Sleep, 20(12):1175-1J84 © 1997 American Disorders Association and Sleep Research Society

Detection of Respiratory Events During NPSG: Nasal CannulaJPressure Sensor Versus Thermistor

Robert G. Norman, Muhammed M. Ahmed, Joyce A. Walsleben and David M. Rapoport

Division of Pulmonary and Critical Care Medicine, New York University Medical Center, New York, New York, U.S.A Downloaded from https://academic.oup.com/sleep/article/20/12/1175/2750010 by guest on 01 October 2021

Summary: Recording of respiratory airflow is an integral part of (NPSG). It is conventionally monitored with a thermistor that measures temperature as a surrogate of flow. The subjectivity of interpreting hypopnea from this signal has prompted us to measure nasal airflow directly with a simple pneumotachograph

consisting of a standard nasal cannula connected to a 2-cm H20 pressure transducer. We manually analyzed respi­ ratory events using simultaneous thermistor and nasal cannula in 11 patients with obstructive sleep syndrome (OSAS) and 9 with upper airway resistance syndrome (UARS). Definite events were scored separately for each signal when amplitude was <50% for> 10 seconds. Events were also scored on the nasal cannula signal when the flattened shape of the signal suggested flow limitation. and these were tabulated separately. Definite events in one signal were tabulated by whether the other signal showed a definite event or not. In addition, nasal cannula events were compared to a more liberal thermistor criterion (any change in the signal for 2:2 breaths). Visually, events were more easily recognized on the nasal cannula signal than on the thermistor signal. In OSAS, 1,873 definite thermistor events were detected. Of these, 99.1 % were detected by nasal cannula, and 0.9% were missed. Of 3,541 definite nasal cannula events, 51.9% were detected by definite thermistor criteria; 75.0% were detected by liberal thermistor criteria; 25.0% were missed. In UARS, 123 definite thermistor events were detected. Of these, 89.4% were detected by nasal cannula and 10.6% were missed. Nine hundred and three nasal cannula events were. detected. However, only 17.2% of these were detected by definite thermistor criteria; 38.6% were detected by liberal thermistor criteria; 61.4% were completely undetected by thermistor. When events identified on the nasal cannula hy flow limitation alone were excluded, the thermistor detected 30.1 % of events by definite criteria and 78.6% by liberal criteria, still leaving 21.4% completely undetected by the thermistor. We conclude that the nasal cannula reliably detects respiratory events seen by thermistor. Additional events (including some characterized only by flow limi­ tation) that help define the UARS, were recognized by nasal cannula but often completely missed by thermistor. We propose that respiratory monitoring during NPSG with nasal cannula significantly improves event detection and classification over that with thermistor. Key Words: Respiratory monitoring-Airflow-Flow limitation-UARS­ OSAS-Hypopnea-Apnea.

