Secondary Logo

Journal Logo

Cardiovascular anaesthesia

Does permissive hypoxaemia during extracorporeal membrane oxygenation cause long-term neurological impairment?

A study in patients with H1N1-induced severe respiratory failure

Holzgraefe, Bernhard; Andersson, Christin; Kalzén, Håkan; von Bahr, Viktor; Mosskin, Mikael; Larsson, Elna-Marie; Palmér, Kenneth; Frenckner, Björn; Larsson, Anders

Author Information
European Journal of Anaesthesiology: February 2017 - Volume 34 - Issue 2 - p 98-103
doi: 10.1097/EJA.0000000000000544
  • Free



Gilbert-Kawai et al.1 reviewed 1651 papers in a Cochrane collaboration study without finding any evidence that hypoxaemia or hyperoxaemia increases morbidity and mortality in patients with respiratory failure. However, it is quite clear that hyperoxaemia is not beneficial in most circumstances and has been found to be associated with an increased mortality in critically ill patients.2–4 However, hypoxaemia has traditionally been considered to cause tissue hypoxia and increase the risk of multiple organ failure as well as cognitive dysfunction after the event.5,6 The main indication for extracorporeal membrane oxygenation (ECMO) in adult patients is hypoxaemia that cannot be managed by other means. These patients have usually suffered a period with substantial hypoxaemia before ECMO is initiated and, indeed, might be hypoxaemic even during ECMO.

The Extracorporeal Life Support Organisation (ELSO) guidelines accept blood oxygen saturation (SaO2) levels of 80% and mixed venous oxygen saturation (SvO2) levels of 70%.7 These SaO2 levels are much lower than those commonly used in critically ill patients and there are no investigations that have explicitly studied long-term neurological outcome after this treatment strategy. There is only one study in an unmixed ECMO cohort that has carefully investigated the cognitive outcome in adult patients, and in that study, the saturation levels achieved were not reported.8 Moreover, only 11 of the 28 patients in that study had acute respiratory failure as the indication for ECMO treatment. It was found in the overall cohort that 41% of the patients had cognitive sequelae.

In our unit, we sometimes use permissive hypoxaemia, that is we accept even lower SaO2 values, if needed, than those accepted in the ELSO guidelines to avoid using other advanced ECMO methods with their inherent risks or increasing airway pressures. In this context, it is important to consider that the only proven therapy in acute respiratory distress syndrome (ARDS) that has been shown to reduce mortality is lung protection, that is reducing tidal volumes, driving pressures and end-inspiratory airway pressures.9 However, in addition, we always aim to keep adequate tissue oxygen delivery by maintaining a sufficient arterial oxygen content by compensating for low saturations with increasing haemoglobin (Hb) concentrations (above 100 g l−1) and cardiac output titrated high enough to keep mixed venous preoxygenator saturation at least 70% and blood lactate concentration of 2 mmol l−1 or less. Although blood lactate concentration is not a marker of the adequacy of cerebral oxygen delivery, it is a marker of the adequacy of systemic oxygen delivery.10,11 We assume that, in hypoxaemic situations when blood lactate concentration starts to increase, it is mainly caused by hypoxia in less vital organs than the brain, as due to the hierarchy of organ oxygen delivery, the brain should theoretically be one of the last organs to suffer.12 Our hypothesis is that this treatment algorithm will give sufficient cerebral oxygen delivery during ECMO to prevent long-term cognitive dysfunction.

The aim of this study was to examine the long-term neurological outcome by neuropsychological tests and MRI in the survivors of the H1N1 pandemic of 2009/2010 who were treated in our ECMO unit. This patient category was chosen for three reasons: the short-term outcome of this group in whom we have employed permissive hypoxaemia has previously been reported by us; we have been unable to find any studies of long-term neurological outcome in ECMO-treated H1N1 patients; and because neurological sequelae from infections with H1N1 are extremely rare, it would be very unlikely that any neurological deficits that might be found were caused by the underlying condition.13–15

Materials and methods

Ethical approval for this study (DNR 2012/986-31/3) was provided by the local Research Ethics Committee in Stockholm (Regionala etikprövningsnämnden i Stockholm), FE 289, 171 77 Stockholm, Sweden (Chairperson Eva Lindeblad) on 20 June 2012. The patients who agreed to participate in this study gave their written informed consent.

All patients (n = 13) were treated during the H1N1 pandemic at the ECMO department of the Karolinska University Hospital in Stockholm, Sweden. This department is an ICU with six beds that is dedicated to ECMO treatment for newborn, paediatric and adult patients. We tried to contact all patients who were still alive 3 years after ECMO (n = 11).

