There are two types of measured Pcrit (Figure 5):

Both the types of Pcrit require a slightly different measurement technique, yet can be measured using the same set-up.

Pcrit measurement techniques are based on continuous positive airway pressure (CPAP) treatment principles, which is the standard non-invasive treatment of OSA. The CPAP creates a pneumatic splint against the collapse of the structures of the pharyngeal airway by blowing the positive pressured air into the airway to keep its structures open (Figure 2) (17).

To measure the Pcrit, a generic set-up has been developed (Figure 3) (12, 18–21). The aim of the study is to supinely position, the subject, irrespective of the effect of the other positions on Pcrit. A full polysomnography (PSG) set-up is used to record various parameters. These include sleep stages through electroencephalography (EEG), surface muscular activity using electromyography (EMG) of the chin and legs, electrocardiography (ECG), and thoracic and abdominal inductive plethysmography bands to record the respiratory effort and oxygen saturation using an oximeter. To measure the airflow, a nasal mask attached to a heated pneumotachometer and a differential pressure transducer is worn by the subject. The flow and the pressure of the mask is recorded on a computer. Special effort should be put in avoiding air leaks. An esophageal pressure catheter is inserted through one of the nostrils for the measurement of the respiratory effort. A modified CPAP device capable of producing both negative and positive pressures is used. This set-up is a closed-loop. Therefore, a whisper swivel to prevent rebreathing and a bias flow of oxygen for periods of zero CPAP flow is added to the set-up after the pneumotachometer and before the CPAP device (12, 18–20).

Various methods for Pcrit measurement have been introduced. The core of most of these methods is to increase the pressure to a level where the pharyngeal airway is open and the airflow is not limited (Figure 1A), called the “holding pressure.” Then, multiple series of pressure drop (Figures 1B, 4) will be performed to obtain a pressure at which the airflow is zero (Figure 1C). Thereafter, the pressureflow relationship is calculated using the linear regression to determine the Pcrit (Figure 1).

Some methods use the peak inspiratory flow of the recorded breath signals as a measure of flow, while others use the inspiratory ventilation for Pcrit calculations (12, 22–24).

As mentioned, two types of Pcrit can be measured. During these pressure measurements, if the pressure is dropped abruptly, the measured critical closing pressure is called, the “passive Pcrit.” Passive Pcrit is measured in this way so that the airflow is recorded before the activation of the dynamic airway muscle activation (25). After finding the “holding pressure,” multiple series of pressure drops will be performed in specific pressure increments (for example, increments of 1–2 cmH₂O for five breaths each) until a pressure at which the airflow is zero. Between each pressure drop, the pressure will be set back to the holding pressure (Figure 5A). The “active Pcrit” is measured by a slow and progressive reduction of the pressure as depicted in Figure 5B. This method allocates enough time for dilator muscles of the upper airway to activate (12, 18–20, 25). In this method, the pressure is dropped from the holding pressure for a longer period of time (10 min) and there will be no returning to the holding pressure between each pressure drop (Figure 5B).

For the first time in 1984, Issa and Sulivan measured the active Pcrit in 18 subjects with OSA (12). Schwartz et al. optimized this method in healthy volunteers (22). Originally, the recording was carried out by setting a holding pressure of −1 cmH₂O for all the subjects. Subsequently, the pressure drop of 3–5 min, allowing for the upper airway muscles to be recruited, was performed throughout the whole duration of the night until the subject was unable to fall back asleep again (22).

To unify the Pcrit measurements, an “abbreviated method” was introduced (23). First, a personalized holding pressure was determined. Subsequently, a pressure drop was performed for the duration of five breaths with 1 min at the holding pressure in between each drop, instead of the long 3-to-5-min pressure drop of prior measurement methods (23). In the abbreviated method, all breaths of the pressure drop period were analyzed to establish whether they were flow-limited or not (23).

In recent studies, 3–5 breaths from a 5-breath pressure drop is used for further analysis (23, 26, 27). During a pressure drop, usually it takes two breaths after the change of the pressure to reach a steady air flow state. This might be due to the peak inspiratory flow adjustments to the end-expiratory lung volume or by the increasing respiratory effort due to arousal stimuli (26–28).

