Introduction. The laryngeal chemoreflex may explain why prone sleeping increases the risk of sudden infant death syndrome (SIDS). Swallowing and arousal are crucial to prevent laryngeal chemoreflex stimulation. Our aim was to examine these reflexes and breathing responses in healthy neonates after pharyngeal infusion of water in the supine versus the prone position, controlling for sleep state.
Methods. A total of 10 term infants were recruited after parental consent and ethics approval. Polygraphic recordings included sleep state (active and quiet sleep by electroencephalogram, eye movements, breathing, and behavior), cardiorespiratory measurements (nasal airflow, chest wall movements, heart rate, and oxygen saturation), swallowing, and esophageal activity (solid state pressure catheter). Initial sleeping position was assigned randomly. Measurements were made for 1 minute before and after 0.4 mL of water was instilled into the oropharynx. To detect a 30% decrease in swallowing, power analysis indicated that ≥10 babies were required. Analysis, blinded to position, was made using nonparametric statistics.
Results. Of the 164 infusions, the most commonly evoked airway protective responses to pharyngeal infusion were swallowing (95%) and arousal (54%). After infusion in active sleep, there was a significant reduction in swallowing and breathing when the prone position was compared with the supine position (prone: 21.3 [1.0] swallows/min and −9.6 [2.1] breaths/min; and supine: 32 (2.2) and −2.9 (1.5), respectively). However, there was no difference in the occurrence of arousal after water infusion.
Conclusion. These data suggest that airway protection is compromised in the prone sleeping position during active sleep, even in healthy infants exposed to minute pharyngeal fluid volumes of 0.4 mL. This is because swallowing rate is reduced significantly, and there is no compensatory increase in arousal. The reduction in airway protective reflexes when in the prone position and in active sleep may be the mechanism for the increased risk of SIDS in the prone position.
Among the identified risk factors for sudden infant death syndrome (SIDS), the prone sleeping position has been one of the most accessible to behavioral change. Public awareness campaigns promoting the supine sleeping position as opposed to the prone sleeping position have been associated with a reduction in SIDS rates of between 50% and 70% in many countries.1
The mechanism by which the prone sleeping position should confer an increased risk of death to a young infant remains unclear. It is important to understand the underlying mechanism to prevent additional deaths and the potential recurrence of SIDS. Any mechanism proposed to explain sudden infant death must be consistent with the epidemiology of SIDS, most specifically, a silent death during presumed sleep associated with the prone sleeping position and largely confined to infants <6 months of age.
Of the mechanisms that have been considered, one of the most striking and potentially lethal is the activation of the laryngeal chemoreflex (LCR). This reflex, which has been described extensively in tracheostomized animals (with or without anesthesia), is activated by direct fluid stimulation of the laryngeal mucosa and leads to a complex series of responses in the young animal, including apnea, bradycardia, swallowing, startle, hypertension, and regional redistribution of blood flow.2 The reflex is age-related and confined to the young infant.3
The LCR has been examined in animals that are surgically intact and asleep,4 surgically intact and anesthetized,5 tracheostomized and anesthetized,2,,36–16 and tracheostomized and sleeping.17–19 The differing methodologies used in these studies result in very different responses. An analysis of these studies demonstrates that LCR-induced apnea is only life-threatening when the subject is anesthetized and when the stimulus is applied directly to the larynx (via a tracheostomy)2,,36–16 or has stimulated the larynx via the pharynx.5 This suggests that after pharyngeal fluid stimulation LCR-induced apnea could be fatal if the mechanisms that protect the airway from fluid entry, such as swallowing, arousal, and expiratory reflexes, are depressed in the infant. Our study in healthy human infants20 supports this hypothesis, because pharyngeal infusion, when supine, did not evoke the LCR complex of responses as described above. The responses that did occur were swallowing, arousal, and expiratory reflexes, whereas breathing was always maintained. Indeed, these responses are most likely mediated by receptors in the pharynx as opposed to laryngeal receptors, because there was no alteration in breathing. Nonetheless, young infants are potentially at-risk for LCR-induced acute life-threatening or fatal apneic events, because responses consistent with the LCR have been reported in the preterm and hospitalized human infant.21–24
Our hypothesis was that the airway protective mechanisms (swallowing and arousal) that protect the larynx from fluid stimulation (and thus LCR-induced decrease in breathing rate and apnea) would be decreased in the prone compared with the supine position. We tested this hypothesis by examining physiologic responses after pharyngeal water infusion during sleep (simulating pharyngeal reflux or nasopharyngeal secretions) in both the supine and prone positions.
