Pulmonary Function after Emergence on 100% Oxygen in Patients with Chronic Obstructive Pulmonary Disease: A Randomized, Controlled Trial

ACCORDING to the World Health Organization, chronic obstructive pulmonary disease (COPD) reflects a disease characterized by obstruction of airflow which interferes with normal breathing and is not fully reversible. In 2007, COPD was the fifth leading cause of death worldwide but will become the third leading cause by 2030.¹  Patients with COPD often present with increased blood flow to lung units with subnormal ventilation—perfusion ratio.^2,3  Loeckinger et al.⁴  documented the underlying modifications in pulmonary gas exchange: With oxygen breathing during emergence, pulmonary blood flow is redistributed to lung units with a subnormal ventilation–perfusion ratio. The consequence of oxygen breathing on postanesthetic pulmonary function in COPD patients remains uncertain and may involve changes in ventilation–perfusion ratios.

On insertion and removal of artificial airways, the patients’ lungs are routinely ventilated with 100% oxygen. This is done to widen the safety margin toward hypoxemia in case airway problems occur. The gain in safety comes at a price: Oxygen breathing may cause alveolar instability and atelectasis formation. Hedenstierna et al.⁵  demonstrated atelectasis formation after induction. Benoît et al.⁶  showed that oxygen breathing during emergence from anesthesia equally results in atelectasis. Amazingly, there is no evidence base for oxygen breathing,⁷  and despite long-standing controversies,^8,9  this routine remains a confirmed habit. Oxygen is also used in patients with COPD particularly because these frequently present with impaired oxygenation. Little is known about the pulmonary gas exchange after emergence on oxygen in COPD patients; however, a deterioration of pulmonary gas exchange is likely.

For examining this theory, we hypothesized that in COPD patients, emergence on 100% oxygen leads to lower arterial oxygen pressures compared with the arterial oxygen pressures with emergence on 30% of oxygen.

In this study, we examined COPD patients after general anesthesia. All patients were treated with 30% of oxygen at induction and during anesthesia. One group was treated with 100% of oxygen during emergence, and the other group remained on 30% of oxygen in this period. After the anesthetic, all patients were continued on room air so that the only difference between groups was the oxygen concentration during emergence from anesthesia. This is of course not the usual practice but was performed in this trial under supervision and with special attention. Pulse oxymetry was monitored continuously and arterial blood gas (ABG) analyses were performed before and 15 and 60 min after anesthesia.

This single-site study was performed from May 2007 to May 2008 at the hospital of the Medical University in Innsbruck, Austria. The type of the study is two-group, parallel arm. An overview of the study protocol is outlined in figure 1. Taken together, patients listed for elective carotid endarterectomy with general anesthesia were asked to participate in this study. Participants were thus recruited at the preanesthetic visit. After the visit on the day before surgery, informed consent was obtained and spirometry was performed on potential participants. Those with spirometry results positive for COPD (see Spirometry) were then further enrolled for pre- and postoperative ABG analyses and randomly allocated (see Statistical Analysis) to either emergence on 100% oxygen (100%, n = 26) or to emergence on 30% oxygen (30%, n = 27). The duration of emergence was measured. After extubation, patients were breathing room air. The study was approved by the local ethical committee and conducted in accordance to their guidelines. The ethics committee also served as a data-safety monitoring board (Innsbruck, Tirol, Austria).

After informed consent was obtained, potential participants were given 400 μg of salbutamol by inhalation. Fifteen minutes later, patients completed spirometry using a hand-held spirometer (Micro Spirometer; CareFusion, Basingstoke, United Kingdom). The best of three attempts was recorded. Parameters obtained included the forced expiratory volume in 1 s (FEV1, liters per second), the forced vital capacity (FVC, Liters), and the FEV1/FVC ratio calculated by the device (FEV1/FVC). Percent-of-predicted value of FEV1 (FEV1%) was calculated thereafter.