To further facilitate Pcrit calculation, a “simple method” was proposed (29). In this method, first, all breaths were plotted against their associated pressures, to find the flow-limited segment. On this pressure-flow plot, an upper inflection point (between portions a and b on Figure 1) and a lower inflection point were identified (between portions b and c on Figure 1). Hereafter, the range of pressure between these two inflection points, over which the flow is markedly varied (Figure 1B), is isolated. This range is used for calculating Pcrit (29).

Pcrit measurements using the gold standard method are performed during natural sleep. However, this is cumbersome and labor-intensive and therefore not feasible for routine practice. To overcome the problem of recurring arousals while measuring Pcrit during natural sleep and reduce the time needed to measure the upper airway collapsibility, other methods were investigated.

The normal adult pharyngeal airway has a negative Pcrit (−13.3 ± 3.2 cmH₂O) during sleep and by lowering the nasal pressure below this Pcrit, occlusion is induced (22). The more positive the Pcrit is, the more collapsible the pharynx is. The Pcrit of primary snorers is estimated to be around −6.5 ± 2.7 cmH₂O. In patients with OSA, more positive values are present, approximately around 2.5 ± 1 cmH₂O and almost never below −5 cmH₂O (3). Table 1 gives an overview of the measured Pcrit from various studies (19, 22, 54, 60, 61).

Healthy subjects are able to decrease their active Pcrit (−11.1 cmH₂O) to a lower level (−4.5 cmH₂O) after passive loading of the upper airway. However, OSA subjects seem to be unable to lower the Pcrit after loading (from active Pcrit = −0.05 cmH₂O to passive Pcrit = −1.6 cmH₂O) (62).

Obesity [body mass index (BMI) ≥ 30 kg/m²], an increase in the proportion of the soft tissue volume to the surrounding bone, enlarged lateral pharyngeal walls and soft palate, and an inferiorly located hyoid bone (increased pharyngeal length as measured by mandibular-hyoid plane distance) increase the pharyngeal collapsibility (60, 63–65).

Neck flexion decreases the pharyngeal patency and therefore increases the collapsibility. Pcrit increases during neck flexion regardless of the body position in supine, prone, or while turning the head to the side (66). In contrast, the sniffing neck position (8 cm elevation) increases the cross-sectional area of the pharynx and results in a Pcrit decrease at the retropalatal and retroglossal levels (retropalatal: 2.68 to −1.78 cmH₂O, retroglossal: 0.80 to −3.72 cmH₂O) (56). In 15 healthy individuals, Pcrit in the neutral neck position was −0.4 ± 4.4 cmH₂O. It increased to 3.7 ± 2.9 cmH₂O by neck flexion, decreased to −9.4 ± 3.8 cmH₂O by extension and remained unchanged by rotation (−2.6 ± 3.3 cmH₂O). These measurements clearly indicate that head posture might have a compelling effect on the pharyngeal collapsibility (55).

Body position during sleep affects the Pcrit as well. Changing the position from supine to lateral decreases the collapsibility (23, 57). Boudewyns et al. (23) compared the effect of supine vs. lateral body position on Pcrit values in 10 patients with obesity having a respiratory disturbance index (RDI) equal to 63.0 ± 14.6/h. Pcrit in the supine position was 1.8 and plummeted to −1.1 cmH₂O by changing the position to lateral during the measurements. However, no significant position dependency was observed in the upstream resistance values (23). Furthermore, in a different study, the lateral position significantly reduced the collapsibility from 2.02 ± 2.55 to −1.92 ± 3.87 cmH₂O, in 20 patients with OSA (57).

Pharyngeal resistance and collapsibility between men and women is not significantly different. The mean Pcrit in men and women without OSA was −10.4 ± 3.1 and −8.8 ± 2.7 cmH₂O, respectively (67). The role of gender was investigated in two groups: the first one consisting of 11 men and 11 women matched for the severity of OSA (AHI: 43.8 ± 6.1 and 44.1 ± 6.6 events/h) and the second one including 12 men and 12 women matched for BMI (31.6 ± 1.9 and 31.3 ± 1.8 kg/m²).