Subjects and Selection
A total of 10 healthy, term infants (37–42 weeks' gestation) were selected for study at Royal Prince Alfred Hospital (Sydney, Australia) between August 1997 and February 1998. The Ethics Committee of Royal Prince Alfred Hospital approved the experimental protocol. Informed and written consent were obtained from the parents of each infant who participated in the study.
Selection criteria for entry into the study required that the infants were born at a gestational age of 37 to 42 weeks (routinely determined by maternal dates in conjunction with ultrasound examination before 20 weeks' gestation), breast fed, not treated for jaundice, appropriate weight and length for gestational age according to Australian growth charts,25 and 3 to 5 days of age at the time of the study. Entry criteria also stipulated that the infants were born by normal vaginal delivery to nonsmoking mothers (confirmed by urinary cotinine levels), who had no complications of pregnancy and who had not received analgesic drugs via intramuscular or epidural routes during labor.
For each recording session, probes were applied to the infant, and the infants were placed to sleep in the prone or supine position with their heads in lateral recumbency. All the studies were conducted in the same quiet, warmed room after the infant had been breastfed.
The infusion studies were recorded on a 12-channel pen recorder (Model 78D; Grass Instruments Co, Quincy, MA) and a Macintosh computer data acquisition system (Model MP 100 Data Acquisition System; Biopac Systems Inc, Goleta, CA; using AcqKnowledge 881 Version 3.1.2 software). Traces were rejected if the results were obscured by movement artifact.
Polygraphic Recording of Sleep State
Sleep state was determined by the examination of multichannel pen and computer on-line recordings and by constant documentation of behavior. Sleep state was recorded using an electroencephalogram (EEG), electroocculogram, and measurement of gross body movement (piezoelectric crystal placed over larynx). Sleep was staged according to the criteria of Anders et al,26 in the 70-second epoch immediately before each infusion of fluid into the pharynx. This included a 60-second control period followed by a 10-second epoch immediately before infusion in which the infant was lying quietly without any apnea or swallows. Sleep was classified as active sleep, quiet sleep, or indeterminate sleep. Behavioral observations such as coughs and sneezes also were noted on the paper traces and entered directly into the computer as a comment.
Breathing measurements consisted of nasal airflow that was recorded using a Premie Airflow Sensor (Edentec Corp, Eden Prairie, MN) placed under the external nares of the infant and of chest wall movements from a pneumogram wrapped around the baby's chest and connected to a plethysmograph (Model 270; Parks Electronics Laboratory, Beaverton, OR).
Respiratory rate was calculated by counting the number of completed breaths in the 60 seconds immediately before infusion. This was compared with the number of completed breaths in the 10 seconds after each infusion of fluid into the pharynx and was expressed as a change in breaths per minute.
Oxygen Saturation and Heart Rate
Oxygen saturation and heart rate were recorded using the Nellcor 200 pulse oximeter (Nellcor Incorporated, Pleasanton, CA) sampling at 3 Hz and averaging every 3 seconds. The sensor probe (Nellcor Oxisensor II N-25 Neonatal/Adult O2 transducer) was wrapped around the foot of the infant and covered with an opaque dressing to decrease interference from ambient light and to maintain good contact with the skin. Beat-to-beat heart rate was obtained from amplification of the analog output signal of the oximeter via a tachograph. Artifact was determined by visualization of the trace and determined as areas of heart rate trace showing rapid variability in pulse readings and/or areas of oxygen saturation showing lost signal.27
Airway Protective Responses
Airway protective responses, namely swallowing, arousal, and expiratory reflexes (cough or sneeze) were determined from the pen and computer on-line recordings in addition to the behavioral observations. Swallowing and esophageal motility were monitored using a custom-made Mikro-tip catheter containing four solid state microtransducers (Model SSD-678; Millar Instruments Inc, Houston, TX) in conjunction with a Millar catheter pressure transducer system (4 X Model TC-510 Transducer Control Unit). The distal pressure transducer was set in the tip of the catheter with the subsequent transducers positioned at distances of 9.0, 12.0, and 15.5 cm from the distal transducer. The catheter (size 6 French gauge) was inserted nasogastrically; thus, the proximal pressure transducer was positioned in the pharynx. Correct positioning of the pharyngeal transducer was achieved by the identification of sharp onset pressure rises characteristic of pharyngeal swallows21,,24on a 12-channel pen recorder (Model 78D; Grass Instruments Co). Once the proximal transducer was positioned in the pharynx as described, the subsequent transducers were situated in the upper esophagus, lower esophagus, and stomach. This arrangement of pressure transducers facilitated the clear monitoring of swallowing and other pharyngeal and esophageal body movements, as well as showing changes in intragastric pressure. The occurrence, rate, and latency of swallowing were analyzed.