For all studied patients, we checked that they fulfilled the Berlin criteria of severe respiratory failure/ARDS before ECMO and of permissive hypoxaemia during the ECMO treatment by examining oxygenation at admission and during the first 10 days on ECMO.16 If the patient was decannulated from ECMO earlier, the whole treatment period was analysed.

The definition of ‘clinically relevant hypoxaemia’ is unclear.5 During the ECMO treatment, we adhered to published definitions of hypoxaemia, that is a peripherally measured oxygen saturation of less than 94%.17,18 Although preoxygenator blood gas analysis was performed, arterial blood gases were not routinely measured during ECMO. For estimation of the arterial oxygen content (CaO2), we calculated the paO2 (mmHg) according to the formula published by Severinghaus19 in 1979:

Ln paO2 = 0.385 ln (S−1 − 1)−1 + 3.32 − (72 S)−1 − 0.17 (S6), where S is the peripherally measured oxygen saturation.

Blood lactate concentrations from preoxygenator blood gas analysis during the same period were used as a surrogate of organ malperfusion and concomitant cellular hypoxia. Peripherally measured oxygen saturation from the right hand, ear or nose was monitored continuously during the whole treatment and was registered, if possible, hourly. The oxygenation index before ECMO was calculated as: FIO2 x 100 x mean airway pressure/paO2.20

Neuropsychological examinations

A trained neuropsychologist administered the tests. Cognitive functioning was assessed by the Wechsler Adult Intelligence Scale, 4th ed. (WAIS-IV), using the average full-scale intelligence quotient (FSIQ) and General Ability Index (GAI).21 The FSIQ is based on four index scores representing major components of intelligence (verbal comprehension index, VCI; perceptual reasoning index, PRI; working memory index, WMI and processing speed, PSI). The GAI is based on VCI and PRI, and reduces emphasis on working memory and processing speed relative to the FSIQ.

Memory functioning (memory functioning index) was assessed using the average test results in the Rey Auditory Verbal Learning Test (RAVLT; learning trials 1 to 5 and delayed recall), immediate recall from Rey Complex Figure and Logical Memory I and II from Wechsler Memory Scale, third edition.22–24 FSIQ, GAI and the MI were established for each patient by comparing the individual scores with age-matched healthy populations.21–24

Native Swedish speakers were assessed using reference values from a Swedish age-matched population, while patients assessed in English were compared with a native English-speaking age-matched population. The FSIQ score was excluded in one patient due to lack of formal education, as literacy and a basic level of formal education are required for reliable FSIQ calculations.25 Individual data for FSIQ, GAI and MI are presented using the IQ standard scale (Mean = 100, SD = 15).


A Philips Ingenia 3 Tesla system (Phillips North America Corporation, Andover, MA, USA) was used with sagittal and axial T1- and T2-weighted images, and coronal T2-weighted and T2-weighted fluid attenuated inversion recovery (FLAIR) images. The slice thickness was 4 mm. All patients were awake during the examinations. Two senior specialists in neuroradiology independently assessed the images.

Statistical analysis

Numbers are expressed as mean ± SD or mean (95% confidence intervals). All calculations were done with SPSS (SPSS version 22 for Macintosh, Chicago, Illinois, USA).


Thirteen patients survived ECMO and 11 were still alive 3 years after ECMO. Three out of the 11 survivors were living abroad and eight in Sweden. We were able to contact one of the three patients living abroad and six patients in Sweden. These seven (mean age 31 years) agreed to participate in this study.

All patients were hypoxaemic and five patients also had respiratory or metabolic acidosis at initiation of the ECMO treatment (Table 1). All patients had low oxygen saturations compared with traditional goals in severe ARDS but sufficient oxygen content and distribution during ECMO treatment (Table 2). Mean CaO2 during the study period was 14.3 ± 1.9 ml dl−1, Hb 126.0 ± 8.5 g l−1, venous pH 7.37 ± 0.03 and body temperature 37.2 ± 0.4°C during the observation period. SpaO2 in all seven patients was registered 1191 times during the first 10 days.

Table 1
Table 1:
Ventilatory and oxygenation parameters before ECMO
Table 2
Table 2:
Mean (±SD) values of oxygenation parameters, tissue perfusion surrogates and blood transfusions during the observation period

Global cognitive functioning (FSIQ and GAI) was at or above the average of 50% of the reference population (FSIQ 85-115) in five study patients and below this average in one patient (Table 3). Memory functioning was normal in all seven study patients when compared with age-matched healthy controls. Patient #2 could not be tested with FSIQ due to a lack of formal education, which is necessary for FSIQ (Table 3). The four index scores that are the components of FSIQ are also displayed.