In the past decade there has been explosive growth couples for this purpose. These devices are based on in the diagnosis of sleep-disordered (SDB)­ measuring relative temperature during expiration and ranging from frank obstructive syndrome inspiration and on the heat content of the air (temper­ (OSAS) to the recently defined upper airway resistance ature and volume) passing over the device. Both syndrome (UARS). In part this has been driven by an thermistors and thermocouples thus only measure flow increased appreciation of the morbidity of excessive indirectly. The use of temperature as a surrogate for daytime sleepiness and the biologically plausible hy­ direct measurement of airflow works well for indicat­ pothesis that significant but mild SOB may lead to ing apnea (absent airflow). However, because the sig­ important cardio- and cerebrovascular consequences nal amplitude bears only a rough correlation to mag­ such as hypertension, myocardial infarction, and nitude of actual airflow [and varies greatly with the stroke. type of thermistor and its position (1)], temperature Polysomnographic diagnosis and assessment of se­ may not be as well related to volume when used to verity of SOB rely on measurement of respiratory air­ assess hypopnea (reduced airflow). Because thermis­ flow. Most laboratories use thermistors and thermo- tors and thermocouples have similar properties we will refer only to thermistors in this paper. Accepted for publication September 1997. Despite the limitations of thermistors it has become Address correspondence and reprint requests to David M. Rapo­ port, M.D., Department of Medicine, New York University Medical conventional to define hypopnea based on reduction in Center, 550 First Avenue, New York, NY 10016, U.S.A. amplitude of the thermistor signal as compared to the 1175 1176 R. G. NORMAN ET AL. local baseline amplitude (2). The exact criteria for hy­ encephalographic (EEG) arousals, AHI < 15 by popnea vary widely, ranging from a 30% to 70% re­ thermistor, and no other diagnosis). Re­ duction to any reduction in amplitude (3,4). In addi­ cordings of central and occipital EEG, electrooculo­ tion, most laboratory use supporting criteria (e.g. pres­ gram (EOG), and submental electromyogram (EMG) ence of 2-4% oxygen de saturation or arousal) to con­ were used to monitor sleep. Leg movements were firm an equivocal event identified on the thermistor monitored with an anterior tibialis EMG. A unipolar channel. While this is useful for preventing excessive electrocardiogram (ECG) was used for cardiac moni­ sensitivity of the thermistor (improving specificity), it toring. Oxygen saturation was monitored with a pulse does not address the issue of identifying events that oximeter. Chest wall and abdominal movement were are entirely undetected by the thermistor. Further com­ monitored with piezoelectric strain gauges. Respira­ plicating the situation is the recent description of tory airflow was simultaneously monitored using a na­ UARS (5). In this syndrome sleep disruption and sal/oral thermistor and a nasal cannula connected to a Downloaded from https://academic.oup.com/sleep/article/20/12/1175/2750010 by guest on 01 October 2021 symptoms similar to OSAS result from repetitive in­ 2-cm H20 pressure transducer (see Appendix for de­ creases in airway resistance (identified by increased tails of cannula setup). Figure 1 shows a picture of the intrathoracic pressure swings) without detectable setup used to collect simultaneous nasal cannula and changes in the thermistor signal. thermistor data. We (6) and others (7) have shown that the identifi­ Respiratory events were manually analyzed from cation of a plateau on the inspiratory flow waveform the thermistor and nasal cannula signals wherever both correlates with an elevated upper airway resistance signals were present and of acceptable quality. Con­ during continuous (CPAP) ti­ tration, and we have proposed that this may be true ventional scoring of and hypopneas was per­ during spontaneous breathing (8). A recent study formed independently from thermistor and nasal can­ shows that there is an excellent correlation between nula signals (with blinding as to the information from the shape of the flow/time curve and the cross-sec­ the other signals). A definite respiratory event (apnea tional area of the airway as well (9). However, a con­ or hypopnea) was scored for one signal when the am­ ventional pneumotachograph signal requires a tight­ plitude of that signal fell below 50% of the local base­ fitting face mask that may be excessively intrusive for line amplitude for more than 10 seconds. The nasal routine sleep monitoring. To circumvent this limita­ cannula signal was also inspected to identify flow lim­ tion, we have been using a simple alternative to the itation events, defined by a flattening of the contour of pneumotachograph that gives a quantitative flow signal successive inspiratory waveforms for at least 10 sec­ without a face mask. This consists of a standard oxy­ onds (but <2 minutes) at a time when the flow am­ gen nasal cannula placed in the nares and attached to plitude did not meet criteria for hypopnea (>50% am­ a sensitive pressure transducer that detects the pressure plitude). These were tabulated separatedly from nasal fluctuations caused by inspiration and expiration. cannula events defined by amplitude criteria alone. Fi­ Monserrat has shown that the signal obtained from nally the thermistor signal was reexamined during such a device is comparable in both shape and ampli­ each nasal cannula event for any visually detectable tude to that of a conventional pneumotachograph (10). change lasting two or more breaths and suggesting hy­ The characteristics of this simple, inexpensive, and popnea. These were classified as liberal thermistor nonobtrusive device make it ideal for quantitative events. monitoring of respiration during sleep. Because neither the thermistor nor the nasal cannula The present study was designed to compare the na­ signal was the a priori gold standard for detection of sal cannula/pressure transducer combination to a con­ events, analysis of their relative sensitivity was per­ ventional thermistor with respect to their ability to de­ formed symmetrically. Events defined by the therm­ tect respiratory events. In particular, we investigated istor were classified as to whether they were simulta­ the ability of each device to (1) detect apnea and hy­ popnea in OSAS, and (2) provide an index of subtle neously detected or missed by the nasal cannula. Con­ in UARS. versely, events defined by the nasal cannula were clas­ sified as to whether they were detected or missed by the thermistor. This analysis was performed separately METHODS for each individual patient. Data were also then pooled A full night of polysomnography was performed on for all OSAS patients and separately for UARS pa­ 11 patients with [apnea-hy­ tients. Percentage agreement and 95% confidence lim­ popnea index (AHI) > 15 by thermistor) and 9 patients its for this percentage were calculated. AHI was cal­ with upper airway resistance syndrome [excessive culated for each signal by dividing the number of daytime sleepiness (EDS), , frequent electro- events by the total sleep time.

