Mask Ventilation during Induction of General Anesthesia: Influences of Obstructive Sleep Apnea

ANESTHESIOLOGISTS perform mask ventilation (MV) with proper airway maintenance maneuvers during induction of general anesthesia. Proper oxygenation is only secured by the MV when tracheal intubation is difficult or impossible. Despite the clinical significance, MV technique has changed little in contrast to significant improvement of tracheal intubation techniques and devices in the past decade. The incidence of difficult and impossible MV determined by the four-point grading scale is reported to be 1.4 to 5% and 0.15%, respectively,^1–3  and difficult MV combined with difficult tracheal intubation occurs 0.14% in adult patients undergoing general anesthesia.⁴  While these recent large epidemiology studies identified various patient characteristics as independent risk factors for difficult MV, we lack knowledge on how these risk factors make MV difficult. Elucidation of the mechanisms of difficult MV might facilitate development of a new MV technique for safer airway management during anesthesia induction.

None of the previous studies preoperatively assessed severity of sleep-disordered breathing (SDB) in all analyzed subjects among the risk factors, whereas prevalence of the SDB is as high as 69% of the general surgical population.⁵  We previously reported that pharyngeal airways of patients with SDB under general anesthesia and paralysis are narrower and more collapsible than those of non-SDB subjects,⁶  and mandibular advancement is less effective particularly in obese subjects.⁷  Anesthesia depth, consciousness, and upper airwaydilating muscle activity, which are significant determinants of the upper airway patency, dynamically change during anesthesia induction.^8–10  Therefore, SDB patients would be more vulnerable to anesthesia induction, possibly leading to insufficient tidal volume (TV) during MV. Despite these indirect evidences, no previous studies have quantitatively characterized dynamic changes of ventilation during MV in SDB patients.

Accordingly, we continuously measured the TV changes during routine MV in order to explore mechanisms of inefficient MV in adult persons. We specifically tested a hypothesis that the presence of SDB, defined as apnea hypopnea index (AHI) greater than 5 per hour, reduces the efficiency of MV assessed by TV achieved by constant mechanical ventilation.

The investigation was approved by the institutional Ethics Committee (Graduate School of Medicine, Chiba University, Chiba, Japan) and registered in University Hospital Medical Information Network Clinical Trial Registry (UMIN000012494, December 5, 2013: https://upload.umin.ac.jp/cgi-open-bin/ctr/ctr.cgi?function=brows&action=brows&type=summary&recptno=R000014629&language=E). Written informed consent was obtained from each subject after the aim and potential risks of the study were fully explained to each. Subject enrollment in this prospective hypothesis testing study was terminated on December, 2015. Inclusion criteria were adult patients (20 to 80 yr old) undergoing scheduled surgeries under general anesthesia in Chiba University Hospital, Chiba, Japan, and exclusion criteria were patients who are eligible for awake intubation, patients with full stomach, patients with severe comorbidities (American Society of Anesthesiologists physical status, more than 2), and patients with allergies to rocuronium or propofol.

For each anesthetic case, airway assessments were performed in addition to a standard history and physical examination, allowing scoring of 12 independent predictors for both difficult MV and direct tracheal intubation reported by Kheterpal et al.⁴  We modified the score reported by Kheterpal et al.⁴  by setting the thresholds for limited thyromental distance, thick neck, diagnosed sleep apnea, and limited or severely limited jaw protrusion as 60 mm, 40 cm, AHI greater than 5 per hour, and upper lip bite test class 3, respectively.

A preoperative sleep study was performed by a nocturnal portable monitor that measured respiratory airflow and arterial oxygen saturation (SpO₂; SAS2100; Nihon Kohden, Japan). All subjects were instructed to attach an oximetry finger probe and nasal cannula before sleep and to remove them on awakening. After checking the quality of the recordings, respiratory and oximetry variables were calculated using computer software. The severity of SDB was quantified by AHI, the frequency of desaturation more than 3% from the baseline, the percent of time spent at SpO₂ less than 90%, and lowest SpO₂. All subjects were divided into two groups based on the results of the overnight sleep study: subjects with AHI equal to or less than 5 per hour were denoted as the non-SDB group, and subjects with AHI greater than 5 per hour denoted as the SDB group.