Arousal was defined by electrophysiologic and behavioral criteria, in which the EEG trace showed a change to low-voltage, fast wave, and body movement occurred (observed visually and seen as a movement artifact on at least two traces for ≥1 second of the chart recording) with or without the eyes opening. Behavioral arousal was selected as the endpoint in this study, because it can be defined clearly, and because it is very difficult to determine a universally accepted definition of polygraphic arousal. The occurrence and duration of arousal were analyzed. Expiratory reflexes were identified aurally and confirmed by a characteristic esophageal pressure trace.24
During active and quiet sleep, fluid boluses of 0.4 mL of sterile water were administered into the pharynx via a French 5-gauge feeding tube. The tube was positioned through the nose and taped to the infant's face to ensure the fluid boluses were administered at skin temperature.21,,24 The infusion catheter was positioned so that the opening was in the oropharynx, 1 cm proximal to the pharyngeal pressure transducer on the manometric catheter. At least 3 minutes elapsed between each infusion. The initial sleeping position was assigned randomly, and after 8 to 10 infusions were made in the initial position (prone or supine), the infants were turned to the other sleeping position, and a similar number of infusions were made. The experiments were halted when the protocol was complete or when the infant was aroused and would not return to sleep. Each infant was recorded in both the prone and supine positions.
Control periods were defined as 1-minute epochs beginning 70 seconds before each infusion of the fluid. To meet the criteria for receiving a stimulus, the remaining 10 seconds immediately preceding the infusion were a period in which the infant was lying quietly without any apnea or swallows with the infant in either active or quiet sleep. The outcome variables were determined in the control periods and in the 10-second period immediately after infusion4,,20,24and expressed as either an occurrence (categorical data) or a rate per minute (continuous data).
The analysis was blinded to a sleeping position. Results for the categorical data are expressed as a percentage of the total number of infusions performed. The results for the continuous data are expressed as mean and SEM, because these parameters were the most appropriate for describing these data and for comparison with our previously published data.4,,20
The responses evoked after each infusion, namely swallowing, arousal, expiratory reflexes, and change in respiratory rate, were compared with the control periods. The effect of sleep state (active or quiet) and position (prone or supine) on each response was examined.
The χ2 test was used to examine the effect of sleep state and position on categorical data and on the occurrence of swallowing, arousal, and expiratory reflexes. The Mann-Whitney U test was used to examine the effect of sleep state and position on the continuous variables, ie, rate and latency of swallowing, latency to and duration of arousal, and change in respiratory rate.
Power analysis indicated that significant differences would be detected with a sample size of ≥10 babies (α set at 0.05; and β set at 0.2). Our studies in the piglet suggested that a 30% decrease in swallowing after pharyngeal infusion (from that elicited during normal sleep), was clinically significant.28 This was calculated using data we recorded, after pharyngeal fluid stimulation in the supine-sleeping human infant.20
Multiple comparisons were chosen as the method of analysis, because the response to each infusion was independent of every other infusion, ie, there was no habituation to the test fluids within each infant. All statistical analyses were completed using the Statistical Package for Social Sciences (SPSS/PC+ Version 4.0) program.29Probability values of P ≤ .01 were considered significant.
Infusion studies were undertaken and completed in 10 healthy term infants. The gestational and postnatal ages of the infants (mean ± SD) were 39.7 ± 1.1 weeks and 3.7 ± 0.9 days, respectively. During the 10 recording sessions, a total of 164 infusions were made with 96 infusions made during active sleep (48 in the supine and 48 in the prone position). In quiet sleep, a total of 68 infusions were made (28 in the supine and 40 in the prone position).