Chronic obstructive pulmonary disease was defined as a FEV1/FVC ratio of less than 0.7 and an FEV1 of 80% or less 15 min after inhalation of 400 μg of salbutamol.¹⁰ 

General anesthesia with intubation was performed. Positive end-expiratory pressure (PEEP) was set 5 cm H₂O. Patients enrolled in this study received no premedication. In terms of drugs, anesthesia was performed in a total intravenous manner using fentanyl (0.005 mg/kg), propofol (2.5 mg/kg), and rocuronium (0.6 mg/kg) for induction and continuous remifentanil (0.25 μg kg⁻¹ min⁻¹) and propofol (50 to 100 μg kg⁻¹ min⁻¹) for maintenance. Additional dosing of the listed drugs or higher infusion rates was left at the discretion of the anesthetist. Before intubation, lungs were ventilated using 30% oxygen balanced with nitrogen. This gas mixture was then used until emergence where—depending on group allocation—patients were kept on 30% oxygen or switched to 100% oxygen. Duration of emergence (from stopping the propofol infusion to extubation) was recorded. Before extubation, relaxometry was performed. A train-of-four ratio of greater than 0.9 was accepted for extubation otherwise neostigmine (25 μg/kg) and atropine (0.01 mg/kg) were given.¹¹ 

Arterial blood gas samples were taken using an indwelling radial arterial catheter before induction and at 5, 15, and 60 min after extubation and immediately analyzed using a Chiron RapidLab 860 (Chiron, Fernwald, Germany). The 15 and 60 min measurements were taken in the postanesthesia care unit (fig. 1). Samples were immediately analyzed for Pao₂, Paco₂, pH, fraction of oxyhemoglobin, bicarbonate, and lactate. Other variables recorded included heart rate and blood pressure.

Aado₂ (alveolar–arterial oxygen partial pressure difference) was calculated using the common alveolar gas equation:

where PA reflects the alveolar pressure of oxygen, Fio₂ is the fraction of the inspiratory oxygen, Patm is the atmospheric pressure, PH₂O is water vapor pressure, Paco₂ is the arterial partial pressure of carbon dioxide, and 0.8 is the respiratory quotient.

This trial was registered in a national database. The analytical framework was to identify superiority when using 30% oxygen instead of 100% during emergence. The estimation of the sample size was based on the comparison of arterial oxygen tension because Aado₂ is difficult to attain. The calculation was based on a two-treatment, parallel design. Assuming that the alternative hypothesis is a reduction in arterial oxygen tension of 8%, we a priori estimated that we needed to enrol 25 patients in each group (totalling 50) to reach a power of 80% with a type I error rate of 0.05 if the true difference between treatments is 0.8 times the standard derivation.

The nature of the randomization scheme was blocking. Randomization was performed by using Research Randomizer,* a long-standing on-line tool. Randomization to the 30% or the 100% group (1:1 ratio) was prepared for 60 patients before the study using opaque envelopes labeled with the sequential number of the patient enrolled and containing the allocation information (100 or 30%). Randomization and envelope preparation were performed by a colleague who was not involved in perioperative care or in data acquisition.

Analyses were performed using the Statistical Package for the Social Sciences (SPSS 15.0; IBM, Armonk, NY). Ahead of the main statistical analysis, results were tested for normal distribution using Pearson chi-square test. A two-way repeated-measures ANOVA was then used to detect inter- and intragroup differences. The hypothesis was tested two-tailed. Significant results were post hoc examined using the Newman–Keuls test. Primary outcome was a difference in Aado₂ at the 60 min measurement. Secondary outcomes included Pao₂, Paco₂, pH, fraction of oxyhemoglobin, bicarbonate and lactate, heart rate, and blood pressure. Results are presented as mean ± SD of the mean. P values less than 0.05 were rendered significant.

The presented data sets are complete. This study involved hypoxemia to a certain degree; however, all participants tolerated the study well. Data are displayed in tables 1 and 2 and figure 2. In this study, we examined 53 patients with COPD listed for elective carotid endarterectomy. The patients we examined were in their sixties, had almost normal weight, and all had a smoking history greater than 50 pack-years. Spirometric findings in the examined cohort revealed COPD at Global Initiative for Obstructive Lung Disease stage 2.¹⁰  Our main findings were that Pao₂ was lower after general anesthesia and this drop was more pronounced in those who had oxygen during emergence from anesthesia.

A total of 76 patients were screened. Fifty-three of 76 recruited patients had an FEV1/FVC of less than 0.7 together with an FEV1 of less than 80% of predicted and were further enrolled and examined for pre- and postoperative ABGs. Patients allocated to the 100% (n = 26) and to the 30% groups (n = 27) had similar anthropomorphic characteristics and spirometric findings (table 1).