The sleep study was performed in 88 subjects whose consents were obtained; however, eight subjects were excluded from the analyses due to failure of TV measurements during anesthesia induction.

Each subject was placed in the supine position and in the neutral head position under cardiorespiratory monitoring (Life Scope J, BSM-9100; Nihon Kohden) in the operating room. In addition to pulse oximetry and an electrocardiogram, respiratory variables such as respiratory flow (accuracy: ± 0.025 l/s), TV (accuracy: ± 10 ml), airway pressure (accuracy: ± 1 cm H₂O; GF-220R Multigas/Flow unit; Nihon Kohden), and carbon dioxide concentration with a mainstream capnometer (CAP-ONE, TG-970P; Nihon Kohden; accuracy: ± 2 mmHg) were continuously monitored. Blood pressure was noninvasively measured every 5 min, and neuromuscular function with acceleromyography (TOF-Watch; Organon Ireland Ltd., Ireland) was assessed by a train of four stimulations every 15 s (no calibration). All these variables were displayed on the cardiorespiratory monitoring screen, and the moving images were captured and stored in a computer for later analyses. Analyses of the respiratory variables were performed by digitized cardiorespiratory monitoring images by an investigator who had no knowledge of the sleep study results. General anesthesia was induced by intravenous administration of 2 μg/kg fentanyl and 1 mg/kg propofol after 3-min inhalation of pure oxygen through an anesthesia circuit. Appropriate anesthesia depth was clinically assessed by the anesthetist without Bispectral Index Score monitoring, and additional propofol was injected if necessary. After confirming the loss of consciousness and starting intermittent train of four stimulations through an ulnar nerve by a 50-mA current, rocuronium 1mg/kg real bodyweight was injected.¹¹ In this one-handed maneuver, the thumb and index fingers hold the anesthesia full-face mask (Air Cushion Face Mask KM202; Koo Medical Equipment Co., Ltd., China), the third and fourth fingers are placed on the left mandibular ramus, and the fifth finger is placed at the left mandibular angle. The anesthetist was instructed to perform his/her best airway opening technique during the anesthesia induction. Twenty-seven anesthesia providers with a wide range of clinical experience of anesthesia management participated in this study without any previous knowledge of the purpose of this study although the results were informed. Twenty seconds after injection of rocuronium, pressure-controlled ventilation by using an anesthesia machine ventilator through the full-face mask (peak inspiratory pressure, 15 cm H₂O; positive end-expiratory pressure, 0; I:E ratio, 0.4; and respiratory rate, 15 cycles per minute) was started and continued for 1 min.

Efficiency of MV with one hand during the initial 15 breaths was assessed by TV achieved by constant mechanical ventilation. We predetermined the breath number needed to initially exceed 5 ml/kg ideal body weight (IBW) of expiratory TV (B-_TV5) as the primary outcome to test the hypothesis and TV of the second breath (TV_2nd), a mean value of the expiratory TV for 1 min (TV_mean), and TV of the 15th breath (TV_15th) as the secondary outcomes in this study. TV was also measured during MV with two hands (TV_2hands) and after either placement of tracheal tube or supraglottic airway (TV_tube; 62 tracheal tubes and 12 supraglottic airways). The efficiency of MV was also assessed by the capnogram waveform. In accordance with 2014 Japanese Society of Anesthesiologists (JSA) airway guideline,¹²  we determined normal capnogram (V1) when all phases of capnogram wave including phase III characterized by a plateau are identified on the waveform, subnormal capnogram (V2) when the phase II waveform characterized by a sole rapid upswing is identified (lack of the phase III), and abnormal ventilation (V3) when no capnogram waveform is identified.