After each infusion, the onset of these responses usually occurred within 10 seconds (Table 1B). An example of the physiologic on-line recordings obtained during the infusion protocol is illustrated in Fig 1.
The control (spontaneously occurring in sleep) occurrence and rate of swallowing (% of the total infusions performed and swallows/min, respectively) were affected by the sleep state in the supine position only. The control occurrence and the rate of swallowing were significantly greater in active compared with quiet sleep in the supine position (Table 1A; P ≤ .01). After pharyngeal water infusion, both the occurrence and rate of swallowing significantly increased from control values for both the supine and prone positions in both sleep states (Table 1A; P ≤ .01). Although pharyngeal infusion significantly increased swallow frequency (swallows/min), after infusion the frequency of swallowing evoked in the prone position was significantly lower than the supine position in active sleep (Table 1A and Fig 2;P ≤ .01).
The control (spontaneous) occurrence of arousal was significantly lower in the prone compared with the supine position in active sleep (Table 1A; P ≤ .01). After pharyngeal infusion, there was a significant increase in the occurrence of arousal in the prone position in both active and quiet sleep. In the supine position, there was a significant increase in the occurrence of arousal in quiet but not active sleep (Table 1A; P ≤ .01). The mean (SEM) duration of behavioral arousal after infusion in active sleep was 33 (4) seconds for the supine and 30 (4) seconds for the prone position. After infusion in quiet sleep, behavioral arousal lasted for a mean of 21 (5) seconds and 34 (4) seconds for the supine and prone positions, respectively. The duration of behavioral arousal was not significantly different between positions or sleep states.
There was a small decrease in respiratory rate during quiet sleep in both the supine and prone positions after pharyngeal water infusion. In active sleep in the supine position in response to pharyngeal infusion, there was also a small decrease in respiratory rate. However, there was a marked decrease in respiratory rate after pharyngeal infusion in the prone sleeping position in active sleep, which was significantly greater than was recorded in the supine position in active sleep (Fig 2; P ≤ .01).
Oxygen Saturation and Heart Rate
Artifact on the heart rate and/or the oxygen saturation trace was identified after 88% of infusions.
The relevance of this study is the demonstration of a possible mechanism that explains how the prone sleeping position increases the risk of the SIDS. Physiologic responses to pharyngeal water infusion are clearly different in the prone position compared with the supine position. When prone and in active sleep, swallowing frequency was significantly lower, and the decrease in respiratory rate was significantly greater with no compensatory increase in arousal, compared with the supine position. These data demonstrate a significant reduction in airway protection in the young infant while sleeping in the prone position compared with the supine position. However, heart rate and oxygen saturation results could not be evaluated adequately, because artifact occurred frequently, during even the slightest movement.27
Although the swallowing rate was rapid in the supine position (≤1 Hz) in response to the small infusions, respiratory rate remained largely unaffected. This demonstrates that, when supine, the infant can coordinate rapid swallowing while maintaining breathing. This maintenance of breathing during rapid swallow runs has also been reported during feeding of both breastfed and bottle-fed infants of similar age.30
These data suggest that different sets of receptors may be stimulated after pharyngeal fluid stimulation in the prone and supine positions; we speculate that one set is pharyngeal and that the other is laryngeal. Although the pattern of responses after infusion was similar in that swallowing was the most common response, followed by arousal and occasionally expiratory reflexes, in the prone position during active sleep, swallowing frequency was 30% lower, and breathing rate decreased significantly after infusion when compared with the supine position. This prone type of response is similar to the LCR complex described in the lamb during wakefulness, after retrograde laryngeal fluid stimulation via a tracheostomy,2 whereas the supine type of response, which involved swallowing and arousal, did not include the concurrent reduction in breathing. This suggests that when prone, the response resembled the LCR, as described in the tracheostomized and sleeping animal,18,,19 and was stimulated after pharyngeal infusion in our study.
This marked response in the prone position occurred after infusion of a small volume (0.4 mL), and, furthermore, occurred despite simultaneous occurrence of arousal. These data suggest that when the infant is sleeping prone, arousal may not be immediately effective in restoring the drive to breathe when there is accompanying fluid stimulation of the upper airway. Moreover, the fact that the healthy infant was vulnerable to such small volume insults (0.4 mL) suggests that the response would be more exaggerated with larger volume insults that may occur, for example, with high gastroesophageal reflux or nasopharyngeal secretions from an upper respiratory tract infection.
