High-Dose Oxygen Therapy and Acute Hypercapnia in Elderly Patients: A Case Series Analysis

Introduction

Current British Thoracic Society (BTS) guidelines recommend administration of oxygen via a non-rebreather mask (NRB) at a flow rate of 15 L/min in patients with acute hypoxemia associated with an SpO2 below 85% on room air [1]. Detection of episodes of desaturation of this degree of severity on pulse oximetry is common in elderly patients in community settings. The modern practice of transferring these patients by ambulance to the Emergency Department (ED), has led to a greatly increased exposure to high-dose inhaled oxygen therapy in this group. At a flow rate of 15 L/min the NRB delivers an FiO2 over 60% (average FiO2: 76%) [2]. In the presence of such high-dose (FiO2 > 50%) inhaled oxygen therapy, a number of mechanisms have the potential to generate an immediate increase in the concentration of carbon dioxide dissolved in the arterial blood (the PaCO2) [3,4]. An individual with normal respiratory function will respond to such an increase in the PaCO2 by increasing pulmonary ventilation, eliminating excess carbon dioxide in exhaled air [5]. This brainstem-mediated response prevents the development of persistent hypercapnia during high-dose oxygen administration. Brainstem sensitivity to hypercapnia declines significantly with age [6]. In addition, some elderly patients have undiagnosed conditions placing them at risk of oxygen-induced CO2 retention. However, even in the absence of such a condition, stressed by an acute respiratory pathology, the aging respiratory system may lack the functional reserve necessary to eliminate the increased CO2 load generated by high-dose oxygen therapy. In theory, therefore, clinically significant oxygen-induced hypercapnia can occur in these subjects even in the absence of an underlying disease currently recognized as a risk factor for oxygen-induced CO2 retention. If the treating physician fails to recognize this possibility, the emergence of acute type II respiratory failure in an elderly patient may be interpreted as a pre-terminal event. Initiation of a purely palliative pathway will result in death. In addition, iatrogenic type II respiratory failure places a significant drain on resources in terms of the application of non-invasive ventilation (NIV). Three cases are presented here in which the initial findings on arterial blood gas (ABG) analysis at presentation to the ED and subsequent clinical course where compatible with type II respiratory failure induced by the application of high-dose oxygen therapy in the pre-hospital setting. The possibility of oxygen-induced CO2 retention was not considered at presentation in these cases, with implications for patient management.

Case Reports

CASE 1:

A woman in her late 80s, who was a nursing home resident dependent in all activities of daily living, developed acute shortness of breath associated with wheezing. An ambulance was called. On arrival, the crew confirmed the presence of hypoxemia (SpO₂ 82% on room air) associated with a respiratory rate of 36 breaths per minute. GCS at that time was 14–15. Nebulized salbutamol (5 mg) and sublingual glycerol trinitrate (800 mcg) were administered. Supplemental oxygen therapy was commenced at 15 L/min via NRB. The SpO₂ rose to 98% over the course of 20 min. Oxygen at this dose was continued during the 30 min taken to reach the local ED. The patient experienced 1 episode of vomiting while in the ambulance.

A chest X-ray after presentation to the ED showed atelectasis at the left lung base (Figure 1A). White cell count was elevated at 21.9×109/l (normal range: 3.5–11.0×109/l). Urea and electrolytes were normal and serum albumin was low at 29 g/L (normal range 35–50 g/L). The patient was treated for both aspiration pneumonia and pulmonary edema with intravenous piperacillin/tazobactam and intravenous frusemide.

Initially, oxygen was continued in the resuscitation room at 15 L/min via NRB. A venous blood gas (VBG) taken 1 h 15 min after commencing high-dose oxygen therapy demonstrated abnormalities consistent with acute respiratory acidosis (Figure 2, VBG1). This was confirmed on ABG analysis (Figure 2, ABG1). The patient was not expected to survive the acute episode and a decision was made not to progress to mechanical ventilation in the event of further deterioration. In the hours following presentation, attempts to correct the patient’s hypoxia by application of an FiO2 of 70%, initially on continuous positive airway pressure (CPAP) and subsequently on bi-level positive airway pressure (BPAP), was associated with worsening hypercapnia and a declining GCS (Figure 2). However, a falling oxygen requirement overnight allowed progressive reduction in the FiO2 delivered on BPAP (Figure 2). Hypercapnia resolved when the FiO2 delivered to the patient was at 40% or below. In the absence of hypercapnia, the patient was fully alert (GCS 15-15).

