Hypercarbia and high-flow nasal oxygen use during anaesthesia – risking a failure to thrive?

Abstract

Prevention of arterial oxygen desaturation during anaesthesia with high-flow nasal oxygen (HFNO) has gained greater acceptance for a widening range of procedures. However, during HFNO use there remains the potential for development of significant anaesthesia-associated apnoea or hypoventilation and the possibility of hypercarbia, with harmful cardiovascular or neurological sequelae. The aim of this study was to determine whether any HFNO-related hypercarbia adverse incidents had been reported on webAIRS, an online database of adverse anaesthesia-related incidents. Two relevant reports were identified of complications due to marked hypercarbia during HFNO use to maintain oxygenation. In both reports, HFNO and total intravenous anaesthesia were used during endoscopic procedures through the upper airway. In both, the extent of hypoventilation went undetected during HFNO use. An ensuing cardiac arrest was reported in one report, ascribed to acute hypercarbia-induced exacerbation of the patient’s pre-existing pulmonary hypertension. In the other report, hypercarbia led to a prolonged duration of decreased level of consciousness post procedure, requiring ventilatory support. During the search, an additional 11 reports of postoperative hypercarbia-associated sedation were identified, unrelated to HFNO. In these additional reports an extended duration of severe acute hypercarbia led to sedation or loss of consciousness, consistent with the known effects of hypercarbia on consciousness. These 13 reports highlight the potential dangers of unrecognised and untreated hypercarbia, even if adequate oxygenation is maintained.

Introduction

A ventilatory effect termed THRIVE (Transnasal Humidified Rapid-Insufflation Ventilatory Exchange) was described in 2015 and suggested to arise largely from the use of high-flow nasal oxygen (HFNO). ¹ This was reported to slow the progression of hypercarbia, which has limited the duration of anaesthesia-induced apnoea. ¹ Subsequent studies have failed to demonstrate a moderation in hypercarbic rise with HFNO use during apnoea,^2,3 and a finding of oxygen desaturation prevention only in low-risk patients when compared with low-flow nasal oxygen (LFNO). ⁴

HFNO has been extensively reported for use during gastroenterological endoscopy ⁴ and bronchoscopy. ⁵ In these instances, the primary aim of HFNO is to sustain oxygenation while simultaneously allowing ease of proceduralist access through the upper airway with a limitation of ventilatory movements. ⁶ A systematic review published in 2022 of HFNO use in gastroenterological endoscopy found reduced rates of hypoxaemic events (SpO₂ < 92%) and need for simple airway intervention of jaw thrust or increase in oxygen flow compared with LFNO, but only in patients at low-risk of hypoxaemia. ⁴ Patients at high risk of hypoxaemia, such as the morbidly obese, obtain no additional benefit from HFNO.^4,5,7

Prevention of desaturation during procedural sedation by increasing the fraction of inspired O₂ (FiO₂) is a logical approach. Desaturation events are nonetheless reported to be very common. For example, during gastroenterological procedures, episodes of SpO₂ < 92% for greater than 15 s are reported in up to 21% of patients receiving HFNO and 33% receiving LFNO. ⁸ Increasing the FiO₂ brings into play two issues related to hypoventilation detection or complications. The first being that only modest increases in FiO₂ can readily maintain SpO₂ > 90% in the presence of substantiative declines in alveolar ventilation. ⁹ This means that when relying on SpO₂, the presence of an emerging obstruction or medication-induced hypoventilation might not be reflected by a pre-warning fall in SpO₂. A progression to complete obstruction of the airway or drug-induced apnoea, in the absence of oxygen delivery to the alveolus, might instead present as a precipitous fall in SpO₂. ¹⁰ The second relates to the direct relationship between alveolar PCO₂ and PO₂ (PACO₂, PAO₂), wherein as PACO₂ increases, the PAO₂ falls.^9,11 This mathematical relationship can be estimated using the alveolar gas equation, which pertains to a perfect theoretical circumstance at equilibrium without contribution from venous admixture (shunt).^12,13 This allows for the estimated lowering in PAO₂ to be calculated when this results not from insufficient oxygen delivery to the alveolus to meet demands, but rather from a rise in the PACO₂. For example, when breathing room air, an increase in the PaCO₂ to 72 mmHg will reduce the SpO₂ to about 90% (arterial oxygen partial pressure, PaO₂ about 60 mmHg). However, with only a modest increase in FiO₂ to 0.3, as obtainable with LFNO, ¹⁰ the alveolar gas equation calculated theoretical estimate for a breach below SpO₂ 90% now climbs to a potential causative PaCO₂ of 120 mmHg.