Table 3
Table 3:
Cognitive and memory function, mean values (95% CI)

MRI did not show any pathological changes consistent with previous hypoxaemia in any patient. However, in four patients, there were other abnormalities; one patient had a small pituitary, one patient had an arachnoid cyst and two patients had small old cerebral or cerebellar infarctions, and in one of these patients, the lesions were known before the ECMO episode.


This is the first study reporting the long-term neurological outcome of permissive hypoxaemia in ECMO-treated patients and, it is also the first to report the long-term neurological outcome in patients treated with ECMO during the 2009/2010 H1N1 pandemic. We found that cognitive function, including memory, was within two standard deviations of the mean FSIQ in a healthy reference population despite the fact that all patients were hypoxaemic before ECMO and subjected to permissive hypoxaemia during the first 10 days of their treatment or the entire treatment if shorter.

In order to detect cognitive sequelae, we used WAIS-IV, a well established and advanced measure of cognitive ability in adults.21 The WAIS-IV provides measures of four cognitive domains (verbal comprehension, perceptual reasoning, working memory and processing speed) as well as a FSIQ based on the results from the four domains. The GAI reduces emphasis on working memory and processing speed relative to the FSIQ. Memory function assessed learning and retention of visual and verbal stimuli using standardised methods.22 Cognitive dysfunction after intensive care treatment is known as a major health problem in short and long-term outcome.26 However, although some studies have found a correlation between hypoxaemia and cognitive dysfunction in patients with severe refractory respiratory failure, it is not fully clear whether hypoxaemia per se is a contributing cause. Tissue oxygenation depends on many factors such as dissolved oxygen, Hb concentration, cardiac output, organ perfusion, pH and body temperature. Most important is the diffusion gradient from blood into the cell, which is not only dependent on paO2 measured in arterial blood gas analysis but on the paO2 in the capillaries. Recently, Mikkelsen et al.6 found that in ARDS, a mean paO2 of 9.2 kPa (72 mmHg) compared with a mean paO2 of 11.6 kPa was significantly associated with cognitive dysfunction. Interestingly, arterial oxygen saturation was above 94% in both groups, indicating that paCO2 was lower or pH was higher in the group with lower paO2, according to the Hb oxygen saturation-oxygen tension relationship. Indeed, hypocapnia and alkalosis reduce cerebral perfusion and might have contributed to the results presented by Mikkelsen et al.6 Unfortunately, Mikkelsen et al.6 did not report whether any of the patients were subjected to ECMO treatment.

Previous laboratory studies have shown that CaO2 and not paO2 is the most important factor in cerebral oxygenation and perfusion.27 In our study, mean CaO2 was kept at 14.3 ± 1.9 ml dl−1 by increasing Hb concentration using erythrocyte transfusions (see Table 2). One potential drawback with this approach is the risk of transfusion-related complications. However, in a recent study28 in septic patients without hypoxaemia comparing two transfusion policies resulting in two levels of Hb concentration, there were no differences in complications between the groups. In another study29 in patients undergoing cardiac surgery, likewise without hypoxaemia, mortality rate was even decreased with higher Hb concentration, and thus a larger number of transfusions. In agreement with a comment by Vincent regarding transfusions in critical illness, we consider that blood transfusion and Hb concentrations need to be individualised also during ECMO treatment depending on the patient's underlying condition, particularly regarding oxygen consumption and oxygen transport ability.30,31 Cognitive dysfunction could be attributed to many different causes in critically ill patients, for example sepsis and inflammation, acute liver and kidney dysfunction, and drugs as well as concomitant conditions, for example old age, atherosclerosis, dementia, alcohol abuse and psychological distress, and therefore, it is not usually possible to identify one single factor, such as hypoxaemia, as the main contributing cause. Furthermore, it is unlikely that irreversible hypoxic cerebral dysfunction or damage occur when hypoxaemia is present with preserved sufficient circulation.32,33 Cerebral hypoxia is usually due to cerebral ischaemia, that is insufficient cerebral blood flow.32,34 In adult intensive care patients, this is often caused by cardiac failure or ischaemic stroke. Therefore, it is our strong opinion that permissive hypoxaemia should not be used in patients with known or suspected cerebral or cardiac ischaemic diseases.

