SEARCH

SEARCH BY CITATION

Keywords:

  • body posture;
  • collapsibility;
  • pharyngeal critical collapsing pressure;
  • repeatability;
  • sleep stage;
  • variability

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Declarations of Interest
  9. Acknowledgements
  10. References

The critical pressure at which the pharynx collapses (Pcrit) is an objective measurement of upper airway collapsibility, an important pathogenetic factor in obstructive sleep apnoea. This study examined the inherent variability of passive Pcrit measurement during sleep and evaluated the effects of sleep stage and body posture on Pcrit. Repeated measurements of Pcrit were assessed in 23 individuals (15 male) with diagnosed obstructive sleep apnoea throughout a single overnight sleep study. Body posture and sleep stage were unrestricted. Applied upper airway pressure was repetitively reduced to obtain multiple measurements of Pcrit. In 20 subjects multiple measurements of Pcrit were obtained. The overall coefficient of repeatability for Pcrit measurement was 4.1 cm H2O. Considering only the lateral posture, the coefficient was 4.8 cm H2O. It was 3.3 cm H2O in the supine posture. Pcrit decreased from the supine to lateral posture [supine mean 2.5 cm H2O, 95% confidence interval (CI) 1.4–3.6; lateral mean 0.3 cm H2O, 95% CI −0.8–1.4, = 0.007] but did not vary with sleep stage (= 0.91). This study has shown that the overall coefficient of repeatability was 4.1 cm H2O, implying that the minimum detectable difference, with 95% probability, between two repeated Pcrit measurements in an individual is 4.1 cm H2O. Such variability in overnight measures of Pcrit indicates that a single unqualified value of Pcrit cannot be used to characterize an individual’s overall collapsibility during sleep. When within-subject variability is accounted for, change in body posture from supine to lateral significantly decreases passive pharyngeal collapsibility.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Declarations of Interest
  9. Acknowledgements
  10. References

The defining feature of obstructive sleep apnoea (OSA) is episodic collapse of the pharyngeal airway during sleep (Remmers et al., 1978). Clinically, the severity of OSA is described in terms of the apnoea–hypopnoea index (AHI), the number of episodes of partial or complete pharyngeal collapse per hour of sleep. More specific measurement of the inherent structural collapsibility of the pharynx can be obtained from its critical collapsing pressure (Pcrit), the nasal pressure at which the hypotonic or ‘passive’ pharynx collapses and inspiratory flow is abolished (Schwartz et al., 1998). Pcrit is used widely in research settings and has significant potential clinical applications (Gold et al., 2002); Pcrit is greater in individuals with severe OSA (2.5 ± 1.5 cm H2O) than snorers (−1.6 ± 1.4 cm H2O) or control subjects (−6.5 ± 2.7 cm H2O) (Gleadhill et al., 1991).

While these group data provide evidence that pharyngeal collapsibility is an important contributor to the pathogenesis of OSA, the extent to which a single value of Pcrit can adequately describe an individual’s propensity to collapse remains unclear. Many factors known to influence Pcrit vary during sleep, including: head and body posture (Boudewyns et al., 2000; Penzel et al., 2001; Walsh et al., 2008b); lung volume (Squier et al., 2010); degree of mouth opening (Ayuse et al., 2004); saliva production (Kirkness et al., 2003); and sleep stage (Schwartz et al., 1998). To date, the variability of Pcrit over the course of a normal, uncontrolled night’s sleep in individuals with OSA has not been thoroughly characterized. Such information would help to define the required magnitude of change in response to an experimental intervention or clinical treatment to be considered clinically significant.

Hence, the primary purpose of this study was to quantify the within-subject overnight variability in Pcrit. A secondary aim was to determine the effects of sleep stage and body posture on Pcrit using statistical analyses that account for this Pcrit variability.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Declarations of Interest
  9. Acknowledgements
  10. References

Overall study design

Each subject underwent a single in-laboratory sleep study throughout which multiple measurements of Pcrit were obtained. Sleep posture was documented but unrestricted.