Sleep, Vol. 20, No. 12, 1997 EVENT DETECTION BY NASAL CANNULA ON NPSG 1177 Downloaded from https://academic.oup.com/sleep/article/20/12/1175/2750010 by guest on 01 October 2021

FIG. 1. Example of nasal cannula and thermistor setup. The nasal cannula is attached to a differential pressure transducer (prototype r' designed for this purpose by Nellcor Puritan Bennett, Lenexa, KS) with DC output connected to recorder. The inset shows the attachment of a thermistor (EPMS, Midlothian, VA) to standard nasal cannula and placement on the patient.

RESULTS nasal cannula (excluding events defined by flow lim­ itation alone). By conventional amplitude criteria Table 1 shows the sensitivity with which the nasal (50% reduction) the thermistor missed a substantial cannula detected thermistor-defined events. By con­ number of nasal cannula events. In OSAS only 55.7% ventional criteria, 98.1% of the thermistor-defined of events were detected by the thermistor and in events (using the same amplitude criteria) were de­ UARS this fell to 30.1 %. Even when liberal criteria tected by the nasal cannula in OSAS and 82.9% in (any change in appearance) were applied to the therm­ UARS. If events defined on the nasal cannula signal istor signal wherever the nasal cannula detected an by the presence of flow limitation were also included, event, 78.4% of nasal cannula events were detected in ~) the percentage agreement increased to 99.1 % for OSAS and 78.6% in UARS. OSAS and 89.4% for UARS. Table 3 shows the sensitivity with which the therm­ Table 2 shows the sensitivity with which the therm­ istor detected nasal cannula-defined events when these istor detected apneas and hypopneas defined by the included the events defined by flow limitation. In

TABLE 1. Detection of thermistor-defined events by nasal cannula il Apnea/hypopnea defined by thermistor

OSAS (n = 1,873) UARS (n = 123) Detection by nasal cannula % Detected 95% CI % Detected 95% CI Apnea/hypopnea 98.1 97.5-98.7 82.9 76.2-89.6 Apnea/hypopnea or flow limitation 99.1 98.7-99.5 89.4 84.0-94.9

Sleep, Vol. 20, No. 12, 1997 1178 R. G. NORMAN ET AL.