Reviewing the patient 17 hours after her arrival in ED, based on the relatively rapid recovery from type II respiratory failure on a combination of BPAP and controlled oxygen therapy, the admitting medical team considered the possibility that the initial use of high-dose oxygen therapy had generated hypercapnia and CO2 narcosis. The patient had never smoked and had no history of respiratory or neurological disease. Her husband had been a life-long heavy smoker. During an admission 1 year previously for treatment of lower-limb cellulitis, she had experienced an episode of acute desaturation. An ABG taken at that time on oxygen (5 L/min via nasal prongs) showed hyperoxemia but no evidence of acute or chronic CO2 retention. The admitting team concluded that, despite the history of passive smoking, there was little to suggest the presence of undiagnosed COPD. Despite the absence of a risk factor for CO2 retention, a target oxygen saturation range was set at 88–92%. The patient quickly transitioned from low-dose inhaled oxygen therapy via nasal prongs to maintaining her target SpO2 on room air. In the days following admission, she experienced intermittent pyrexia and episodes of desaturation. One such episode was accompanied by evidence of acute pulmonary edema on chest X-ray (Figure 1B). Of note, at one point she became unrousable when a nebulizer was administered on 100% oxygen. Seven days after admission, she was discharged back to her nursing home with appropriate palliative measures in place. She died peacefully 2 days following discharge.

CASE 2:

A woman in her 90s, living at home but dependent on family and carers for all activities of daily living, experienced a distressing persistent choking episode. An ambulance was called. On arrival, the crew confirmed the presence of hypoxemia (SpO2 61% on room air) associated with a respiratory rate of 40 breaths per minute. At the time of arrival, the crew recorded a GCS of 13–15 (patient baseline: 14–15). The crew administered intravenous frusemide (40 mg) and nebulized salbutamol (5 mg). Supplemental oxygen therapy was started at 15 L/min via NRB. The SpO2 rose to 87% over the course of 25 min. Oxygen at this dose was administered for 1 h prior to arrival at the local ED. On arrival in the ED, the SpO2 was 76% on the NRB and the GCS had fallen to 9–15. Initial arterial blood gas analysis showed hypoxemia and significant acute hypercapnia (Figure 3, ABG1). She was not expected to survive the acute episode, and a decision was made not to progress to mechanical ventilation in the event of further deterioration. The patient was started on BPAP at an initial FiO2 of 60%, subsequently increased to 70%. This was associated with persistent hypercapnia and low GCS (Figure 3). In the hours after admission, a decreasing oxygen need allowed reduction of the FiO2. At an FiO2 of 35%, hypercapnia was absent (Figure 3, ABG2, ABG3). A chest X-ray at presentation showed volume loss in the right and left lung fields, but no acute abnormality (Figure 1C). White cell count was elevated at 16.2×109/l (normal range: 3.5–11.0×109/l). ECG showed sinus tachycardia. Urea and electrolytes were normal and serum albumin was low at 33 g/L (normal range 35–50 g/L). A diagnosis of aspiration pneumonia was made and the patient was treated with intravenous amoxicillin/clavulanic acid.

The patient had been diagnosed with Parkinson’s disease 11 years prior to the events described here and had a history of epilepsy. However, the episode of choking had been witnessed and no seizure activity had been noted. She had never smoked and had no history of chronic obstructive pulmonary disease (COPD). There was no history of passive smoking. Despite the absence of a recognized risk factor for oxygen-induced CO2 retention, the admitting medical team considered it possible that CO2 retention induced by high-dose oxygen therapy had contributed to the type II respiratory failure at presentation. BPAP was discontinued, and oxygen therapy was titrated to a target oxygen saturation range of 90–94%. Hypercapnia was absent on controlled oxygen therapy and the GCS returned to baseline (GCS 14–15). Despite manifesting worsening pneumonic changes on her chest X-ray in the 24 h following admission (Figure 1D), hypercapnia remained absent on low-dose oxygen therapy (Figure 3, ABG2, ABG3). A second reduction in GCS (10–15) on day 2 after admission was not associated with hypercapnia or hypoxemia on ABG analysis (Figure 3). This low GCS preceded palliative sedation and no seizure activity was noted. Seventeen days after admission, with appropriate palliative measures in place, the patient died peacefully on the ward in the presence of her family.