Physiological data and ABG results are given in table 2. Aado₂ is presented in figure 2. Emergence on 100% oxygen leads to decrease in Pao₂ (table 1) based on an increase in Aado₂ when compared with emergence on 30% (intergroup) and when compared with baseline (intragroup) at 15 and 60 min after emergence from anesthesia. Paco₂ was comparable between groups at all measurements. Other parameters including time-to-extubation and circulatory variables were not different between groups.

All demographic data and all results were normally distributed; hence, a parametric test (ANOVA) was adequate. When testing for a correlation between the drop amplitude in Aado₂ and FEV1% (reduction from 100%), we could not detect a dependence (Pearson correlation = 0).

In the ANOVA results, group effect had a P value of 0.02, time effect and group × time interaction had a P value of less than 0.001. A point estimate was calculated for the difference between the two conditions at the 60-min measurement, where the mean difference was 5.32 (95% CI, 1.05 to 9.98).

This investigation showed that in COPD patients subjected to surgery for carotid artery stenosis, general anesthesia impaired postoperative oxygenation in all patients. Those who received 100% oxygen during emergence showed a statistically significant increase in Aado₂ and hence lower Pao₂ as compared with patients receiving 30% oxygen during emergence.

The findings by Gunnarsson et al.¹²  using the multiple inert gas technique showed that COPD patients develop an increase in pulmonary blood flow heterogeneity—but no increase in shunt. This means that our observed changes in Aado₂ are interpretable as changes in heterogeneity rather than increases in shunt. The relative inability of COPD patients to develop shunt has been first noted by Wagner et al.²  Previous work from our group where emergence on oxygen has been examined in anesthetized animals also indicates blood flow heterogeneity as the leading cause of decreases in Pao₂; however, these have been healthy animals.⁴ 

In the observed setting, increases in Aado₂ suggest changes in diffusion, ventilation–perfusion ratio, or in shunt. Shunt, however, has been shown to be of little importance in patients with COPD.^2,12  In the absence of shunt and with no changes in diffusion, Aado₂ reflects the heterogeneity of pulmonary blood flow. An increase in Aado₂ as in our results indicates increased ventilation–perfusion heterogeneity or—put differently—an intensification of the pre-existing pathology.

With oxygen breathing in normal subjects, a shunt usually develops within 30 min and is also readily reversible.¹³  The mechanism behind is alveolar denitrogenation. In lung units, where oxygen uptake into the blood exceeds the delivery of fresh gas, atelectasis occurs.¹³ 

In terms of anesthesia, oxygen is not the sole source of atelectasis formation: supine posture, reduction in the functional residual capacity, and airway closure all contribute to the formation of atelectasis.^5,14 

In COPD patients, lungs may show clinically significant pulmonary blood flow to lung units with normal and low ventilation–perfusion ratios.⁸  In awake oxygen breathing, COPD patients rarely develop intrapulmonary right-to-left shunt, which might be associated with collateral ventilation, where alveoli ventilate other alveoli distal of an airway obstruction.² 

Some of the observed changes during oxygen breathing may also be attributed to the hypoxic pulmonary vasoconstriction (HPV). HPV is a vascular reflex, where airway hypoxia causes pulmonary arteries to constrict, diverting blood flow from poorly to better-oxygenated lung units.¹⁵  Although HPV is of little importance in the healthy lung (because there are no poorly ventilated units), it is critical for maintaining a balanced distribution of ventilation and perfusion in the lungs of COPD patients.^16,17  Oxygen breathing obviously dampens the HPV as nifedipine does.¹⁷  In the classic experiments performed by Mélot et al., the difference in Pao₂ with a vigorous HPV compared with a dampened HPV response is approximately 20 mmHg—roughly the same as in our comparison between 100 and 30% oxygen during emergence from anesthesia.

From the alveolus, oxygen moves into the pulmonary capillary blood. This uptake is usually completed in a quarter of a second. Alveolar nitrogen does not readily move into the blood because there is only a negligible partial pressure gradient and because it has a low solubility in blood. Nitrogen helps to keep an alveolus open. Also, our perioperative treatment helps to keep the alveoli open: PEEP is an airway and alveolar pressure above the atmospheric pressure at the end of expiration. PEEP is considered to keep alveoli open.