TABLE 2. Detection of nasal cannula-defined events (apnea/hypopnea but not flow-limitation events) by thermistor

Apnea/hypopnea defined by nasal cannula

In OSAS (n = 3,159) In UARS (n = 339) Detection by thennistor % Detected 95% CI % Detected % CI Definite criteria 55.7 54.0-57.4 30.! 25.3-35.3 Liberal criteria 78.4 76.9-79.8 78.6 73.8-82.8

OSAS only 51.9% of events were detected by the ulated as events but suggest sustained elevated upper thermistor and in UARS this fell to 17.2%. Even when airway resistance without arousal, which may be part

the liberal criteria were applied to the thermistor signal of the syndrome of UARS. Downloaded from https://academic.oup.com/sleep/article/20/12/1175/2750010 by guest on 01 October 2021 wherever the nasal cannula detected an event, only Figures 4 through 6 show selected tracings of the 75.0% of nasal cannula events were detected in OSAS dual detection of events by thermistor and nasal can­ and only 38.6% were detected in UARS. Thus in nula. Figure 4 shows an example of an apnea that is UARS over half of the events identified by the nasal well detected by both the thermistor and nasal cannula. cannula (including flow limitation) remained unde­ Figure 5 shows a hypopnea with flow limitation that tected by the thermistor. is detected by the nasal cannula but missed by the Figure 2 shows data from individual patients com­ thermistor. Figure 6 shows an example of repetitive paring the nasal cannula and thermistor as to overall flow limitation events that are detected in the therm­ ability to detect total number of events without regard istor signal only with liberal criteria. to simultaneity. The number of events detected by the nasal cannula was expressed as a percentage of the DISCUSSION number of thermistor events (100% represents an equal number of events) and was plotted against the Improved detection and classification of respiratory number of thermistor events. In each patient the nasal events should allow better discrimination between nor­ cannula was more sensitive than the thermistor mal and disease states, especially at the milder end of (> 100%) and this increased sensitivity was greatest in the spectrum. This has particular importance in char­ mild disease (UARS). For individual patients, up to 23 acterizing patients with UARS who have prominent times as many events were detected by the nasal can­ daytime symptoms but have little detectable respira­ nula as the thermistor. tory abnormality on conventional polysornnography The diagnostic implications of the differences in unless invasive techniques are used. Techniques for event detection are suggested in Fig. 3, which plots improving upper airway patency (surgery and dental the AHI calculated from the nasal cannula signal devices) have been shown to be effective for snoring against the AHI calculated from the thermistor signal. but have had a variable effect on apnea and hypopnea All UARS subjects had thermistor AHIs at or below as detected by thermistors (11,12). Use of a technique 15 events per hour; this could have led to their being sensitive to subtle respiratory events may help resolve classified as normal. By nasal cannula, six of nine of some of the ambiguity associated with these therapies. these subjects had an AHI greater than 15 per hour The data presented in this paper show that the nasal and three had AHIs greater than 10. Because of the cannula has the desired increased sensitivity. It detects greater sensitivity of the cannula, in our laboratory we essentially all of the apneas and hypopneas detected have used a cutoff of 20 nasal cannula events per hour by a thermistor and significantly more events in pa­ for respiration in normals. Even with this higher cut­ tients with UARS. off, six of nine of the patients are at or above this An important issue is the relationship of both the normal limit. The three patients falling below this cut­ thermistor and the nasal cannula signals to physiolog­ off all exhibited long runs of flow limitation and snor­ ical truth. This gold standard could be either a pneu­ ing on the nasal cannula tracing. These were not tab- motachograph signal if one wishes to confirm that the

TABLE 3. Detection of nasal cannula-defined events (apnea/hypopnea and includes flow-limitation events) by thermistor

Events defined by nasal cannula (including flow limitation) In OSAS (n = 3,541) In UARS (n = 903) Detection by thennistor % Detected 95% CI % Detected % CI Definite criteria 51.9 50.2-53.5 17.2 14.9-19.4 Liberal criteria 75.0 73.6-76.5 38.6 35.8-41.5