Participants

Subjects with varying severities of OSA were recruited from sequential patients attending a hospital sleep clinic. All were currently on continuous positive airway pressure (CPAP) therapy. No specific selection criteria in terms of OSA severity, age or body mass index (BMI) were applied. Each subject provided written, informed consent prior to participation in the study, which was approved by the hospital’s Human Research Ethics Committee.

Equipment and measurements

Following standard polysomnography (PSG) (E-series; Compumedics, Abbotsford, VIC, Australia), scoring of sleep stage and arousals was performed by the same experienced technologist according to American Academy of Sleep criteria (Iber et al., 2007). Body posture was scored as lateral or supine using the position sensor and confirmed via a real-time video recording. Respiratory flow was measured using a calibrated pneumotachograph (Flowstat Prototype; Key Technologies, Baltimore, MD, USA) attached to a continuous positive airway pressure (CPAP) mask (Mirage Activa; ResMed, BellaVista, NSW, Australia; or Zest; Fisher-Paykel Healthcare, Ringwood, VIC, Australia). A sample port on the mask allowed mask pressure to be measured continuously (143PC, micro Switch; Honeywell, Morristown, NJ, USA). Continuous airway pressure was delivered via a customized machine (nCPAP Pcrit; ResMed, BellaVista, NSW, Australia) which permitted rapid changes to any pressure from −20 cm H2O to +20 cm H2O when required.

Each subject was also instrumented with a thin intranasal catheter to record pressure downstream of the site of pharyngeal collapse/narrowing (MPC-500; Millar Instruments, Houston, TX, USA) to provide an indication of respiratory effort. Mask pressure, airflow and epiglottic pressure were recorded on a PowerLab data acquisition and analysis system (16s; ADInstruments, Sydney, NSW, Australia).

Protocol

Throughout the night a positive ‘holding pressure’ was applied sufficient to eliminate evidence of inspiratory flow limitation (Boudewyns et al., 2000). Once uninterrupted sleep was evident on PSG, the applied nasal pressure was reduced for five breaths to a level sufficient to induce flow limitation, as described previously (Fig. 1), and then returned to the holding pressure as per passive Pcrit protocol (Boudewyns et al., 2000). Once all physiological measurements including oxygen saturation had stabilized, nasal pressure was again reduced to induce a different degree of flow limitation. Sequences of at least three pressure reductions sufficient to produce different levels of flow limitation were required to generate a single Pcrit measurement. These sequences were repeated as many times as possible throughout each sleep study. Subjects were offered Temazepam (up to 20 mg) to aid with sleep maintenance and a chin strap or mouth tape was applied to prevent mouth leak.

image

Figure 1.  A single pressure reduction during stage N2 sleep encompassing flow limitation and the development of maximal inspiratory flow accompanied by further reductions in epiglottic pressure. EEG: electro-encephalography represented by lead C3-A2; EOG: electro-oculography; ECG: electrocardiography.

Download figure to PowerPoint

Pcrit analysis

Passive Pcrit was determined from linear regression of the upper airway pressure–flow relationship obtained during flow limitation as described previously (Schwartz et al., 1998; Walsh et al., 2008b). Briefly, only sequences of at least three pressure reductions encompassing flow limitation in breaths 3–5 of the pressure reduction were considered, and any breaths associated with arousal or sleep stage transitions were excluded (Boudewyns et al., 2000). Flow limitation was identified as the deviation of the tracking of the epiglottic pressure trace from the airflow trace (Boudewyns et al., 2000; Walsh et al., 2008b). On the occasions when the epiglottic pressure sensor moved into or above the collapsible segment of the upper airway and a reliable trace was not obtained, flattening of the flow profile early in inspiration was taken as evidence of the presence of flow limitation (Penzel et al., 2001).

