The effect of flow limitation on the cardiorespiratory response to arousal from sleep under controlled conditions of chemostimulation in healthy older adults


Mary J. Morrell, Royal Brompton Hospital, Sydney Street, London, UK.
Tel.: +44 207 351 8911;
fax: +44 (0)20 7351 8911;


The influence of flow limitation on the magnitude of the cardiorespiratory response to arousal from sleep is of interest in older people, because they experience considerable flow limitation and frequent arousals from sleep. We studied older flow-limiting subjects, testing the hypothesis that the cardiorespiratory activation response would be larger when arousal occurred during flow limitation, compared to no flow limitation, and chemical stimuli were controlled. In 11 older adults [mean ± standard deviation (SD) age: 68 ± 5 years] ventilation was stabilized using continuous positive airway pressure, and flow limitation was induced by dialling down the pressure. Partial pressure of end-tidal carbon dioxide (PetCO2) was maintained by titration of the inspired CO2 and hyperoxia was maintained using 40% O2 balanced with nitrogen. Flow limitation at the time of arousal did not augment cardiovascular activation response (heart rate P = 0.7; systolic blood pressure P = 0.6; diastolic blood pressure P = 0.3), whereas ventilation was greater following arousals during flow limitation compared to no flow limitation (P < 0.001). The pre–post-arousal differences in ventilation reflected significant pre-arousal suppression (due to flow limitation) plus post-arousal activation. In summary, the cardiovascular response to arousal from sleep is not influenced by flow limitation at the time of arousal, when chemical stimuli are controlled in older adults. This finding may contribute to the decreased cardiovascular burden associated with sleep-disordered breathing reported in older adults, although our data do not exclude the possibility that flow limitation in the presence of mild hypoxic hypercapnia could increase the cardiovascular response to arousal.


Sleep-disordered breathing (SDB) encompasses a wide spectrum of severity, ranging from snoring and mild flow limitation (FL) to hypopnoea and apnoea. At the severe end of this continuum, epidemiological studies have shown that patients with obstructive sleep apnoea (OSA) have a threefold increased likelihood of developing hypertension over 4 years, independent of other risk factors (Nieto et al., 2000; Peppard et al., 2000). The mechanisms that could contribute to this association include repetitive intermittent hypoxia (Fletcher et al., 1992) and large swings in negative intrathoracic pressure (Somers et al., 1993) and acute cardiovascular and respiratory activation on arousal from sleep (Davies et al., 1993; O’Driscoll et al., 2004; Trinder et al., 2001). More subtle respiratory events during sleep, such as FL and hypopnoea, also produce transient arousals (Exar and Collop, 1999). However, the effect of FL on the cardiorespiratory activation response during arousal from sleep is unclear. Some data suggest that the heart rate (HR) response to an arousal from sleep is elevated in OSA patients treated suboptimally with continuous positive airways pressure (CPAP) to induce FL (Jordan et al., 2004a).

The influence of FL on the magnitude of the cardiovascular response to arousal from sleep is of considerable interest in older adults, as they tend to experience FL and have relatively large numbers of arousals from sleep (Browne et al., 2003; Mathur and Douglas, 1995). The aim of our study was to investigate the effect of FL on the cardiorespiratory response to arousal from sleep in older people while controlling for any differences in chemostimulation. We tested the hypothesis that the cardiorespiratory activation response at arousal from sleep would be larger during acute periods of FL, induced by a dialling down of CPAP in hyperoxia, compared to when ventilation was maintained by CPAP, and thus no FL limitation occurred. In addition subjects were also under hyperoxic conditions and matched partial pressure of end-tidal carbon dioxide (PetCO2).



Healthy older people (60–75 years) were recruited through the volunteer registry of the National Ageing Research Institute, Melbourne. Twenty-one subjects, who were non-smokers, free from physical illness and not on medication, underwent an overnight screening sleep study to ascertain if they experienced periods of FL. Fifteen subjects had FL and returned for a second night of data collection; however, four of these subjects were subsequently excluded because of insufficient FL arousals (i.e. fewer than three FL arousals). Thus, all the 11 subjects included in the final analysis were prone to FL (see Table 1 for details). The study was approved by the University of Melbourne, Human Research Ethics Committee; subjects signed consent forms prior to their participation in the study and were reimbursed for their time commitment to the study.