Sleep, Vol. 20, No. 12, 1997 EVENT DETECTION BY NASAL CANNULA ON NPSG 1179

25000/0 r------~--_, AHltherm=15 ~ 120 v 0 2250%} 1 100

0 i) 1000% 80 750% 0 0 UARS 60 I~ OSAS 500% •• 0 40 0 250% 0 ~) ".1§l0 0 •••••••••••••0 ..(). ••••• ·G •••• .(). ---- I'. •• ~16

20 - -e : - • ------. ------AHI =20 Downloaded from https://academic.oup.com/sleep/article/20/12/1175/2750010 by guest on 01 October 2021 0% Identi Nc 100 200 300 400 500 600 o OL-•__ ~ __~ ____~ __~ __~ __~ # Thermistor Events o 20 40 60 80 100 120 (Definite + Liberal) AHI by THERMISTOR FIG. 2. Number of events detected by the nasal cannula expressed FIG. 3. Comparison of the AHI calculated from nasal cannula sig­ as a percentage of the number of events detected by thermistor in nal and calculated from thermistor signal in each patient. Note that each patient. Open circles are patients with OSAS and closed circles by definition UARS patients have thermistor AHI values at or below are patients with UARS. Note that the nasal cannula always detects the common clinical cutoff for disease of 15 events per hour (vertical more events (> 100%) than the thermistor and that the largest dif­ line). The increased number of events detected by the nasal cannula ference in sensitivity occurs in patients with the least number of allows one to quantitate their disease. Six of nine of the UARS events (UARS). patients fall at or above a cutoff of 20 events per hour (we use this higher cutoff to account for the greater sensitivity of the nasal can­ nula). The remaining three patients had moderate numbers of events. clinical signal being tested represents flow or an They also had long runs of flow limitation and snoring easily de­ tected by the nasal cannula but not classified as events by our cri­ esophageal pressure signal if one wishes to represent teria. effort. In the present study, however, we sought to de­ termine the relationship between our measure and the usual thermistor measure. A separate series of exper­ mation of this pressure (18)] may identify subtle iments would be necessary to address the relationship changes in airway resistance. of the nasal cannula signal to significant events. As Although increased sensitivity of a monitoring de­ used here, significance relates to the presence of an vice may be beneficial, excessive sensitivity to minor

event with consequences (e.g. O 2 desaturation or respiratory irregularities is clearly undesirable. It has arousal), rather than the distinction between apnea and been suggested that detection of flow limitation might hypopnea, and there is an increasing trend to report fall in this category, but we have shown that repetitive clinical data in this fashion. Our data show that essen­ occurrences of flow limitation have physiological con­ tially all the thermistor events, plus an additional num­ sequences. Short repetitive runs of flow limitation are ber of events previously undetected, are picked up by invariably terminated by EEG arousal (19) and may the nasal cannula signal. Identifying the significance thus contribute to daytime sleepiness. Changes in heart of these events is beyond the scope of the present rate similar to the autonomic changes seen with apnea study, but the usual criteria of linkage of event to con­ usually accompany flow limitation events (20). In pa­ sequence could still be applied to such data. tients with UARS these flow limitation events coincide Other techniques have been proposed for detecting with the events described by Guilleminault et a1. (5) respiratory events during sleep and some of these may (which are not reliably detected by thermistors). Fi­ also have greater sensitivity than thermistors. In the nally, although flow limitation occurs in normal sub­ Sleep Heart Health Study (13) the definition of hy­ jects, repetitive flow limitation events occur with a popnea was based on fluctuations in chest and abdom­ greater frequency in symptomatic individuals with inal movement as well as data from the thermistor. UARS (21). The ability to recognize flow limitation Others have suggested that movement alone can be events in addition to apnea and hypopnea provides a sufficient to detect respiratory events using develop­ justification for the use of the nasal cannula instead of ment of paradox or changes in phase angle (14,15). the thermistor during polysomnography. However if Stradling has proposed that the effects of changes in flow limitation is included in the tabulation of respi­ intrathoracic pressure and sympathetic tone related to ratory events [respiratory disturbance index (RDI)], apnea and arousal on the pulse transit time may be some readjustment of the threshold for disease may be clinically useful (16). Finally, direct measurement of necessary. esophageal pressure swings (5,17) [or indirect esti- It should be noted that if a very liberal definition is

Sleep, Vol. 20. No. 12, 1997 1180 R. G. NORMAN ET AL.