There has been no previous study, to date, that compares the primary outcome (B-_TV5) between SDB and non-SDB subjects. Based on the B-_TV5 (mean ± SD: 5.5 ± 5.3 breaths) obtained in our preliminary study of 10 anesthesia patients, we expected a 2.5 difference between the non-SDB and SDB groups and calculated the samples size as 73 subjects for this study assuming α = 0.05 (two tailed) and β = 0.8 (SigmaPlot 12.0; Systat Software Inc., USA). Accordingly, total sample size was set to be 80. Since the prevalence of SDB patients in Japanese general anesthesia population is unknown, we intentionally recruited participants who were suspected of having SDB to ensure that we would observe similar numbers of SDB and non-SDB patients based on clinical and physical features of SDB such as higher STOP-Bang score,^13,14  obesity, and small maxilla-mandibular structure with excessive submandible soft tissue.^15,16 

For the baseline variables, summary statistics were constructed using frequencies and proportions for categorical data and means and SDs for continuous variables. Patient characteristics were compared using Fisher exact test for categorical outcomes and Student’s t test or Wilcoxon rank sum test for continuous variables, as appropriate. For the primary analysis, we compared the efficiency of MV between groups using an analysis of covariance (ANCOVA) model, taking into account the variation caused by MV effects and using the baseline body mass index (BMI) as covariate. Multiple linear regression analyses were performed with six potential predictors including severity of expiratory flow limitation (EFL) during one-hand MV and five factors (age, BMI, Mallampati class, thyromental distance, upper lip bite test, and AHI), which were selected from 12 previously reported independent risk factors for difficult and impossible MV.^1–3  The definition of SDB is arbitrary and varies among the studies. Our previously determined SDB definition AHI more than 5 includes mild SDB patients and therefore possibly failed to discriminate the influence of SDB on efficacy of MV. As post hoc subgroup analyses, we compared efficacy of MV between 20 non-SDB patients with lower AHI and 20 SDB patients with higher AHI. A value of P < 0.05 was considered statistically significant, and all P values were two sided. All statistical analyses were performed using SAS version 9.4 (SAS Institute, USA).

The study was successfully completed in 80 adult subjects. Neither an oral nor nasal airway was inserted during the MV. None had cardiorespiratory complications such as hypoxemia and hypotension during anesthesia induction. Table 1 represents patient characteristics and results of the sleep study. SDB patients (n = 42) were significantly more obese than non-SDB patients (n = 38) and had more hypoxemic SDB episodes by definition. No SDB patient was preoperatively treated with nasal continuous positive airway pressure.

As indicated in table 2, TV greater than 5 ml/kg IBW (primary outcome) was achieved significantly earlier by one breath with greater TV in non-SDB patients (n = 38) than in SDB patients (n = 42). Other respiratory variables representing efficiency of MV such as TV_2nd, TV_15th, and TV_mean (secondary outcomes) were also significantly better in non-SDB patients than in SDB patients. These statistical differences disappeared when the variables were adjusted by BMI, which significantly differed between the groups among the background patients’ characteristics (table 3). No statistically significant difference of the MV efficiency was observed during two-hand MV.

Subgroup analyses for 20 lower AHI non-SDB patients and 20 higher AHI SDB patients clarified the impact of SDB on the efficiency of MV (tables 2 and 3). The primary outcome (B-_TV5) was significantly earlier by two breaths in non-SDB patients than in SDB patients. The secondary outcomes representing efficiency of MV such as TV_2nd, TV_15th, and TV_mean were also significantly better in non-SDB patients than in SDB patients (table 2). Even after adjustment by BMI, ANCOVA revealed significant differences of variables representing efficiency of MV such as B-_TV5, TV_2nd, and TV_mean between the subgroups (table 3).

Changes of TV during one-hand MV and during two-hand MV are presented for each of the subgroups in figure 1. Smaller TV despite the same ventilator setting in the severe SDB patients than in the non-SDB subjects was clearly illustrated during one-hand MV. The difference between the groups was not eliminated by the two-hand MV (table 2). It is also notable that TV was initially smaller but progressively increased by more than 70% in 1 min during the MV in both groups.