Given the above, it is possible procedures conducted with oxygenation maintained by HFNO in the presence of prolonged apnoea or hypoventilation associated with anaesthesia could present as unintended or unrecognised cases of severe hypercarbia. The purpose of this study was to interrogate the webAIRS database for reports of adverse clinical incidents occurring due to hypercarbia when HFNO is utilised during procedural anaesthesia. A secondary aim was to identify other incidents, unrelated to HFNO use, in which hypercarbia had resulted in sedation or unconsciousness in the postoperative period.

Methods

WebAIRS is a voluntary internet-based anaesthesia incident database established in 2009 for anaesthetists in Australia and New Zealand to assist in identifying patient safety issues and guide safer practice. The ethics approval process for the webAIRS data collection methodology has previously been published. ¹⁴ It is briefly repeated here: webAIRS data collection complies with the current ethics requirements for the collection of de-identified quality assurance data in Australia, as outlined by the National Health and Medical Research Council (NHMRC). ¹⁵ In order to ensure data collection meets the NHMRC requirements ethics approval was been sought and obtained from two hospitals in Australia: Royal Brisbane and Women’s Hospital Human Research Ethics Committee (HREC/11/QRBW/311) and Nepean Blue Mountains Local Health District (HREC/12/NEPEAN/18). Ethics approval in New Zealand was obtained from the Health and Disability Ethics Committee, (MEC/09/17/EXP). In addition, reporters at each site had to comply with local institutional approval requirements. The reports are confidential and protected by qualified privilege in both Australia and New Zealand.

A search of the webAIRS database was conducted by the authors (GP, MC, YE) in December 2022. The word search criteria included ‘THRIVE’, ‘HFNO’, ‘Optiflow’, ‘HFNP’, ‘hyperca’ (to match hypercarbia, hypercarbic, hypercapnia and hypercapnic), ‘high CO₂’, or ‘acido’ (to match acidotic or acidosis). The narrative for each report was then assessed by the searching authors. Reports for which there was agreement by all the searching authors of a likely association were included for the purpose of the study.

Results

The webAIRS database yielded 493 reports using the word search criteria. After a review of the report narratives by the researchers and agreement of a likely association, two reports of HFNO use during anaesthesia with the development of hypercarbia and an adverse incident were identified. The details of these reports are summarised in Table 1.

Table 1.

Anaesthesia and high-flow nasal oxygen associated incidents.

Report	Age (years) and sex	Comorbidities	Procedure	Airway	Anaesthesia	Event	ABG	Therapy	Outcome
1	50–59 F	Pulmonary hypertensionCardiac failureChronic kidney disease	Bronchoscopy and oesophagoscopy	HFNO	Propofol TCI Cet 3.3 µg/ml and remifentanil 0.1 µg/kg/min	Pulseless electrical activity	pH 7.1, PaCO2 69 mmHg (after period of resuscitation)	CPR, ETT, IPPV	Return of spontaneous circulation, wakened and extubated
2	40–49 F	Depression	Bronchoscopy	HFNO (10–30 l/min)	Remifentanil 0.15 µg/kg/minProcedure duration 30–40 min	Unresponsive at case completion HypertonusFacial muscle dystonia	PaCO2 elevated above reportable range	IPPV	Restoration of consciousness with normalisation of PaCO2

Note: ABG: arterial blood gas; HFNO: high-flow nasal oxygen; TCI: target-controlled infusion; Cet: effect site targeted concentration; CPR: cardiopulmonary resuscitation; ETT: endotracheal intubation; F: female M: male; IPPV: intermittent positive pressure ventilation.

Case 1 was attributed by the reporter to hypercarbia leading to an exacerbation of pulmonary hypertension and cardiac arrest. The patient, in her fifties with multiple comorbidities, was undergoing bronchoscopy and oesophagoscopy. Preoperative echocardiogram demonstrated impaired right ventricular function and right ventricular systolic pressure (RVSP) of 52 mmHg (normal < 37 mmHg ¹⁶ ). Intravenous infusion of propofol to effect a site target concentration of 3.3 µg/ml and remifentanil at 0.05 µg/kg/min were administered to achieve deep sedation. SpO₂ was maintained at 99% with HFNO (FiO₂ and flow rate not detailed). During the procedure, remifentanil was increased to 0.1 µg/kg/min. Shortly thereafter the systolic blood pressure, as measured non-invasively, fell to 50 mmHg and rapidly progressed to pulseless electrical activity. A successful resuscitation was achieved with immediate endotracheal intubation, positive pressure ventilation, external chest compressions and intravenous adrenaline of 900 µg. The end-tidal CO₂ (EtCO₂) once intubation was achieved was 74 mmHg. A short time later a PaCO₂ of 69 mmHg was recorded while the EtCO₂ was in the ‘low 50 s’. The reporter noted ‘hence PaCO2 would have been close to 90s’, and ‘Hypoventilation was masked by the high flow nasal O₂’. The procedure was abandoned.