03: 13: 00 I I'll ght Si dE' I I ME'an Sa02: j~/A I Wlndow 97 ( 120 sec> EEG

EOG

EMG

LEGS Downloaded from https://academic.oup.com/sleep/article/20/12/1175/2750010 by guest on 01 October 2021 THERMISTOR

ABDOMEN NASAL CANNULA

Sa02 FIG. 4. A 120-second section from the NPSG in one subject showing apneas that are well detected by both the thermistor and the nasal cannula. taken for defining events on a thermistor (or any other such as desaturation and discarding all hypopneas signal), many false-positive events will be detected as without a 2-4% desaturation. In contrast, hypopneas hypopneas. This excessive sensitivity has been avoid­ and flow limitation events on a nasal cannula are ob­ ed by linking events to a physiologic consequence jectively defined and thus may not have as great a need

" 01:11:30 I Left SldE' I I Mean Sa02: N/A I Wlndow 51 ( 120 sec > EEG

EOG

EOG

EMG

LEGS

THERMISTOR

RIB

ABDOMEN NASAL CANNULA

Sa02 FIG. 5. A 120-second section from an NPSG in one subject with a hypopnea that is easily detected on the nasal cannula signal but missed by the thermistor.

Sleep, Vol. 20, No. 12, 1997 EVENT DETECTION BY NASAL CANNULA ON NPSG 1181

02:45:00 1 RighI Sid~ I I M~an Sa02: 93.3 1 Window 83 ( 120 sec)

EEG

EOG

EOG

EMG

LEGS Downloaded from https://academic.oup.com/sleep/article/20/12/1175/2750010 by guest on 01 October 2021

THERMISTOR

RIB

ABDOMEN NASAL CANNULA

Sa02 FIG. 6. A l20-second epoch from an NPSG in one subject with repetitive events of flow limitation and arousal that are detected in the thermistor signal only by liberal criteria. There is no desaturation present during these events. In this case these events are also observable ., on the thoracic and abdominal inductance band signals .

for a confirmatory signal as part of their definition to through the mouth during sleep. Furthermore, during avoid false positives. As pointed out above, the con­ stable sleep stages switching from nose only to nose sequences of these events (which often occur without plus mouth as a breathing route was unusual, occurring significant desaturation, see Fig. 6) can then be as­ only two times per subject per night. Our own experi­ sessed independently. ence is similar to that of Monserrat and coworkers (23) Subjects frequently complain about the discomfort that during use of the nasal cannula, events identifiable of oral/nasal respiratory monitors during polysomnog­ as occur only very rarely during sleep raphy (Sleep Heart Health Study, unpublished data). (1-2% of recording time). With any degree of nasal After using the nasal cannula for 5 years in the clinical airflow, as is seen during partial mouth breathing, we laboratory setting, it is our impression that this device have observed that there is sufficient fluctuation in the is at least as comfortable as most thermistors. The ox­ amplitude and shape of the signal from the nasal can­