The reduction in swallowing and breathing when prone is of particular relevance. This is because previous studies have demonstrated that the occurrence of gastroesophageal reflux in sleep is confined largely to active sleep in both healthy infants31 and in infants presenting with acute life-threatening episodes.32,,33Furthermore, in infants with acute life-threatening episodes, when reflux did occur, it frequently reached the level of the pharynx,32 an event that is very uncommon in other infants.34
The reduction in swallowing and breathing after infusion in the prone compared with the supine position occurred, although the infants were healthy and the stimulus minimal (so as not to cause potential harm). In piglets, when swallowing and arousal were both depressed, the infusion of small volumes of pharyngeal fluid in sleep-produced life-threatening apnea and death.5 Therefore, impaired or depressed swallowing or arousal by extrinsic causes such as drugs or intrinsic causes such as infection or abnormal brainstem neurology35 are likely to further inhibit breathing when the infant is prone. For example, phenothiazine drugs, which are readily available and sold to parents in many over-the-counter preparations,36 have been implicated in SIDS37,,38 and have been shown to affect both swallowing and breathing. In piglets, phenothiazines have been observed to reduce the effective clearance of pharyngeal fluid by swallowing.28 In human infants, phenothiazines resulted in reduced spontaneous arousal and increased apneas.38Alternatively, the potentiation of LCR apnea by respiratory syncytial virus,39,,40 the demonstration of abnormal hypoglossal nuclei in SIDS compared with control infants41 with the potential for impaired swallowing, represents the one or more additional factors needed to convert a decrease in breathing rate described in the prone sleeping infants in this report to life-threatening apnea.
Although SIDS is rare in the first month of life (when our observations were made), the present study may suggest a mechanism indicating why infants die more frequently in the prone position. Additional evidence may suggest why SIDS is rare in the first postnatal weeks. First, epiglottal taste buds on the laryngeal surface (that may mediate the reflex apneic response during chemical stimulation of the larynx) increase sixfold during the early postnatal period (from birth to 18 to 25 days of age) in the kitten.42 Second, during the first postnatal months, the larynx is descending from its high, fetal position to the permanent lower adult position. During this time, the anatomic relationship between the larynx and the epiglottis is unstable and may increase the risk of exposure of epiglottal taste buds to fluid stimulation.43
There are three major variables operating in the causal pathway, namely, the prone position, the active sleep state, and the activation of an upper airway reflex, most likely the LCR. All three variables are age-related and are most common in the young infant. This is consistent with SIDS epidemiology, which is largely confined to young infants <6 months of age, with a peak at 2 to 4 months. Possible explanations for the attenuation of SIDS at this age range is that between 4 and 6 months of age, infants who have been placed prone to sleep are beginning to roll spontaneously and variably assume the supine position for sleep.44 Furthermore, active sleep is no longer the dominant sleep state, occupying 35% of sleep time by 6 months of age in healthy infants.45 Finally, animal studies indicate a rapid attenuation of the LCR with advancing maturation.3
The other notable epidemiologic risk factors, namely exposure to smoking, overheating, and a recent viral illness, may also operate through prolongation of LCR-induced apnea. In Norway, recent population data indicate that when the prone position decreased from 64% to 1.4%, the SIDS rate decreased from 3.5 to 0.2 per 1000 live births.46 In Hong Kong, where back sleeping is the norm, SIDS is virtually nonexistent.47 These observations would suggest that the prone sleeping position is the single major risk factor. Therefore, other risk factors are likely to operate via the mechanism that produces death in the prone position. Therefore, the results of our study suggest that smoking, overheating, and recent infection increase the risk of LCR stimulation by depressing or impairing swallowing or arousal or by prolonging apnea once stimulation of the LCR has occurred.