Case 2 was of a decreased level of consciousness identified by the reporter as a complication of hypoventilation-induced hypercarbia. The patient, a female in her forties underwent an investigative bronchoscopy. Airway topicalisation with lignocaine and intravenous remifentanil infusion 0.1 µg/kg/min was provided for the procedure. Oxygenation was supported with THRIVE at 10–30 L/min. During the procedure an increase in remifentanil to 0.15 µg/kg/min was required. The report goes on to state, ‘Breathing efforts were noted to be slow. Given THRIVE was used, there was no capnography monitoring, but pulse oximetry confirmed adequate saturations throughout (99–100%)’. Procedure duration was 30–40 min and was followed by failure to wake 20 min after cessation of the remifentanil. The PaCO₂ at this time was above the reportable upper limits for the arterial blood gas machine. The management plan was for tracheal intubation. However, during preparation for intubation the patient showed signs of response to pain and voice. Following this, ‘Respirations were supported by PSV (pressure support ventilation) via facemask and CO₂ was 70 mmHg on a subsequent blood gas’. The patient continued to regain full responsiveness completely over the next few minutes.

There were 11 additional reports identified of decreased level of consciousness postoperatively, during which the patients were confirmed to have hypercarbia. These are summarised in Table 2. Ten of these patients had had successful removal of a formal airway after surgery followed by a decreased level of consciousness in the post anaesthesia care unit. Hypercarbia was detected during assessment or resuscitative intervention. Administration of naloxone was reported in seven patients, with none resulting in an observable response. In the majority, hypercarbia was managed with intubation, an extended period of positive pressure ventilation, and transfer to an intensive care unit. The reports contained statements of a clinical suspicion of hypercarbia-induced sedation, which included ‘good response to overnight ventilation’, ‘hypoventilation and subsequent hypercarbia causing patient to be obtunded’, ‘further hour of assisted ventilation until she awoke’, ‘on arrival in ICU she was starting to wake as CO₂ decreased’, and ‘on low FiO₂ until less drowsy and CO₂ normalised’.

Table 2.

Postoperative hypercarbia-associated sedation incidents.

Report	Age (years) and sex	Comorbidities	Procedure and airway type	Postoperative oxygen supplementation	Recovery event	ABG	Naloxone therapy	Outcome
1	70–79 M	Chronic obstructive pulmonary disease	LaparotomyETT	Hudson mask	Postoperative hypoventilation, drowsiness progressed to loss of consciousness	PaCO2 100	No response to repeated naloxone	ETT, IPPV, ICU
2	90–99 F	Acute kidney injury	LaparotomyETT		Postoperative hypoventilation, drowsiness progressed to loss of consciousness	PaCO2 120	No response to naloxone	ETT, IPPV, ICU
3	60–69 M	Chronic kidney disease	LaparoscopyETT		Postoperative hypoventilation, drowsiness progressed to loss of consciousness	PaCO2 70 pH 7.0	No response to naloxone	ETT, IPPV, extubated successfully once PaCO2 normalised
4	70–79 M		LaparoscopyETT	Non-rebreather mask	Postoperative hypoventilation, drowsiness	PaCO2 ‘elevated’ pH ‘low’	No response to naloxone	ETT, IPPV, ICU
5	70–79 F	–	LaparoscopyETT	–	Postoperative drowsiness	EtCO2 70	–	PPV until consciousness returned
6	70–79 M	Chronic obstructive pulmonary disease	Endobronchial ultrasoundLMA	–	Postoperative drowsiness	PaCO2 98	No response to naloxone	ETT, IPPV, ICU
7	80–89 F	–	LaparoscopyETT	–	Postoperative hypoventilation and drowsiness after opioids	PaCO2 109	No response to naloxone	ETT, IPPV, ICU. Extubated following day once PaCO2 normalised
8	60–69 F	Dementia	ORIF femur fractureSpinal (no airway)	–	Postoperative drowsiness	PaCO2 81	No response to naloxone	Non-invasive ventilation until consciousness returned
9	40–49 M	Morbid obesity Smoker	Hand lacerationETT	–	Delayed emergence	PaCO2 95	–	ICU
10	80–89 M	–	Insertion of pacemakerETT	–	Postoperative drowsiness	PaCO2 90	–	ETT, IPPV, ICU
11	60–69 F	–	LaparotomyETT	Non-rebreather mask	Delayed episode of unconsciousness in recovery	PaCO2 96pH 6.97	–	ETT, IPPV, ICU