The mean mid-inspiratory flow and corresponding mask pressure from breaths 3–5 for each pressure drop within a sequence of pressure reductions were calculated and plotted (Walsh et al., 2008b). Extrapolation of the linear regression to the x-intercept defined Pcrit. Pcrit values more than 3.0 cm H2O away from the lowest obtained nasal pressure were excluded from statistical analysis due to the magnitude of extrapolation required. All analyses were performed by a single investigator.

Statistical analyses

A mixed-effects model, which accounted for within-subject variability within sleep stages and body postures, was used to evaluate the within-group effects of sleep stage and body posture. This model also allowed for the designation of subject-within-body posture by sleep stage as a random effect, thereby accounting for the possibility that each subject might be affected differently by sleep stage and body posture effects on Pcrit (Cleophas and Zwinderman, 2008). Initially, sleep stage, body posture and their interaction were designated as fixed effects. AHI and body mass index (BMI) were added as covariates. Backwards elimination of factors not found to have a statistically significant effect on Pcrit was also performed.

Coefficients of repeatability were selected as the measure of within-subject variability and were calculated overall, within body postures, within non-rapid eye movement (NREM) in each body posture and within each body posture with and without Temazepam use. Briefly, the coefficient of repeatability is defined as the value below which two single test results may be expected to lie with 95% probability (Altman and Bland, 1983) and was calculated from the mean-squared error from one-way analysis of variance (anova) analysis. If this value is small, the test can be considered ‘repeatable’. To ensure that the variability of repeated measurements was not related to the magnitude of the measurement, linear regression of the standard deviation of repeated measurements against mean Pcrit and linear regression analysis was applied (Altman and Bland, 1983).

All analyses were performed on PASW version 18.0 (SPSS Inc., Chicago, IL, USA). A significance level of < 0.05 was regarded as statistically significant in all analyses. Results are displayed as mean ± standard error unless otherwise stated.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Declarations of Interest
  9. Acknowledgements
  10. References

Twenty-three subjects (15 male/eight female) were studied. Data were only able to be used from 20 subjects (Table 1), as one was unable to tolerate the intranasal catheter and aroused during all pressure reductions, in another it was difficult to induce flow limitation and another had difficulty maintaining sleep despite pharmacological assistance. A total of 146 Pcrit measurements were obtained in 20 subjects (mean number of repeated measurements ± standard deviation; 1.7 ± 2.2 in stage N2 lateral, 2.7 ± 3.2 in stage N2 supine, 1.5 ± 2.0 in stage N3 lateral, 0.7 ± 0.9 in stage N3 supine, 0.2 ± 0.4 in REM lateral, 0.6 ± 0.7 in REM supine). It was not possible to obtain measurements of Pcrit in all sleep stages (N2, N3, REM) and postures (lateral, supine) in all subjects. The correlation coefficient for linear regression of inspiratory flow against nasal pressure across all 146 Pcrit measurements was 0.94 ± 0.07 (range 0.63–1.00).

Table 1.   Subject characteristics
 Males (n = 14)Females (n = 6)All (n = 20)
Mean ± SDRangeMean ± SDRangeMean ± SD
  1. BMI: body mass index; AHI: apnoea–hypopnoea index; NREM: non-rapid eye movement; REM: rapid eye movement; SaO2: oxyhaemoglobin saturation; SD: standard deviation.