'( ygen cannula was designed to be comfortably worn up nula signal to detect respiratory events even if nasal to 24 hours a day by those requiring low flow oxygen flow is only a fraction of total airflow (because that therapy, and it is worn without complaint by many fraction appears to be relatively constant). In particular, patients during sleep. In addition the nasal cannula is intermittent appearance of flow limitation can still be significantly easier to apply correctly than a thermistor observed and used to deduce variation in upper airway and can even be self-applied by a patient during un­ resistance. An additional point is that when intermittent attended studies. mouth breathing occurs, ventilation tends to occur most Whereas conventionally positioned thermistors detect during the portion of time when the subject is awake both nasal and oral airflow, the nasal cannula detects or awakened. Thus the decrease in nose signal occurs only nasal airflow. One could argue that intermittent with arousal rather than sleep, making such events look mouth breathing with either an absence or reduction in very different from apnea or hypopnea associated with nasal airflow can mimic apnea or hypopnea. Gleeson et sleep. al. (22) showed that partial mouth breathing is common Concurrent use of an oral/nasal thermistor with the during sleep in normal subjects, but complete absence cannula may provide redundancy as well complemen­ of nose breathing is very rare. In that study of 14 nor­ tary information. The strength of the cannula lies in r') mal subjects, no subject was observed to breathe solely its ability to separate hypopnea from normal respira-

Sleep. Vol. 20, No. 12, 1997 1182 R. G. NORMAN ET AL.

50 nights. Subjects did not consistently appear to prefer i..~' the comfort of one monitor over the other.

Sleep, Vol. 20, No. 12, 1997 EVENT DETECTION BY NASAL CANNULA ON NPSG 1183

IS. Eaton EJ, Atkins NCM, Stone PA, Woodock AA, Hulme K. motachograph, flow is approximately proportional to Respiratory paradox and arousal from non-REM sleep. Am J Respir Crit Care Med 1997;ISS:AI28. the pressure across the resistor. As pointed out by 16. Pitson DJ, Sandell A, van den Hout R, Stradling JR. Use of Monserrat et al. (10), this proportionality may be fur­ pulse transit time as a measure of inspiratory effort in patients ther improved by taking the square root of the pressure with obstructive sleep apnoea. Eur Respir J 1995;8:1669-74. 17. Milic-Emili J, Mead J, Turner JM, Glauser EM. Improved tech­ drop, but this is not clinically necessary for most ap­ nique for estimating pleural pressure from esophageal balloons. plications. The pressure reading (referenced to atmo­ J Appl Physiol 1964;19:207-11. spheric pressure) is taken from the distal end of a stan­ 18. Tobin MJ, Jenouri GA, Watson H, Sackner MA. Noninvasive measurement of pleural pressure by surface inductive plethys­ dard oxygen nasal cannula with prongs that extend into mography. J Appl Physiol 1983;SS:267-7S. the nose, thus measuring the pressure inside the nose. 19. O'Malley EB, Walsleben JA, Norman RG, Rapoport DM. De­ tection of unappreciated respiratory-related EEG arousals. Am J This pressure drop can be directly recorded and is Respir Crit Care Med 1996;lS3:AS68. taken to represent airflow at the nose. The pressure Downloaded from https://academic.oup.com/sleep/article/20/12/1175/2750010 by guest on 01 October 2021 20. O'Malley EB, Farkas DT, Norman RG, Walsleben JA, Rapoport signal is difficult to calibrate to an absolute value of DM. Frontal EEG arousals are associated with autonomic changes. Am J Respir Crit Care Med 1997;ISS:A77S. flow, as the proportionality constant between pressure 21. Rapoport DM, Norman RO, Krishansamy I, Walsleben lA. Flow and flow is dictated by the resistance of the nasal or­ limitation during sleep in normal subjects. Am J Respir Crit ifice, modified by any blockage caused by the insertion Care Med 1994;149:A493. 22. Gleeson K, Zwillich CW, Braier K, White D. Breathing route of the nasal prongs. However, over short periods (min­ during sleep. Am Rev Respir Dis 1986; 134: IIS-20. utes to hours), the pressure signal accurately tracks 23. Hernandez L, Ballester E, Reolid A, Fornas C, Rodriguez-Roisin airflow (10). In general, during quiet breathing the R, Montserrat 1M. Breathing route detected by conventional de­ vices (thennistor). Am J Respir Crit Care Med 1997;ISS:AI30. pressure swings seen in a typical adult breathing with