Note: ETT: endotracheal intubation; F: female; M: male; LMA: laryngeal mask airway; IPPV: intermittent positive pressure ventilation; ICU: intensive care unit; ‘–’: not stated. ORIF: open reduction and internal fixation. All CO₂ values are in mmHg.

Discussion

The two anaesthesia incidents during HFNO use are a timely reminder to consider the potential for unrecognised hypercarbia and its sequelae during HFNO maintenance of oxygenation in anaesthetised or sedated patients. While we cannot determine the rate of adverse incidents using HFNO by interrogating a voluntary de-identified database, it is possible similar but perhaps less severe cases went unrecognised, or other cases of similar severity went unreported. In any event, the severity of the two incidents is sufficient to alert anaesthetists to the potential for hypercarbia during HFNO when it is associated with hypoventilation or apnoea. The incidents should prompt anaesthetists to consider alternative techniques depending on a patient’s premorbid condition or if the duration of the procedure may be such that the possibility of severe hypercarbia might go undetected.

Given the potential risks of unrecognised severe hypercarbia, it is opportune to review the physiology of apnoeic oxygenation, and the literature in support of its application. High FiO₂ has long been known to prolong oxygenation in apnoea. ¹⁷ This results from a denitrogenating effect enhancing oxygen supply and a drawing effect of the mass movement of oxygen toward the alveoli as O₂ is taken up by pulmonary blood in greater volumes than CO₂ is released into the alveoli.^17,18 The practice of apnoeic oxygenation was described in 1959 when a cuffed endotracheal tube with an open circle circuit flushed with 100% O₂ at 8 L/min maintained SaO₂ 100% for 55 min of apnoea. ¹⁸ Reports of severe hypercarbia and fatal acidosis during apnoeic oxygenation tempered ongoing enthusiasm for its use during anaesthesia.^1,17

The THRIVE study purported the hypercarbia of apnoeic oxygenation could be moderated with HFNO. ¹ A median apnoea of 14 min without desaturation below SpO₂ 90% was reported in 25 patients with the use of humidified HFNO (Optiflow™ 70 l/min, FiO₂ 1.0). The rate of CO₂ rise when measured using EtCO₂ was 1.13 mmHg/min. This contrasted markedly with historical control values in apnoeic oxygenation of 2.6–3.4 mmHg/min – measured mostly as arterial partial pressure CO₂ (PaCO₂).^18–21 An ‘insufflating’ effect of HFNO was proposed and the term THRIVE introduced. In a follow-on study the authors then concluded HFNO uniquely created vortices above the glottis which combined with cardiac oscillations to enable more effective elimination of CO₂ from the alveoli. ²²

In 2017 Gustafsson et al. aimed to recreate the conditions of the THRIVE study. ⁶ Similar to the THRIVE study, they found a median EtCO₂ rise of 0.9 mmHg/min and no desaturations below SpO₂ 90% after 23 min of apnoea. However, when measured in arterial blood, the rise in PaCO₂ at 1.8 mmHg/min was closer to historical values. The mean peak PaCO₂ rise was 77 mmHg and pH nadir 7.2.

Randomised trials have since investigated the benefit of HFNO over alternatives for the rate of CO₂ rise. Two were conducted in a paediatric population. Compared with no supplemental oxygen, HFNO did not alter the rate rise of transcutaneous CO₂ but did expectedly double the apnoeic time. ²³ When HFNO was compared with LFNO, neither a difference in the rate of rise in transcutaneous CO₂ nor in apnoea time was seen. ³

A large randomised controlled trial in adults reported in 2022 further challenged the preferential use of HFNO to either maintain oxygenation or prevent hypercarbia. ² This compared 125 patients with nasal oxygen flow rates of 2, 10 or 70 L/min over 15 min of apnoea. Importantly, two control groups were included: one with an endotracheal tube delivering 0.25 l/min FiO₂ 1.0, and the other HFNO at 70 l/min while airway patency was maintained with a laryngoscope. No significant difference in the rate of rise of PaCO₂ was found between groups, and median values were 2.0–2.1 mmHg/min. Five percent of study participants desaturated below SpO₂ 92% with no significant difference between groups. The findings did not support the concept of HFNO-induced ventilation leading to additional CO₂ elimination. ²⁴ The authors concluded CO₂ elimination was dependent on cardiac oscillations and not on nasal oxygen flow rate.