Age, years55 ± 939–7260 ± 454–6556 ± 8
BMI, kg m234.8 ± 7.728.0–51.135.2 ± 7.431.0–44.734.9 ± 7.4
AHI, h−148.3 ± 27.26.5–101.049.2 ± 22.618.2–79.348.5 ± 25.3
NREM AHI, h−144.2 ± 28.85.1–101.842.2 ± 16.717.9–64.143.6 ± 25.3
REM AHI, h−146.0 ± 20.710.0–79.639.0 ± 33.70.0–77.343.9 ± 24.6
Mean SaO2, %91 ± 482–9592 ± 289–9491 ± 3
Minimum SaO2, %82 ± 866–9385 ± 479–8983 ± 7
Central apnoea index, h−10.9 ± 1.30.0–2.10.4 ± 0.80.0–2.00.7 ± 1.2
Obstructive apnoea index, h−18.2 ± 20.20.0–75.91.3 ± 2.90.0–7.16.1 ± 17.1
Mixed apnoea index, h−11.0 ± 3.40.0–12.80.1 ± 0.10.0–0.20.7 ± 2.9
Hypopnoea index, h−138.2 ± 21.06.3–70.147.4 ± 21.617.6–79.341.0 ± 21.1

The overall repeatability coefficient was 4.1 cm H2O, which incorporated within-subject variability due to sleep stage and body posture. After stratification by body posture and considering only NREM sleep, the coefficients were 4.8 cm H2O in the lateral posture and 3.4 cm H2O for the supine posture. There was no tendency for within-subject variability to change with magnitude of Pcrit mean (= 0.94 for overall, = 0.56 for supine only, = 0.32 for lateral only).

Temazepam was requested in 10 subjects to assist sleep transition. Characteristics, including Pcrit of these 10 subjects (age 57 ± 10 years; AHI 44 ± 27 events h−1; BMI 33 ± 7 kg m2; Pcrit 1.5 ± 0.4 cm H2O) were not significantly different from the 10 subjects who did not use Temazepam (age 54 ± 6 years; AHI 52 ± 26 events h−1; BMI 37 ± 8 kg m2; Pcrit 1.6 ± 0.4 cm H2O). The coefficients of repeatability were similar for the with Temazepam and without Temazepam groups within each posture (lateral posture with Temazepam 5.0 cm H2O; without Temazepam 4.6 cm H2O; supine posture with Temazepam 3.8 cm H2O; without Temazepam 2.7 cm H2O).

Pcrit was unaffected by sleep stage (Fig. 2a), although body posture significantly influenced Pcrit with mean values of 0.3 ± 0.5 cm H2O for lateral body posture and 2.5 ± 0.5 cmH2O for supine body posture (Fig. 2b). AHI was a statistically significant covariate in the mixed effects analysis (= 0.002), although BMI was not (= 0.283). There was no significant interaction between sleep stage and body posture (= 0.491).

image

Figure 2.  (a) The estimated mean critical collapsing pressure (Pcrit) in each sleep stage as calculated by the mixed effect model. REM: rapid eye movement; N2: stage non-rapid eye movement (NREM)-2; N3: stage NREM-3;. error bars represent standard error of the mean. (b) The estimated mean critical collapsing pressure (Pcrit) in each body posture as calculated by the mixed effect model. Error bars represent standard error of the mean.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Declarations of Interest
  9. Acknowledgements
  10. References

The present study is the first to quantify within-subject variability of passive Pcrit by calculating coefficients of repeatability. The overall coefficient of repeatability was 4.1 cm H2O, implying that the minimum detectable difference, with 95% probability, between two repeated Pcrit measurements in an individual is 4.1 cm H2O. The coefficient of repeatability for measurements obtained in the lateral and supine postures, irrespective of sleep stage, was 4.8 cm H2O and 3.3 cm H2O, respectively. In NREM sleep, the coefficient of repeatability for measurements obtained in the lateral and supine postures was 4.8 cm H2O and 3.4 cm H2O, respectively.

Recently, the within-subject limits of agreement of differences between two within-night repeated Pcrit measurements in the supine posture was quantified as ±3.1 cm H2O (Kirkness et al., 2010). Methodological differences between this study and ours may account for the modest difference between the present study’s coefficient of repeatability of 3.3 cm H2O in the supine posture and the limits of agreement reported by Kirkness et al. Specifically, Kirkness et al. generated two Pcrit values using two sequences of Pn:Vmax data for each Pcrit at two time-points across the night. In contrast, the present study included Pcrit measurements generated from single sequences of Pn:Vmax data and did not have a maximum number of repeated Pcrit measurements, capturing within-subject variability across the sleep period.