Table 1.   Subject demographics
GroupMean ± SD
  1. The subjects were prone to flow limitation during sleep, which was reversed using continuous positive airways pressure (CPAP), and induced during dial-down. BMI, body mass index; BP, blood pressure; SD, standard deviation.

n (males/females)11 (7/4)
Age (years)68 ± 5
BMI (kg m−2)24 ± 3
Office BP (mmHg)
 Systolic BP130 ± 9
 Diastolic BP76 ± 8

Sample size

Previous data show that OSA patients experience a 12 ± 2% [mean ± standard deviation (SD)] increase in HR during arousal from sleep when on optimal CPAP (i.e. no-FL), compared to 9 ± 2% on suboptimal CPAP (Jordan et al., 2004a). To detect a similar difference in the present study, a minimum of eight FL subjects would be required, power = 0.9, rejection rule α = 0.05%.


Subjects were studied on two to four non-consecutive nights over 4 weeks; the number of nights depended on how long it took to collect sufficient trials, up to a maximum of four nights. Both FL and no-FL trials were collected on each night. Prior to each night subjects were asked to abstain from alcohol and caffeine for 24 h. After attachment of the recording devices, subjects went to bed. All data were collected in the supine position. Arousals were induced using an adjustable auditory stimulus (5 dB increments from 50 dB up to 70 dB maximum) following at least 2 min of undisturbed non-rapid eye movement (NREM) Stage 2 sleep (Trinder et al., 2006). CPAP was used to maintain a stable respiratory baseline. On FL trials, FL was induced by lowering the CPAP level over five breaths; the tone was then sounded to arouse the subject during FL (Fig. 1). On resumption of sleep, CPAP was increased and the tone was sounded to induce an arousal during no-FL. FL and no-FL trials were induced alternately throughout the night. Inspired CO2 was titrated to maintain constant PetCO2 during FL and no-FL trials. A hyperoxic gas mix (40% O2 balanced with nitrogen) was added to the breathing circuit via capillary tubing at the distal end of the CPAP tubing to maintain hyperoxic conditions, limiting any hypoxic stimulation during the FL trials.

Figure 1.

 Original trace showing induction of flow limitation during sleep. In this example an arousal is induced using an auditory tone to produce a flow limited trial. CPAP, continuous positive airway pressure; ECG, electrocardiogram; EEG, electroencephalogram; PetCO2, partial pressure of end-tidal carbon dioxide (PetCO2).


Sleep–wake status was monitored via electroencephalograms (EEG: C3/A2, C4/A1, O1/A2), electro-oculograms (EOG: right and left) and electromyograms (EMG: submentalis muscle), HR from a single-channel ECG and beat-by-beat blood pressure (BP) via finger plethysmography (Model 2; Portapres, Amsterdam, Netherlands). CPAP was applied via a full face mask, and airflow was measured using a heated pneumotachometer (model 3719, flow range 0–100 L min−1; Hans Rudolf, Shawnee, KS, USA) plus a differential pressure transducer (DP45-14, range ± 2.25 cm H2O; Validyne, Northridge, CA, USA). Mask pressure (DP45-14, range ± 2.25 cm H2O; Validyne), end-tidal CO2 and O2 were also monitored via mask ports (CD-3A, S-3A/I; Ametek, Berwyn, PA, USA). All signals were amplified, filtered (7D polygraph; Grass Instruments, West Warwick, RI, USA) and digitized (CED 1401; Cambridge Electronic Design, Cambridge, England).


Arousals were identified by two experienced sleep scorers according to standard criteria (Sleep Disorders Atlas Task Force of the American Sleep Disorders Association, 1992). An additional scorer made the final decision in the event of disagreements. Arousals associated with body movements, or which lasted longer than 15 s, were excluded. The five breaths before the arousal were classified as FL or no-FL trials by the same two scorers, with adjudication by a third scorer. The presence of FL was based on any discernable flattening or distortion of the inspiratory air flow signal on at least three of the five breaths (Ayappa et al., 2000; Condos et al., 1994; Hosselet et al., 1998). For each trial, the variables analysed were minute inspiratory ventilation (Vi), HR, BP. The PetCO2 values that were used to match chemostimulation across conditions were the first two arousal breaths, i.e. when clear end-tidal values can be seen.