We noted three different flow patterns during MV (fig. 2). After achieving maximal expiratory flow, the expiratory flow gradually decreased, and no difference between inspiratory and expiratory TV was observed in 57 patients (EFL-1: fig. 2A). Normal capnogram waveforms (V1) were observed in the no-EFL breaths. In contrast, expiratory flow abruptly decreased immediately after peak expiratory flow (EFL), and both inspiration and expiration were significantly impaired with V3 capnogram waveform for more than four breaths in nine patients (EFL-3: fig. 2C). In 14 patients, expiratory flow was abruptly decreased but maintained during expiration for more than four breaths. Inspiratory flow was not impaired as noted by the difference between the inspiratory and expiratory TV (EFL-2: fig. 2B). V2 capnogram waveforms were observed in the EFL-1 breaths.

TV_2hands was significantly greater than TV_15th, and patterns of capnogram wave form significantly improved by the two-hand MV compared with one-hand MV (table 2; fig. 1), which suggest the advantage of using two hands for improving efficiency of MV. However, the EFL pattern was not normalized even by using two hands in seven of 23 patients (fig. 2D), while the abnormal flow pattern disappeared in majority of patients (fig. 2, B and C). Interestingly, TV_2hands was significantly greater than the TV_tube, indicating that two-hand airway maneuvers can establish wider upper airway lumen compared to a tracheal tube (table 2). It should be also noted that the TV_tube did not differ between non-SDB and SDB patients, implying similar respiratory compliance between them.

Results of the multiple linear regression analyses with six potential predictors are presented in table 4. Occurrence of EFL was a statistically significant factor increasing B-_TV5, decreasing TV_2nd, TV_15th, and TV_mean, and therefore, explaining the reduction of one-hand MV efficiency. Obesity was a significant factor in the decrease of TV_15th and TV_mean in the model. Notably, both are still the significant factors for explaining reduction of MV efficiency with two hands.

We took this opportunity to assess the relationship between TV and capnogram waveform (V1, V2, and V3) by using data at the fifth breath MV. As illustrated by the box plots in figure 3, TV ranges were overlapped among the subjects with V1 and V2 capnogram waveforms. Notably, minimum TV of patients with V2 capnogram waveform was approximately 2 ml/kg IBW, which is close to the physiologic dead space for adults, and the TV was less than the 2 ml/kg IBW in all subjects with V3 capnogram waveform pattern. Similarly, the minimum TV of subjects with V1 capnogram waveform was approximately 6 ml/kg IBW. End-tidal carbon dioxide concentration was higher in subjects with V1 than those with V2 capnogram waveform, indicating the increasing alveolar-arterial difference of carbon dioxide concentration. Two-hand MV significantly increased the number of subjects with V1 and end-tidal carbon dioxide concentration (table 2).

This is the first study that quantitatively assessed the efficiency of MV during anesthesia induction and explored whether the presence of SDB worsens the efficiency of MV. We found that (1) TV during one-hand MV progressively increased by more than 70% in 1 min, (2) one-hand MV was more difficult in patients with severe SDB than in non-SDB patients, (3) higher BMI and occurrence of EFL independently decreased the efficiency of one-hand MV, (4) two-hand MV maximized the efficiency of MV although limited by obesity and occurrence of EFL, and (5) capnogram waveform patterns well reflected the efficiency of MV.