Thus, one component of smoking, nicotine, has been shown to depress hypoxic chemoreceptor sensitivity.48 Therefore, once LCR apnea is induced, the normal hypoxic chemoreceptor stimulation to ventilation is depressed further by nicotine, and apnea is prolonged. Similarly, both evidence of recent viral infection39,,40and elevated temperature (whether attributable to extrinsic causes, such as overwrapping, or intrinsic causes) can prolong LCR-induced apnea.49
In addition to the physiologic differences between the supine and prone positions observed in this study, others have shown that when supine, the piriform fossae are available for the pooling of fluid introduced via the nasopharynx.24 In the prone position, the piriform fossae are unavailable to potential fluid entry into the pharynx. In addition, when prone, the esophagus is superior to the laryngeal opening, and therefore, fluid from the pharynx flows directly into the interarytenoid notch and laryngeal opening as a result of gravitational force. This region is known to be richly supplied with laryngeal chemoreceptors.16
In summary, the LCR is a well described, potent cause of apnea and bradycardia, age-limited to the young infant, that can lead to profound apnea and death. For both physiologic and anatomic reasons, the prone position makes activation of this reflex more likely. This study demonstrated that even in healthy infants, the swallowing rate is decreased significantly, arousal is not augmented, and respiratory rate was decreased significantly in the prone position compared with the supine position. Any impairment of swallowing or arousal or potentiation of the LCR response would augment this potentially lethal reflex response.
This project was supported by the SIDS Research Foundation of South Australia and the National Health and Medical Research Council project grant.
We thank the mothers and infants for their participation in this study. We also thank Dr Peter Stewart for his biochemical analysis of the infant urine samples, all found negative for cotinine; Dr Crista Wocadlo for her assistance with the statistical analysis; and Danielle Ius for her help with the infant studies.
- Received October 28, 1998.
- Accepted March 15, 1999.
Reprint requests to (H.E.J.) Department of Neonatal Medicine, Royal Prince Alfred Hospital, Missenden Rd, Camperdown 2050 Australia. E-mail:
- SIDS =
- sudden infant death syndrome •
- LCR =
- laryngeal chemoreflex •
- EEG =
- Dwyer T,
- Ponsonby A
- ↵Johnson P, Salisbury DM, Storey AT. Apnoea induced by stimulation of sensory receptors in the larynx. In: Bosma JF, Showacre J, eds. Development of Upper Respiratory Anatomy and Function. Washington DC: US Department of Health, Education, and Welfare; 1975:160–178
- Downing SE,
- Lee JC
- Page M,
- Jeffery HE,
- Marks V,
- Post EJ,
- Wood AKW
- Page M,
- Jeffery HE,
- Post EJ,
- Wood AKW
- Boggs DF,
- Bartlett DJ
- Harding R,
- Johnson P,
- Johnston BE,
- McClelland ME,
- Wilkinson AR
- ↵Johnson P. Laryngeal induced apnea. In: Robinson RR, ed. SIDS Proceedings of the Francis E Camps Symposium. Toronto, Canada: Canada Foundation Study SIDS; 1974;231–242
- Lee JC,
- Stoll BJ,
- Downing SE
- Sullivan CE,
- Murphy E,
- Kozar LF,
- Phillipson EA
- Davies AM,
- Koenig JS,
- Thach BT
- Pickens DL,
- Schefft GL,
- Thach BT
- ↵Anders T, Emde R, Parmalee A, eds. A Manual of Standardized Terminology: Techniques and Criteria for Scoring of States of Sleep and Wakefulness in Newborn Infants. Los Angeles, CA: UCLA Brain Information Service; 1971
- ↵McKelvey G, Post E, Jeffery HE, Wood A. Sedation with promethazine profoundly affects spontaneous airway protection in sleeping neonatal piglets. Clin Exp Pharmacol Physiol. 1999
- ↵Norusis MJ. SPSS/PC+ 4.0. Chicago, IL: SPSS, Inc; 1990
- ↵Jeffery HE, Page M, Ius D. Repetitive swallowing depresses breathing in preterm infants even in wakefulness. Presented at: 5th SIDS International Conference; April 20–23, 1998; Rouen, France
- Jeffery HE,
- Heacock HJ
- Jeffery HE,
- Rahilly P,
- Read DJC
- MacFadyen UM,
- Hendry GMA,
- Simpson H
- Kahn A,
- Blum D
- Kahn A,
- Hasaerts D,
- Blum D
- ↵Griffiths R. The Griffiths Mental Development Scales From Birth to 2 Years: Manual 1996. Revision 18. Henley, Oxon: The Test Agency Ltd, Cray House;1996
- Copyright © 1999 American Academy of Pediatrics