Our findings suggest that factors other than posture and sleep stage contribute to the within-subject variability of Pcrit measurement. Such factors could include overnight changes in neck posture (Walsh et al., 2008b), arousal-related saliva production (Kirkness et al., 2003), degree of mouth opening (Ayuse et al., 2004) and/or variations in end-expiratory lung volume (Squier et al., 2010), all of which influence pharyngeal collapsibility. While it is difficult to control these factors adequately during natural sleep and thereby reduce the Pcrit variability, it is possible to do so using general anaesthesia, under which condition the variability of repeated Pcrit measurements is reported to be markedly decreased (coefficient of repeatability 2.8 cm H2O) (Walsh et al., 2008b).

When this variability was accounted for, the supine posture was associated with a more collapsible passive upper airway than the lateral posture, consistent with previous investigations (Boudewyns et al., 2000; Penzel et al., 2001). The magnitude of Pcrit change between body postures (2.2 cm H2O) was slightly less than observed previously (2.9 cm H2O) (Boudewyns et al., 2000). This may have been due to the incorporation of Pcrit measurement variability in the mixed-effect model, as well as the possibility that repeated measurements were made in slightly different head positions despite maintaining a similar whole body posture. Mechanisms underlying this postural effect include changes in upper airway lumen shape and end-expiratory lung volume (Squier et al., 2010; Walsh et al., 2008a).

Sleep stage was not found to have a statistically significant effect on passive Pcrit in this study, whereas AHI is known to be influenced by sleep stage (Ratnavadivel et al., 2010). Passive Pcrit reflects collapsibility of the passive upper airway whereas AHI is subject to both anatomic and neural influences. Hence, it is likely that this apparent discrepancy reflects the influence of neurophysiological factors which are not captured in the passive Pcrit measurement, such as altered arousal threshold (Ratnavadivel et al., 2010), upper airway dilator muscle activation (Remmers et al., 1978) and unstable respiratory control (Penzel et al., 2001). Our use of ‘passive’ Pcrit measures rather than ‘active’ Pcrit (Schwartz et al., 1998) acts to reduce the influence of varying degrees of muscle activation, which would be an added influence on within-subject variability of the measure.

One limitation of this study was its observational design, which resulted in differing numbers of repeated measurements between individuals and between sleep stage and posture combinations. Thus the data were vulnerable to the effect of regression towards the population mean. Further, the relatively smaller number of Pcrit measurements made in REM sleep limited the strength of the mixed-effect analyses.

Conclusion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Declarations of Interest
  9. Acknowledgements
  10. References

This study demonstrated that there is substantial variability in overnight measures of Pcrit in individuals such that a single unqualified value of Pcrit cannot be used to characterize an individual’s overall collapsibility during sleep. This variability appears likely to be due to variations in factors such as body position, neck posture, mouth opening and lung volume. It suggests that multiple measures should be obtained during sleep that record and account for the influence of these factors in analysis, or that the measures are made under strictly defined and controlled conditions. The variability documented here suggests that a single measure of Pcrit made under undefined conditions could not be regarded confidently as different from other measures unless that difference exceeded 4.1 cm H2O, the coefficient of repeatability of the measurement. Where the within-subject variability is accounted, for we found that change in body posture from supine to lateral significantly decreases passive Pcrit, whereas sleep stage did not appear to affect this passive measure.