Mean values were calculated on a breath-by-breath, or beat-by-beat, basis for the five breaths or 15 beats occurring immediately before and after the arousal. To assess the affect of FL during the pre-arousal period, FL trials were compared to no-FL trials using a 2 (trial type) × 5 (breaths), or 15 (beats), repeated-measures analysis of variance (anova). Further overall mean pre-arousal values were calculated. The effect of FL on the response to arousal was assessed by comparing FL and no-FL trials over the 10 pre- and post-arousal breaths, or 30 beats, using a 2 (trial type) × 10 breaths, or 30 beats, repeated-measures anova. The level of PetCO2 was not included in analyses, as it was matched across all trials.


The median (range) number of FL, and no-FL arousal trials per subject was five (3–14) and six (4–14), respectively. The duration of the arousals under both conditions was similar [arousal duration: FL trials, 10 (8–13) s versus no-FL trials, 10 (7–12) s]. During FL trials CPAP was dialled down to a significantly lower pressure than the pressure applied to maintain stable breathing during no-FL trials (CPAP: FL trials, 2.2 ± 0.2, versus no-FL trials, 3.2 ± 0.2 cm H2O; P = 0.006). PetCO2 was similar during the FL compared to no-FL trials 37.9 ± 3.9 mmHg and 37.4 ± 3.1 mmHg, respectively, t(10) = 0.838, P = 0.42.

Respiratory and cardiovascular data for five breaths or 15 heartbeats prior to arousal from sleep are presented in Table 2. As expected, the mean Vi prior to FL trials was less than that prior to no-FL trials (P < 0.001), with the difference being significant at each breath position (Fig. 2). In contrast, there was no difference in HR or diastolic BP prior to FL, compared to no-FL trials (HR, P = 0.3; diastolic BP, P = 0.2), although systolic BP was elevated significantly in the FL trials (P = 0.04; Fig. 3).

Table 2.   Mean respiratory and cardiovascular data prior to arousal from sleep
Trial conditionNo-FL arousalsFL arousals
  1. Values are mean ± standard error of the mean (SEM) group data for an average of five breaths (respiratory data) or 15 heartbeats (cardiovascular data) prior to arousals from Stage 2 sleep, during no-FL and FL trials. BP, blood pressure; HR, heart rate; FL, flow limitation; no-FL, no flow limitation; Vi, inspiratory minute ventilation. Significant difference between conditions: *P < 0.05; ***P < 0.001.

V i (L min−1)7.1 ± 0.55.0 ± 0.4***
HR (beats min−1)55 ± 256 ± 3
Systolic BP (mmHg)98 ± 5102 ± 5*
Diastolic BP (mmHg)45 ± 447 ± 3
Figure 2.

 Group mean ± standard error of the mean (SEM) breath-by-breath changes in ventilation (Vi) responses to arousals (>3 s) from Stage 2 sleep following periods of flow limitation (FL: filled circles) and no flow limitation (no-FL: open circles). Vertical dashed line represents start of arousal. *P < 0.05, **P < 0.01, ***P < 0.001

Figure 3.

 Group mean ± standard error of the mean (SEM) beat-by-beat changes in heart rate (HR), systolic blood pressure (BP) and diastolic BP responses to arousals (>3 s) from Stage 2 sleep following periods of flow limitation (FL: filled circles) and no flow limitation (no-FL: open circles). Vertical dashed line represents start of arousal. *P < 0.05

The pre–post analysis (10 breaths or 30 beats) indicated a significant condition × breath interaction (P < 0.001), with the post-arousal values for FL being significantly higher than for no-FL at each breath position (Fig. 2). The differences in ventilation pre- versus post-arousal reflected two components, a pre-arousal suppression (due to FL) and a post-arousal activation (the effect we wished to test), with the latter being indicated by the significant elevation of ventilation in the FL condition above the no-FL condition post-arousal.

As indicated above, systolic BP was elevated in the FL condition prior to arousal (Table 2), an effect that extended five beats into the post-arousal period (Fig. 3). However, there was no overall post-arousal difference between conditions and no evidence for greater activation in the FL condition post-arousal. Finally, there was no difference in HR or diastolic BP between conditions either prior to or post-arousal (P > 0.05; Fig. 3).


The main finding of this study was that FL did not augment the cardiovascular response during arousal from sleep in older adults when chemostimulation was controlled with hyperoxia and matched PetCO2. However, the ventilatory response was greater during FL compared to no-FL arousals. This was a consequence of the reduction in Vi prior to arousal, and an elevation in the post-arousal response. Our findings support the notion that the respiratory response to arousal from sleep is dependent on the pre-arousal state of the airway, whereas the cardiovascular response is independent of respiratory conditions prior to the arousal.