When oxygen breathing (partially) washes the nitrogen out of the alveoli, the alveolus may become increasingly critical and eventually collapses.¹³  This occurs when all nitrogen is washed out and all oxygen is taken up by the blood.

When oxygen breathing is applied during emergence from anesthesia, ventilated alveoli first have their nitrogen washed out and then after extubation also have the PEEP taken away. In contrast to anesthetic induction, this sequence is now inverted leaving oxygen-filled alveoli suddenly unpressurized and thus prone to become partially emptied by the blood flow. Regional ventilation will be impaired as alveoli become critical and ventilation–perfusion ratios become lower. So in this concept, extubation on oxygen may particularly help to develop lung units with subnormal ventilation–perfusion ratios. The latter partly explains the observed increases in Aado₂.

In this experiment, we examined a widely practiced maneuver that generates pulmonary blood flow heterogeneity in a cohort of patients already presenting with such a heterogeneity. Our patients presented with Pao₂ values of approximately 80 mmHg which is at the lower end of the normal range (80 to 105 mmHg). After extubation on either 100 or 30% oxygen, we observed a consistent difference in Pao₂. The observed values (table 2) are all relatively low, and given that a Pao₂ of 60 mmHg is generally considered as an indication for intubation, this issue deserves attention, because oxygen breathing leads to Pao₂ values of approximately 60 mmHg. Furthermore, these Pao₂ values are on the steep section of the oxygen dissociation curve, where a small decrease in oxygen partial pressure results in a large drop in saturation.

The decrease in Pao₂ between preanesthetic values and those measured at 15 and 60 min after extubation is very large in both groups (P = 0.0001; table 2) and reflects why we need to be cautious with general anesthesia in COPD patients.

In a survey performed in the United Kingdom and Ireland in 2005, 54% of the anesthetists claim to always use pure oxygen during emergence, 32% say, they mostly use it and 10% state, they occasionally use oxygen on extubation.¹⁸  And recently, Mackintosh et al.¹⁹  published that high intraoperative concentrations of oxygen do not alter the postoperative oxygen requirements in those with healthy lungs. Emergence on oxygen is still a common practice and it seems to be harmful for only those with a pre-existing pulmonary problem. A recommendation or guideline for those presenting with COPD for general anesthesia might help to establish an adapted and adequate treatment.

When we give thought to postoperative care of a COPD patient with a low Pao₂, we come across the nurse in the postanesthesia care unit. As in the observed cohort, such low Pao₂ values indicate that supplemental oxygen is required. It is uncertain how supplemental oxygen affects pulmonary function of the already compromised COPD lung. Research may thus also be directed on the postoperative care of the COPD patient after general anesthesia. Noninvasive ventilation via face mask may be a strategy to keep the airways open, compensate for the work of breathing and avoid dynamic hyperinflation. Positive end-inspiratory pressure could be set to values at the patient’s intrinsic PEEP,²⁰  and delivering low oxygen concentrations may help to avoid further redistributions in pulmonary blood flow.

In our study, we examined a homogenous cohort, with individuals similar in terms of demographics, spirometry results, and treatment received. Postoperative oxygenation deficits are thus attributable to the only difference between the two groups: oxygen breathing during emergence.

Some limitations should also be mentioned: First, the anesthetist was not blinded in terms of group allocation; however, this is impossible due to patient safety. Second, we examined predominantly men, however, fewer women with a smoking history present for carotid endarterectomy. Another limitation is our failure to perform measurements later than 60 min after extubation. This is in part due to the fact that we did not expect such long-term changes. Last, it has to be noted that induction at 30% of oxygen is uncommon, and some may even consider it dangerous. However, having done proper airway assessment (i.e., anesthetic history and Mallampati score), we were able to induce anesthesia with 30% safely. It also needs to be mentioned that omitting supplemental oxygen after general anesthesia is not the usual practice in the European Union.

After emergence on oxygen sustained alterations in pulmonary gas exchange could be observed in patients with COPD. Emergence on 30% oxygen also caused gas exchange disorders in these patients, but to a lesser extent. The mechanism can be attributed to an increase in pulmonary blood flow heterogeneity because COPD patients show a relative inability to develop shunt. The routine use of 100% oxygen during emergence might need reconsideration in the treatment of COPD patients.

Support was provided solely from institutional and/or departmental sources.

The authors declare no competing interests.