CASE 3:

A woman in her 90s, living at home with a home-care package in place, had been unwell for 2 weeks with a persistent productive cough. She developed sudden onset of shortness of breath at rest. An ambulance was called. The ambulance crew confirmed the presence of hypoxemia with an SpO2 of 76% on room air. GCS was 15-15. The crew documented an audible generalized wheeze on auscultation of the chest. The respiratory effort was described as ‘moderately distressed.’ Nebulized salbutamol was administered. Intolerant of a face mask, oxygen was initially started at 6 L/min via nasal prongs. The SpO2 rose to 87%. The patient remained on this oxygen dose during the 15-min journey to the local ED. On arrival in the ED, the GCS was 15-15. An ABG on oxygen 5 L/min via nasal prongs demonstrated type II respiratory failure (Figure 4, ABG1). A trial of BPAP was considered appropriate. After 1 h on BPAP at an increased FiO2 of 55%, a repeat ABG showed a significant rise in the PaCO2 compared to the admission ABG (Figure 4, ABG2). The FiO2 was further raised to 60%, and on this oxygen dose the patient experienced a sustained decrease in her GCS overnight (Figure 4). No ABGs were taken at this oxygen dose. Guided by a rising SpO2, the FiO2 on BPAP was reduced. By morning, at an FiO2 of 35%, the GCS had risen to near baseline (GCS 14-15).

The cause of the patient’s initial hypoxemia was not clear. A chest X-ray on arrival showed no acute abnormality (Figure 1E). A large hiatus hernia in the left hemithorax and bibasilar atelectasis were longstanding X-ray findings, present on earlier films. White cell count was normal at 6.4×109/l (normal range: 3.5–11.0×109/l). Urea and electrolytes were normal and serum albumin was low, 26 g/l on admission, falling to 19 g/L 4 days after admission (normal range 35–50 g/l). NT-proBNP was elevated at 1481 pg/ml (normal <450 pg/ml). ECG showed no evidence of cardiac ischemia. Intravenous frusemide, a course of intravenous piperacillin/tazobactam, and regular salbutamol nebulizers were administered.

The admitting medical team considered the possibility that the acute type II respiratory failure observed at presentation had resulted from CO2 retention secondary to high-dose oxygen therapy. The patient had never smoked and had no history of passive smoking. There was no history of respiratory or neurological disease. Despite the absence of a risk factor for oxygen-induced CO2 retention, a target oxygen saturation range was set at 88–92%. On low-dose oxygen therapy, the patient remained fully alert and comfortable. Hypercapnia, in association with metabolic alkalosis, persisted on low-dose oxygen therapy 8 days after admission (Figure 4, ABG5). At that time, the patient and the admitting doctor agreed that any further investigation or intervention would be inappropriate. One month after admission, she had completed rehabilitation on the ward and was discharged home with a care package in place. Six months following discharge, she remained well.

Discussion

In all 3 cases presented here an assumption was made that the emergence of type II respiratory failure in the acute phase of the patient’s illness reflected ventilatory failure due to progression of a severe underlying acute lung pathology. The emergence of type II respiratory failure was considered a pre-terminal event. Key decisions with regard to patient management were made based on this interpretation of events. The alternative possibility that high-dose oxygen therapy had induced hypercapnia was not considered at presentation in these patients, all 3 of whom lacked a recognized risk factor for oxygen-induced CO2 retention. While it is possible that the application of BPAP reversed hypercapnia in Case 1 and Case 2, it is also possible that the initiation of controlled oxygen therapy while on this modality, removed the stimulus generating hypercapnia and that this was responsible for the relatively rapid and clinically unexpected resolution of type II respiratory failure in these cases. It is also of interest to note that in Case 2, a deterioration in chest X-ray findings after admission (Figure 1D) was not accompanied by type II respiratory failure at a time when oxygen therapy was controlled.