a standard nasal oxygen cannula are about 0.5 cm H 20. In young children or individuals with very small nares, APPENDIX: TECHNICAL ASPECTS OF pressure swings will be higher; significant nasal ob­ RECORDING RESPIRATION WITH A NASAL struction can be avoided by using a nasal oxygen can­ CANNULAJPRESSURE TRANSDUCER nula with smaller (neonatal or pediatric) prongs. It is SYSTEM also possible to obtain a dual lumen cannula that al­ The nasal cannula system functions essentially as an lows simultaneous monitoring of airflow and admin­ uncalibrated pneumotachograph. The flow signal is in istration of low flow oxygen. fact measuring a pressure drop across a relatively con­ Detection of the pressure swings from the nasal can­ stant resistance (the inlet of the nose). As in any pneu- nula is accomplished with any stable pressure trans- "

G~ '-'U< is ~ e O~ ....;l t: f;I;,,g en ~------1 ,-., c U !! < :a ~i O~ ....;l 01) f;I;, c ',0 ~~------~

"

o 5 10 15 20 25 30 TIME (sec) FIG_ 8. A flow limitation event recorded from the nasal cannula simultaneously amplified by three different amplifiers. The bottom signal is from a DC amplifier with no filtering. The top two signals are from AC amplifiers with low-frequency filters with time constants of 1.6 in the top trace and S.3 in the middle trace. The short time constant filter (top trace) causes the flow signal to decay to baseline rapidly during any period of relatively constant flow (flow limitation plateau). The longer time constant filter (middle trace) provides reasonably good reproduction of these constant flows.

Sleep, Vol. 20, No. 12, 1997 1184 R. G. NORMAN ET AL. ducer able to detect differential pressures in the range are appropriate to record this signal, but AC amplifiers of ±5 cm H 20 (4 torr). This type of transducer is avail­ intended for EEG and for thermistor signals usually able from several of the companies that make poly­ have too short a time constant or are excessively fil­ graph amplifiers (Grass, Validyne, Nicolet), although tered, altering the appearance of flow-limited breaths. many are intended for monitoring either esophageal An example of signals that might be obtained with pressure or CPAP pressure and thus tend to be tuned various types of amplification is shown in Fig. 8. The to a larger pressure signal. The output of these trans­ unfiltered DC signal is shown on the bottom, and the ducers generally needs to be amplified and condi­ appearance of the identical signal passed through two tioned, as well as sometimes needing a source of pow­ AC amplifiers (T = 1.6 vs 5.3 seconds) is shown er. If the device is not used in conjunction with a poly­ above. As can be seen, normal breaths are well iden­ tified on all tracings, but in the short time constant AC graph intended to interface with it, this may require Downloaded from https://academic.oup.com/sleep/article/20/12/1175/2750010 by guest on 01 October 2021 signal flow-limited breaths tend to be broken into two some special adaptations. In contrast, some devices portions and may lose their characteristic inspiratory sold for these purposes may have some of these fea­ plateau. tures bundled with the transducer but may have sig­ The final requirement of an optimal recording sys­ nificant filtering hard-wired into the circuitry and are tem, if it is based on digital recording of the data, is thus unacceptable for nasal flow monitoring. If it is an adequate sampling frequency and resolution. This desired to detect the shape of the inspiratory airflow should be such that display of the signal allows anal­ waveform (which is proportional to the cannula pres­ ysis of the shape of the inspiratory contour and gen­ sure), it is necessary that the electrical output from the erally requires sampling rates of at least 16-20 Hz and pressure transducer be DC amplified (or with a long an 8-bit resolution. Computerized algorithms to ana­ AC time constant >5 seconds). Filtering should be lyze the shape of the inspiratory curve may require a minimal. Most DC amplifiers on polygraphic systems slightly higher sampling frequency and resolution.

Sleep, Vol. 20, No. 12, 1997