: Little, Brown and Co, 1987.
- Eckmann D, Glassenberg R, Gavriely N. A pilot study: acoustic reflectometry and the difficult airway. Anesth Analg 1996; 83: 1084–9.[Abstract]
- Jackson AC, Butler JP, Miller EJ, et al. Airway geometry by analysis of acoustic pulse response measurements. J Appl Physiol 1977; 43: 523–36.[Abstract/Free Full Text]
- Fredberg JJ, Wohl MEB, Glass GM, Dorkin HL. Airway area by acoustic reflections measured at the mouth. J Appl Physiol 1980; 48: 749–58.[Abstract/Free Full Text]
- Louis B, Glass G, Kresen B, Fredberg JJ. Airway area by acoustic reflection: the two-microphone method. J Biomech Eng Trans Am Soc Mechanical Engineers 1993; 115: 278–85.
- Rosner B. Fundamentals of biostatistics. 5th ed. Belmont, MA: Duxbury Press, 1999.
- Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143: 29–36.[Abstract/Free Full Text]
- Brooks LJ, Castile RG, Glass GM, et al. Reproducibility and accuracy of airway area by acoustic reflection. J Appl Physiol 1984; 57: 777–87.[Abstract/Free Full Text]
- D’Urzo AD, Rubinstein I, Lawson VG, et al. Comparison of glottic areas measured by acoustic reflections vs computerized tomography. J Appl Physiol 1988; 64: 367–70.[Abstract/Free Full Text]
Accepted for publication May 14, 2002.
Top
Abstract
Introduction