While type II respiratory failure may emerge secondary to progression of common acute conditions such as pulmonary edema or pneumonia [7], it is important to be aware of the possibility of oxygen-induced CO2 retention in a patient deteriorating on high-dose oxygen therapy. If a case of oxygen-induced hypercapnia is missed, and that patient is placed on a purely palliative pathway, BPAP may be withheld and oxygen withdrawn. The latter intervention will result in a precipitous fall in the PaO2 to a level lower than that present before oxygen was administered. This increased hypoxemia results from persistence of a high level of CO2 in the alveoli (PACO2) for up to 40 min after inhaled oxygen is stopped [8]. The resulting sustained degree of rebound hypoxemia can be fatal. Also, the resource implications of oxygen-induced hypercapnia, in terms of the use of NIV, are highly significant [9].

The 3 elderly patients described in this report had no documented history of a chronic condition known to be a risk factor for oxygen-induced CO2 retention. The association between COPD and oxygen-induced or aggravated hypercapnia is well known [10] and undiagnosed COPD is common in adult populations [11]. However, all 3 patients in this series were lifelong non-smokers. While Case 1 did have a history of significant passive smoking, there were no features in the patient’s medical history or previous hospital investigations to suggest the presence of underlying undiagnosed COPD. Parkinson’s disease is common in individuals aged over 85 years of age and is associated with impaired respiratory function [12,13]. Case 2 suffered from advanced Parkinson’s disease. In theory, the disorder could have placed the patient at risk of CO2 retention in response to high-dose oxygen therapy. In all 3 cases, the minor elevation of arterial bicarbonate concentration above the normal range at presentation suggests that hypercapnia was acute. However, a degree of chronic CO2 retention cannot be definitively excluded. In practice, neither a history of passive smoking nor Parkinson’s disease influences the current approach to acute oxygen therapy in the pre-hospital setting. Furthermore, with no access to ABG analysis, ambulance crews have no way of detecting unsuspected chronic CO2 retention.

It is possible that the physiology of aging places individuals of very advanced age at risk of CO2 retention triggered by high-dose oxygen therapy, even in the absence of a chronic disease predisposing to this phenomenon. Hypoxic pulmonary vasoconstriction (HPV) is a physiological mechanism diverting blood from poorly ventilated areas in the lungs, minimizing ventilation–perfusion mismatch (Figure 5). In the very elderly, coalescence of airspaces (‘senile emphysema’) gives rise to extensive, poorly ventilated areas in the lung parenchyma [14]. Therefore, it is possible that very elderly individuals are particularly dependent on HPV to maintain relatively normal pulmonary gas exchange. When an individual inhales supplemental oxygen, elevation of the alveolar partial pressure of oxygen (the PAO2), eliminates HPV in poorly ventilated areas. An increased ventilation–perfusion mismatch results in an increase in the arterial partial pressure of carbon dioxide (PaCO2) (Figure 5). Calculated using the alveolar gas equation, in Case 2, administration of oxygen via the NRB will have generated a PAO2 of 405 mmHg (Figure 3, ABG1). On low-dose oxygen (FiO2: 24%) the PAO2 will have fallen to 117 mmHg.

In a young individual, brainstem respiratory centers will detect the elevation of the PaCO2 resulting from the mechanism described above, triggering a rise in the pulmonary ventilation [5]. Excess CO2 is eliminated in exhaled air and hypercapnia is transient (Figure 5). Compared to normal young individuals, elderly subjects (aged >64 years) have a significantly reduced brainstem response to hypercapnia [6]. It is possible that age-dependent loss of the brainstem response to a rising PaCO2, is compounded by the effects of undiagnosed CNS pathologies on brainstem respiratory centers in very old people. However, even if capable of detecting a rise in the PaCO2, an aged respiratory system, already taxed by an acute severe lung insult, may be incapable of further elevating the alveolar ventilation to eliminate the increased CO2 load associated with the application of high-dose oxygen therapy. Age-related degenerative changes in thoracic anatomy were prominent in Case 2 and Case 3 in this series (Figure 1). Skeletal muscle weakness associated with advanced age may have further reduced respiratory reserve in all 3 patients.