In the OSA-severity matched group; women had a higher BMI. Yet, no difference among genders were found for the collapsibility (0.35 ± 0.62 vs, −0.18 ± 0.87 cmH₂O, P = 0.63). In the BMI-matched group; women had less severe OSA and a lower Pcrit compared to men (−2.01 ± 0.62 vs. 1.16 ± 0.83 cmH₂O; P = 0.005). These data suggest that the effect of being overweight/obese on OSA might be greater in men compared to women and hence women have a less collapsible pharynx for any given BMI as in the male who have the same BMI (68). A recreated finite element model of the male and female pharyngeal airway confirmed this finding (69). This might be due to the different anatomical characteristics, such as a longer pharyngeal airway, an increased cross-sectional area of the soft palate, and increased airway volume in men compared to women (69).

Ethnicity and genetics affect the shape and structure of the pharyngeal airway. As an example, a comparison of Pcrit between the Whites and the Japanese Brazilians showed no statistically significant difference. However, the craniofacial characteristics that affect the Pcrit were different between the two groups. In the Whites, the collapsibility was correlated with the ratio of the tongue volume to the mandible volume. In the Japanese Brazilians, it was correlated with the cranial base angle (85).

The upper airway is more prone to collapse during sleep (3). During both the awake and sleep state, the human airway has compensatory mechanisms, such as neuromuscular reflexes to the negative pressure, to stay open. These compensatory reflexes are reduced during sleep. In subjects with OSA, this loss is more prominent during sleep compared to the healthy subjects. Therefore, as patients with OSA rely mainly upon these neuromuscular reflexes to keep the airway open, this will result in a more collapsible upper airway (33, 86, 87). Collapsibility during sleep, in healthy controls in the lateral posture was significantly higher than during the awake state (54.4 vs. 27.5%; p < 0.01) (34).

Sleep stage and the quality of sleep may affect Pcrit. According to one study, no difference between Pcrit during sleep stages N1/N2 (3.1 ± 0.4) and rapid eye movement (REM) sleep (2.4 ± 0.2 cmH₂O) was found (12). However, there was a significant difference between the collapsibility in other sleep stages and that of slow-wave sleep (4.2 ± 0.2 cmH₂O), showing a higher collapsibility during the deeper sleep (12). Contradictory to this, another study found that Pcrit during REM was 4.9 ± 1.4 cmH₂O higher than that of deeper sleep stages, implying a more collapsible upper airway during REM sleep (58).

Sleep deprivation (one night of total sleep deprivation prior to measurements) and sleep fragmentation (using auditory stimuli) in subjects with OSA resulted in a more collapsible airway (12.3 ± 6.3 cmH₂O). However, sleep deprivation in healthy subjects had no effect on collapsibility (−17.1 ± 6.8 cmH₂O) (88).

Within a single respiratory cycle, the collapsibility and upstream resistance varies between inspiration and expiration, with Pcrit during expiration being up to 4 cmH₂O more than during inspiration. This variation might be due to both local mechanical and neuromuscular factors (89).

Therapeutic CPAP levels show a strong relationship with collapsibility. The needed CPAP level, to resolve OSA, is lower in patients with a milder Pcrit (≤ -2 cmH₂O) compared to a higher Pcrit (>-2 cmH₂O) collapsibility. Therefore, the level of therapeutic CPAP might be indicative of the level of collapsibility (90). The CPAP did not affect Pcrit during N1/N2 sleep stages. However, during deep sleep, for every 1 cmH₂O rise in CPAP pressure, a 0.65 cmH₂O decrease in the Pcrit was shown (12).

Weight loss decreases collapsibility. In 23 (BMI: 42 ± 7 kg/m²) patients with obesity, severe OSA (disordered breathing rate: 83.3 ± 31/h), and high collapsibility (Pcrit: 3.1 ± 4.2 cmH₂O), the effect of weight loss was evaluated (91). Thirteen patients reached the goal of 5% reduction in body weight and were included in the final evaluations. After 16.9 ± 10.0 months of follow-up, the disordered breathing rate decreased to 32.5 ± 35.9/h and the Pcrit to −2.4 ± 4.4 cmH₂O. The reduction in collapsibility was significantly correlated with the reduction in BMI. A decrease of Pcrit to below −4 cmH₂O was associated with a near-complete elimination of OSA (disordered breathing rate <20/h) (91).