We first demonstrated that one-hand MV is more difficult at the beginning of the procedure than during the latter period. This well-recognized phenomenon for clinical anesthesiologists is commonly attributed to increasing depth of either anesthesia or muscle relaxant,¹⁷  while no previous study systematically explored its mechanisms. Joffe et al.¹⁸  assessed TV changes during one-hand and two-hand MV during anesthesia induction in their randomized cross-over study. Unlike the results of our study, they found no TV change during one-hand MV but progressive increase of the TV during two-hand MV in anesthetized nonparalyzed patients with oral airway in place. Differences of the results and study designs between the study by Joffe et al.¹⁸  and our studies suggest involvement of mechanisms other than depth of anesthesia and muscle relaxant. In fact, the ratio of train of four responses at the 15th breath was significantly greater in the SDB group than in the non-SDB group (table 2), but the MV efficiency did not differ between the groups when ANCOVA was performed (table 3). Furthermore, in the subgroup analyses, the depth of neuromuscular block did not differ between the subgroups, but efficiency was higher in the non-SDB subgroup. However, the results of this study should be carefully interpreted when rocuronium is given based on ideal body weight. Learning effects,¹⁹  pharyngeal airway hysteresis due to surface adhesive force on the pharyngeal mucosa,²⁰  change of lung volume influence on pharyngeal airway collapsibility,²¹  and opioid-induced rigidity of chest wall and vocal cords²²  would be additional candidates for the mechanisms. Recognition of this time-dependent nature of MV should help anesthesiologists diagnose truly difficult MV during anesthesia induction.

Our SDB patients comprise a wide spectrum of sleep-related breathing abnormalities: mild asymptomatic SDB to severe symptomatic untreated SDB patients. We found that MV during anesthesia induction is more difficult in the subgroup patients with severe SDB than in non-SDB patients although the 5 per hour AHI cutoff for the SDB group used in this study is not yet clear cut. Although the AHI is a marker for SDB severity, AHI thresholds vary and differ depending on the SDB-related symptoms and comorbidities.²³  To our knowledge, no study has determined such an AHI threshold for difficult MV.

Anesthetized paralyzed patients with severe SDB are reported to have higher pharyngeal closing pressures than non-SDB and mild SDB patients.⁶  Furthermore, improvement of pharyngeal collapsibility by various airway interventions is limited by the severity of the preexisting collapsibility of the pharynx.²⁴  In the previous epidemiologic studies that identified SDB as an independent risk factor, SDB was preoperatively documented and treated by nasal continuous positive airway pressure that is usually prescribed for symptomatic moderate to severe SDB patients.^1–4  These physiologic and epidemiologic studies suggest a higher threshold AHI for difficult or impossible MV in agreement with our findings. However, our primary hypothesis was to test the mild SDB criteria (AHI, more than 5), and this study did not test the priorly defined severe SDB criteria. A critical AHI threshold needs to be determined using an appropriate study design for the purpose.

We found that the presence of EFL and obesity independently accounts for reduction of MV efficiency. Safar et al.^25,26  first described the occurrence of expiratory obstruction during mouth-to-nose, not mouth-to-mouth, ventilation particularly in obese anesthetized patients. They speculated the cause to be a valve-like behavior of the soft palate without direct observance of the pharyngeal airway. Despite the absence of air leak from the facemask as confirmed by a stable flow trace, we noticed that the expiratory TV was smaller than the inspiratory TV, in agreement with the observation by Goodwin et al.²⁷  who, however, attributed this to high fresh gas flow used in their study. Our pneumotachograph was placed between the facemask and anesthetic circuit, which prevented the influence of fresh gas flow on the airflow measurement. We consider that greater inspiratory TV over expiratory TV reflects actual gas movement and indicates the increase of lung volume when this phenomenon occurs as Safar²⁶  documented in lung volume increase during the expiratory obstruction.²⁶  In fact, Buffington et al.²⁸  used chest rise but not fall and lack of a carbon dioxide waveform on the capnograph as a definition of expiratory obstruction during anesthesia induction. They identified age and enlarged pharyngeal soft tissue as independent factors for occurrence of the expiratory obstruction. It should be noted that patient characteristics reported by both Safar et al.²⁶  and Buffington et al.²⁸  are common features of SDB patients. Although the expiratory obstruction is believed to be caused by valve-like obstruction at the soft palate, no study has directly proved this and extensively explored how the EFL is caused in association with SDB and/or obesity. Clearly, future studies should focus on this interesting and clinically important phenomenon.