In addition to loss of HPV, several additional mechanisms can contribute to the emergence of hypercapnia during high-dose oxygen therapy delivered by the NRB. In the frail elderly patients presented here, hypercapnia may have been compounded by an increase in dead space associated with the face mask of the NRB. Also, in some patients, this modality will deliver an FiO2 above 80%, a level sufficient to induce significant absorption atelectasis of alveoli [15]. Both Case 1 and Case 2 spent approximately 1 h on the NRB prior to initial blood gas analysis. Experiments in patients on mechanical ventilators indicate that absorption atelectasis manifests within 30 min of inhalation of high-dose oxygen [16]. Shunting associated with absorption atelectasis may have further aggravated hypercapnia in Case 1 and Case 2. In Case 1 and Case 3, high-dose oxygen therapy elevated the SaO2 on ABG analysis into the normal reference range (Figures 2, 4). In these cases, hypercapnia will have been compounded by the Haldane effect [3,4], which depends on elevation of PaO2. When hemoglobin binds oxygen in the alveolar capillaries, the conformation of the molecule changes and significant quantities of carbon dioxide bound by the molecule in the tissues are released into the bloodstream. At the initial level of hemoglobin desaturation present in Case 1 and Case 3, raising the hemoglobin saturation to the normal range would be expected to elevate the PaCO2 by approximately 5–7 mmHg due to the Haldane effect [17]. In contrast, in Case 2, with little change in the oxygen saturation of hemoglobin in response to oxygen administration, the Haldane effect would not have contributed to the emergence of hypercapnia seen on the initial ABG.

The physiological response to high-dose inhaled oxygen in the very elderly has not been studied. In 2004, Ting reported the case of an 80-year-old woman presenting to the ED with chest pain. She developed CO2 narcosis with loss of consciousness shortly after commencing inhaled oxygen therapy (6 L/min via nasal prongs) [18]. She had never smoked. However, unlike the cases presented in this report, she did have a history of chronic respiratory disease, specifically, late-onset asthma, and had evidence of chronic CO2 retention on her initial ABG. In her case, a significant increase in the FiO2 delivered in the early phase of presentation resulted in a paradoxical fall in the PaO2 associated with a rise in PaCO2. This phenomenon was also observed in Case 3 in this series (Figure 4, ABG1, ABG2). Also, it should be noted that the magnitude of the rise in arterial bicarbonate concentration in Case 3 following admission was is in excess of that expected purely on the basis of renal compensation occurring in response to an acute respiratory acidosis (Figure 4, ABG5). The patient had, in fact, developed a metabolic alkalosis. The etiology of this metabolic alkalosis was not determined. However, the metabolic disturbance may have been responsible for the persistence of hypercapnia. Type II respiratory failure (associated with hypoxemia) maintained by an underlying metabolic alkalosis has been described previously [19].

An elderly patient, confused and agitated due to delirium or CO2 narcosis, may be intolerant of BPAP. Having ‘failed a trial of BPAP’, they may then be quickly moved on to an end-of life management protocol. However, if the clinical scenario proposed in this case series is recognized, oxygen delivered by high-flow nasal cannula (HFNC) at a controlled oxygen dose may be a useful alternative to NIV. By ‘washing out’ dead-space carbon dioxide, HFNC may help reduce the PaCO2 [20]. Ideally, the clinical scenario described here should be avoided. Although generally considered to be of therapeutic value, there is no evidence that oxygen therapy in acute hypoxemia improves prognosis, and there is evidence that oxygen therapy in excess is harmful [21]. Humans are relatively tolerant of acute hypoxemia [22]; therefore, it seems reasonable to advocate a less aggressive approach to oxygen supplementation in the pre-hospital setting, particularly in the elderly.

In the conscious self-ventilating hypoxemic patient, presenting with an SpO2 below 85%, it may be appropriate to start oxygen therapy at a low or moderate dose and to titrate the dose upwards if necessary. In my opinion, we rarely need to exceed an SpO2 of 92% on supplemental oxygen therapy, and this could prevent a significant number of iatrogenic problems which, at the present time, go largely unrecognized in the ED.

Conclusions

The possibility of oxygen-induced CO2 retention should be considered in all elderly patients developing acute type II respiratory failure while on high-dose oxygen therapy. If oxygen-induced CO2 retention is suspected, inhaled oxygen therapy must not be withdrawn abruptly.

A rapid, often described as a ‘Lazarus-like’ recovery from type II respiratory failure following a period on BPAP, should raise the possibility that oxygen-induced hypercapnia contributed to the development of type II respiratory failure.

Key events in patient management occur in the pre-hospital phase. Ambulance notes should be carefully analyzed by the receiving ED and admitting medical teams.