Declarations of Interest

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Declarations of Interest
  9. Acknowledgements
  10. References

This study was funded by a NHMRC Project Grant (No. 572647). J.O. received a BMedSc Foundation Professor’s Scholarship from the University of Western Australia. P. E. is funded by a NHMRC Senior Research Fellowship no. 513704. Nil other interests declared by G. T., S. T., D. H. and J. W.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Declarations of Interest
  9. Acknowledgements
  10. References

We would like to thank Adam Benjafield and Glenn Richards from the ResMed Science Center for providing the Pcrit machine with which we made our measurements.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Declarations of Interest
  9. Acknowledgements
  10. References
  • Altman, D. G. and Bland, J. M. Measurement in medicine: the analysis of method comparison studies. J. R. Stat. Soc. Series D, Statistician, 1983, 32: 307317.
  • Ayuse, T., Inazawa, T., Kurata, S. et al. Mouth-opening increases upper-airway collapsibility without changing resistance during midazolam sedation. J. Dent. Res., 2004, 83: 718722.
  • Boudewyns, A., Punjabi, N., Van De Heyning, P. H. et al. Abbreviated method for assessing upper airway function in obstructive sleep apnea. Chest, 2000, 118: 10311041.
  • Cleophas, T. and Zwinderman, A. Random effects models in clinical research. Int. J. Clin. Pharmcol. Ther., 2008, 46: 421427.
  • Gleadhill, I. C., Schwartz, A. R., Schubert, N., Wise, R. A., Permutt, S. and Smith, P. L. Upper airway collapsibility in snorers and in patients with obstructive hypopnea and apnea. Am. Rev. Respir. Dis., 1991, 143: 13001303.
  • Gold, A. R., Marcus, C. L., Dipalo, F. and Gold, M. S. Upper airway collapsibility during sleep in upper airway resistance syndrome. Chest, 2002, 121: 15311540.
  • Iber, C., Ancoli-Israel, S., Chessonn, A. and Quan, S. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications, 1st edn. American Academy of Sleep Medicine, Westchester, IL, 2007.
  • Kirkness, J. P., Madronio, M., Stavrinou, R., Wheatley, J. R. and Amis, T. C. Relationship between surface tension of upper airway lining liquid and upper airway collapsibility during sleep in obstructive sleep apnea hypopnea syndrome. J. Appl. Physiol., 2003, 95: 17611766.
  • Kirkness, J.P., Peterson, L.A., Squier, S.B. et al. Performance characteristics of upper airway critical collapsing pressure measurements during sleep. Sleep, 2011, 34: 459467.
  • Penzel, T., Moller, M., Becker, H. F., Knaack, L. and Peter, J. H. Effect of sleep position and sleep stage on the collapsibility of the upper airways in patients with sleep apnea. Sleep, 2001, 24: 9095.
  • Ratnavadivel, R., Stadler, D., Windler, S. et al. Upper airway function and arousability to ventilatory challenge in slow wave versus stage 2 sleep in obstructive sleep apnoea. Thorax, 2010, 65: 107112.
  • Remmers, J. E., Degroot, W. J., Sauerland, E. K. and Anch, A. M. Pathogenesis of upper airway occlusion during sleep. J. Appl. Physiol., 1978, 44: 931938.
  • Schwartz, A. R., O’donnell, C. P., Baron, J. et al. The hypotonic upper airway in obstructive sleep apnea: role of structures and neuromuscular activity. Am. J. Respir. Crit. Care Med., 1998, 157: 10511057.
  • Squier, S. B., Patil, S. P., Schneider, H., Kirkness, J. P., Smith, P. L. and Schwartz, A. R. Effect of end-expiratory lung volume on upper airway collapsibility in sleeping men and women. J. Appl. Physiol., 2010, 109: 977985.
  • Walsh, J. H., Leigh, M. S., Paduch, A. et al. Evaluation of pharyngeal shape and size using anatomical optical coherence tomography in individuals with and without obstructive sleep apnoea. J. Sleep Res., 2008a, 17: 19.
  • Walsh, J. H., Maddison, K. J., Platt, P. R., Hillman, D. R. and Eastwood, P. R. Influence of head extension, flexion, and rotation on collapsibility of the passive upper airway. Sleep, 2008b, 31: 14401447.