Mandibular advancement devices protrude the mandible to open up the airway (92). This forward positioning of the mandible affects the pharyngeal collapsibility and decreases the Pcrit (93–95). The MAD therapy significantly increases both the passive ventilation (p = 0.002) and active ventilation (P < 0.001) (95). Pcrit in four different positions was measured. Neutral (−4.2 ± 2.9 cmH₂O) centric occlusal meaning zero mm advancement (−7.1 ± 5.2 cmH₂O), incisors aligned (−10.7 ± 4.4 cmH₂O), and 75% of maximal advancement of the mandible (−13.3 ± 3.2 cmH₂O). Mandibular protrusion to the “incisors aligned” position significantly decreased the collapsibility. Further advancement toward the maximal position reduced the Pcrit even more (96). These results were also confirmed by another study. Ayuse et al. showed that the advancement of the mandible from centric occlusion to maximal protrusion by 7.1 ± 1.2 mm, significantly reduces Pcrit (from 1.9 ± 2.9 cmH₂O to −7.3 ± 1.9 cmH₂O) (93). It has been demonstrated that the effect of mandibular advancement is dose-dependent and works through improving the passive pharyngeal airway rather than affecting the function of the genioglossus muscle (94). In 12 patients, the measurement of collapsibility in three different positions of neutral, 50%, and 100% of maximum comfortable mandibular advancement clearly manifested this stepwise dose-dependent effect of mandibular advancement in reducing the collapsibility with a decrease in Pcrit from 1.8 ± 3.9 cmH₂O to −0.9 ± 2.9 cmH₂O, and −4.0 ± 3.6 cmH₂O, respectively (94).

A less collapsible airway is an independent predictor of response to MAD, defined as a percent reduction in AHI (non-linear multivariate model, r² = 0.70) (95). Non-responders to MAD therapy were shown to have a higher baseline collapsibility (bivariate model, p = 0.014) (97). This shows the importance of Pcrit not only in the follow-up of MAD therapy and efficacy evaluations, but also in the prediction of treatment success and the choice of the therapy option (95, 98).

Using FE modeling techniques, the effect of three different therapy options on collapsibility was investigated. The benefit of using such a model is that it allows for comparing all manipulations in the same individual. The baseline Pcrit was −13 cmH₂O. All three treatment modalities, such as mandibular advancement (−21 cmH₂O), palatal resection (−18 cmH₂O), and palatal stiffening by palatal implants (−17 cmH₂O) decreased the collapsibility (99). In another model with baseline Pcrit of −2 cmH₂O, uvulopalatopharyngoplasty (UPPP) reduced the collapsibility with tissue resection to a certain sufficient level. However, after sufficient palatal resection, Pcrit was improved. However, with further palatal resection, retro-lingual obstruction occurred and no amount of further resection could improve the collapsibility (100). Palatal implants with or without tongue stiffening did not change the Pcrit. While stiffening of the tongue via implanting in the same direction as the genioglossus muscle fibers and in the C region of the tongue (adjacent to the soft palate), without any palatal component, gave the best results in the reduction of collapsibility (−7 cmH₂O) (100). Similar to MAD therapy, these studies again emphasize the value of baseline Pcrit in response to prediction and in the follow-up of OSA therapies.

There are several pharmacological options to treat OSA. Acetazolamide reduces the breathing events and increases the inspiratory drive (105, 106). However, acetazolamide does not affect the collapsibility (107). Recently a combination of a noradrenergic drug atomoxetine plus the antimuscarinic oxybutynin has been shown to reduce the severity of OSA (108). Both the combination of atomoxetine-oxybutynin and each medication alone improved the collapsibility of the pharyngeal airway (109). Trazodone increases the respiratory arousal threshold in subjects with OSAs without affecting the pharyngeal collapsibility (110).