Although obesity has been repeatedly reported to be an independent risk factor for difficult and impossible MV,^1–4  no study in this field has systematically elucidated why obesity leads to difficult MV. We previously demonstrated that mandible advancement is unable to restore the airway patency behind the soft palate in obese subjects⁷  and suggested the importance of MV through the mouth in accordance with the recommendation by Safar et al.²⁶  However, by using only one hand, it is practically difficult to appropriately perform airway maneuvers including mouth opening in obese patients with relatively large facial structures. Furthermore, obesity worsens SDB by increasing pharyngeal airway collapsibility through structural mechanisms such as excessive soft tissue for the craniofacial size and reduced lung volume.^29,30  Our results indicate that the use of two hands can effectively, though not perfectly, compensate for the structural abnormalities resulting in acceptable MV even in severe SDB patients. The use of oral or nasal airways would prevent the EFL at the soft plate and further improve the MV as documented by Safar et al.²⁵ 

Since this study mimicked routine anesthesia induction, we believe that its results provide clinically useful information for predicting, diagnosing, and managing difficult MV. The efficiency of MV decreased only in severe SDB patients indicating the importance of preoperative SDB screening by questionnaires and portable sleep monitor for diagnosis of the SDB severity.³¹  Clinical values of the preoperative SDB diagnosis for decreasing incidence of difficult MV are worth assessing in future studies.

Although prompt diagnosis of difficult MV could maximize risk management strategy, most anesthesia providers rely on their subjective assessments such as thoracic movement and resistance of the anesthetic bag, and Han scale for grading difficulty of MV also includes subjective items.³²  Our results are in agreement with the JSA airway management guideline that suggests the use of three distinct capnogram waveforms to continuously and instantaneously grade ventilation status before development of severe hypoxemia.¹²  Furthermore, airflow tracing, not a TV value, if available, would help to know a possible precursor of difficult MV, i.e., EFL, and indicate time to start two-hand MV or insert an oral airway. It should be also noted that two-hand MV did not normalize the EFL in 30% patients with EFL. Clearly, more effective airway intervention for improving MV needs to be tested or developed in the near future.

Despite these possible clinical implications, this study has several methodological limitations. First, we did not standardize the MV technique, and the anesthesia providers performed their best to achieve adequate MV. Although 27 anesthesia providers participating in this study have a wide range of clinical experience and their own MV techniques, this single-institute study could not cover and test the variety of MV techniques used in other institutes and countries. Second, only eight patients had BMI greater than 30 kg/m² while our results clearly indicated the contribution of obesity to difficult MV and that definition of obesity in Japan is 25 kg/m². Clearly, future studies should address how MV becomes difficult in morbidly obese SDB patients in whom SDB pathogenesis may not be completely identical to that of nonobese SDB patients. Third, we did not objectively assess the depth of anesthesia, which is a significant determinant of the upper airway patency, while the amount of propofol and fentanyl used for induction of anesthesia did not differ between the non-SDB and SDB groups (propofol: 1.0 [1.0 to 1.2] vs. 1.0 [0.98 to 1.05] mg/kg, respectively, P = 0.317; fentanyl: 2.0 [2.0 to 2.1] vs. 2.1 [1.9 to 2.0] μg/kg, respectively, P = 0.106). Lastly, our study design does not allow exploring how muscle paralysis influences efficacy of MV during routine anesthesia induction, while recent clinical studies did not indicate unfavorable effects.^27,33,34 

In conclusion, we first characterized breath-by-breath dynamic changes of ventilation during routine MV in 80 adult patients undergoing general anesthesia and found that (1) one-hand MV is more difficult in patients with severe SDB than non-SDB subjects and (2) obesity and occurrence of EFL are involved in mechanisms of inefficient one-hand MV. The use of two hands is strongly recommended to restore efficient MV.

The authors thank Sara Shimizu, M.D. (Shimizu Orthopedic Plastic Surgery Clinic, Tokyo, Japan), who greatly helped to improve this manuscript.

Supported in part by grant-in-aid (Kakenhi; 4390363) from the Ministry of Education, Culture, Sports, Science and Technology, Tokyo, Japan.

The authors declare no competing interests.