The cardiovascular response on arousal from sleep following an apnoeic event is almost double that which occurs during a spontaneous arousal from sleep (Ali et al., 1992). Moreover, OSA patients experience an augmented HR response following arousal from sleep during FL induced using suboptimal CPAP (Jordan et al., 2004b). Therefore, we hypothesized that FL would also produce an augmented cardiovascular response to arousal from sleep in older adults. We interpret the absence of an augmented cardiovascular response during FL as support for a ‘reflex’ cardiovascular response, which is dependent on a change in state from sleep to wake, but not the state of the upper airway, and the intensity of the concurrent ventilatory stimuli (Horner et al., 1997; Trinder et al., 2001).

Most explanations of the cardiovascular arousal response incorporate the concept that the respiratory stimuli developed during sleep drives an exaggerated response at the onset of wakefulness. However, the failure of hypercapnia, hypoxia, hypercapnic hypoxia or brief occlusions (Catcheside et al., 2001; O’Driscoll et al., 2004, 2005; Ringler et al., 1990a) to increase the magnitude of the activation in humans favour the idea of a reflex response, supporting the findings of this study.

Possible age-related changes in the mechanisms underlying the cardiovascular response to arousal from sleep

We have shown previously that the cardiovascular response on arousal from sleep is reduced in healthy older people compared to younger people (Goff et al., 2008). Our current findings show that under conditions of controlled chemostimulation FL does not contribute to the cardiovascular response in older adults. In younger people, the transient increases in BP associated with arousal are attributed to increased peripheral resistance (Morgan et al., 1996). We speculate that in older people the reduction in vascular compliance and an increase in arterial stiffness (Lakatta, 1993) inhibits the influence of the negative intrathoracic pressure swings on sympathetic activation. Older adults may also experience less of an increase in cardiac vagal activity in response to activation of the autonomic nervous system, caused by increased negative intrathoracic pressure. This may, in turn, mean that FL in older adults does not affect the magnitude of vagal withdrawal that occurs on arousal from sleep (Horner et al., 1995).

Methodological considerations

A number of limitations need to be taken into consideration when interpreting the findings of this study. Arousals from sleep were induced during hyperoxia with controlled PetCO2; therefore, we cannot rule out the possibility that under normal conditions the mild hypoxia associated with FL would increase sympathetic drive and the subsequent cardiovascular responses to arousal from sleep. In both humans and an animal model of OSA, brief periods of airflow obstruction are associated with small reductions in oxygen saturation that are sufficient to increase sympathetic activation in the absence of any obvious arousal from sleep (O’Donnell et al., 1996, 2002). Conversely, if the hypoxia associated with an apnoea is eliminated by the administration of supplemental oxygen in OSA patients, the cardiovascular response at the termination of an apnoea is not diminished (Ringler et al., 1990a). In summary, our findings suggest that while severe hypoxia is known to increase sympathetic drive, the role of mild hypoxia with FL requires further investigation.

FL can be identified via a flattening of the inspiratory flow signal measured non-invasively via a nasal canula, thermistor or pneumotachograph (Clark et al., 1998; Hosselet et al., 1998). In the present study we used a pneumotachograph and trained observers to detect the presence of discernable FL plus a reduction in Vi. Although we used observational criteria, we are confident that our measurements of FL represent an increase in of upper airways resistance and negative intrathoracic pressure.

CPAP was used in order to maintain a stable respiratory baseline during sleep. However, CPAP increases thoracic pressure which can, in turn, reduce cardiac preload and afterload (Shekerdemian and Bohn, 1999). There is a theoretical possibility that this could have influenced the BP during the no-FL conditions. Although, in practice, we think this is unlikely, as the older healthy adults in this study were on low levels of CPAP (<4 cm H2O) and the average difference in the CPAP applied across conditions was 1 cm H2O, therefore excessive changes in afterload due to the CPAP were unlikely.


The main finding of this study is that when chemical stimuli are controlled FL does not augment the cardiovascular activation response to arousal from sleep in older adults, despite exaggerated respiratory responses. Therefore, although older people are likely to experience more arousals from sleep and FL, the cardiovascular burden from the arousal activation response is unlikely to contribute to an increased cardiovascular risk. These findings do not rule out an augmentation of the cardiovascular response due to the mild hypoxia that occurs normally during FL.