The cardiovascular effects of hypercarbia are predominantly mediated by sympathetic stimulation. ²⁵ In the pulmonary vasculature, rises in PACO₂ markedly increase pulmonary vascular resistance (PVR) and may be a greater determinant of pulmonary blood flow than oxygen. ²⁶ Echocardiographic studies in healthy volunteers calculated the increase in RVSP as 0.6–1.0 mmHg for every 1 mmHg increase in PACO₂ and being greater incrementally than the increase in RVSP seen with a fall in PAO₂. ²⁶ In a patient with pre-existing pulmonary hypertension, a rise in PVR resulting from hypercarbia can lead to rapid onset of acute right ventricular decompensation and cardiac arrest. ²⁷

Elevations in PaCO₂ can present neurologically as facial twitching, irritability, or gross myoclonic limb movements similar to those in Report 2.^28,29 A sedative (narcotising) effect of hypercarbia has long been espoused. ¹⁰ A decreased level of consciousness only eventuates if hypercarbia results in (respiratory) acidosis. ³⁰ This explains why patients with chronic obstructive pulmonary disease have been recorded with severely elevated PaCO₂ of up to 175 mmHg, but as a result of chronically compensated pH changes maintain full consciousness.^31–34

In 1958, Westlake reported on respiratory failure patients and concluded, ‘in general the critical level for loss of consciousness in carbon dioxide intoxication is at a pH of 7.1–7.14 and PCO₂ of 120–130 mm(Hg)’. ³⁵ Others have stated that abnormalities of mental state appear above PaCO₂ 90 mmHg or when pH falls below 7.25. ³⁶ Experimental research has concluded that the resultant cerebrospinal fluid (CSF) pH is a greater determinate of CO₂ sedation than arterial pH, ³⁰ and the maximal change in CSF pH as PaCO₂ alters can be delayed by up to 30 min. ³⁷ These findings together may have implications in prolonged exposure to hypercarbia where, despite normalisation of PaCO₂ and arterial pH with artificial hyperventilation, there may be ongoing sedation owing to the delay in CSF pH normalising.^28,34,36 The mechanism by which acidosis leads to sedation has not been fully elucidated.^38–40

Thus, it seems prudent to consider benefit and risk when utilising HFNO to support oxygenation during procedural anaesthesia when hypoventilation or apnoea results. The benefit is limiting the frequency in some patients of hypoxaemic events requiring remediation with simple airway manoeuvres. Risk can arise from failing to detect hypoventilation or apnoea and an insidious rise in PaCO₂, leading to significant cardiovascular and neurological sequelae. In this circumstance the use of HFNO can lead to a presentation counter to expectation and be viewed as a ‘failure to thrive’.

The 11 incidents involving postoperative hypercarbia (without HFNO) and decreased level of consciousness are reminders of the interdependence of hypoventilation, hypercarbia, and sedation or loss of consciousness. In many of the cases, opioid-induced ventilatory impairment (OIVI) was suspected by the clinicians and naloxone administered. However, in none did naloxone lead to an improvement. This should not discourage the use of naloxone if OIVI is suspected. However, there may be a subset of cases in which CO₂ narcosis has progressed so far that naloxone alone may be insufficient. The four cases of loss of consciousness are noteworthy for their extreme PaCO₂ (100–120 mmHg) and pH (6.97–7.0) values being largely in the ranges reported by previous authors to be associated with loss of consciousness.^35,36 The remaining seven cases of decreased level of consciousness, similarly recorded PaCO₂ values (70–109 mmHg) in the range previously reported for such an occurrence, or else were potentially capable of inducing the pH values of concern.

In summary, we present 13 incidents, two involving HFNO and 11 involving postoperative narcosis, related to the development of hypercarbia. The two incidents involving HFNO are reminders of the risks of hypercarbia when using this technique, even if adequate oxygenation is maintained. The other 11 are a reminder of the potential severe consequences of postoperative hypercarbia.