The ELSO accepts a SaO2 down to 80%, and in this study, even much lower SaO2 values were observed.7 However, it is important to note that neither ELSO nor ourselves propose permissive hypoxaemia as a goal, because in the ELSO recommendations, mixed venous oxygen saturation should be more than 70%, and in our protocol, the hypoxaemia is compensated for by increased perfusion and Hb concentration. In all seven patients in our study, both preoxygenator SvO2 and lactate concentrations indicated general tissue normoxia. Importantly, the global results from WAIS IV (FSIQ and GAI) as well as MI in this study showed normal cognitive functioning in all patients 3 years after ECMO. In three patients, one or more WAIS index scores were below the average. This is not uncommon in the general population in which most people have at least one index score that deviates from their average performance.35 Even motivation when performing the tests is an important factor and can influence the results.

Risnes et al.8 published the only previous study in an unmixed ECMO group, investigating the possible correlation of ECMO treatment and long-term cognitive dysfunction. In their study, 11 of 28 patients were treated with ECMO for severe refractory respiratory failure. In all, 41% of patients in their study showed neuropsychological impairment. In addition, intracerebral lesions were associated with cognitive decline. In our study, four patients had intracranial disease. These findings were, however, not related to previous hypoxaemia, as independently assessed by two experienced neuroradiologists. Furthermore, none of the patients with intracerebral lesions had, in contrast to the study by Risnes et al.,8 impaired cognition at follow-up. Nevertheless, intracerebral lesions are not uncommon in ECMO-treated patients and may depend, for example, on the cannulation technique; they are more frequent in veno-arterial ECMO when central vessels or the subclavian artery are used for returning the blood into the patient.8 Cerebral lesions could therefore be a consequence of the attempt to improve oxygenation by changing from veno-venous to veno-arterial ECMO in patients without restored lung during veno-venous ECMO.

Our study has some limitations. First, the study is an observational case series in a small population and without a control group, and it is therefore difficult to draw general conclusions. However, the number of patients in the previous study of long-term cognitive follow-up of a similar patient category is only 11, and we believe that our study contributes much to the knowledge of cognitive outcome in adult patients after ECMO. In addition, there is no other study that has examined long-term cognitive outcome, either in patients treated with permissive hypoxaemia during ECMO or in patients treated with ECMO due to H1N1 infection. Second, we could not assess cognitive function before the ECMO treatment; if the participants had significantly higher FSIQ than normal before they became ill, they could be suffering from cognitive impairment but still be within the normal range of the age-matched healthy population. Third, we did not measure cerebral oxygenation but used preoxygenator saturation and blood lactate concentrations as surrogates for assessing organ perfusion. The median venous oxygen saturation was normal or supranormal, but in two patients, it was higher than SaO2, which might indicate a significant amount of recirculation in the extracorporeal circuit. However, blood lactate concentrations were normal in these patients.


To our knowledge, this is the first study reporting the long-term neurological outcome of permissive hypoxaemia in ECMO-treated patients and also the first to report the long-term neurological outcome in patients treated with ECMO during the 2009/2010 H1N1 pandemic. We found that cognitive function was within the normal range in seven study patients despite the fact that all patients were hypoxaemic before ECMO and subjected to permissive hypoxaemia during the first 10 days of their treatment, or their whole treatment if shorter. The results of this small retrospective study indicate that the ELSO recommendation that SaO2 down to 80% during ECMO treatment might be reasonable if adequate organ perfusion is preserved, regarding long-term cognitive outcome. Because of the small sample size of our study, general conclusions for patients with severe ARDS and ECMO treatment cannot be drawn and further studies are therefore needed.

Acknowledgements relating to this article

Assistance with the study: we would like to thank Dr Peter Radell for his help with language editing.

Financial support and sponsorship: BH received departmental funding; AL received a grant from the Swedish Heart and Lung foundation and the Swedish Research Council.

Conflicts of interest: none.

Presentation: preliminary data for this study were presented as a poster at the European Society of Anaesthesiology (ESA) Euroanaesthesia Congress, 31 May to 3 June 2014, Stockholm, Sweden.


1. Gilbert-Kawai ET, Mitchell K, Martin D, et al. Permissive hypoxaemia versus normoxaemia for mechanically ventilated critically ill patients. Cochrane Database Syst Rev 2041 2014; 5:CD009931.
2. Hafner S, Beloncle F, Koch A, et al. Hyperoxia in intensive care, emergency, and peri-operative medicine: Dr Jekyll or Mr. Hyde? A 2015 update. Ann Intensive Care 2015; 5:42.
3. Davies A, Jones D, Bailey M, et al. Australia and New Zealand Extracorporeal Membrane Oxygenation (ANZ ECMO) Influenza Investigators. Extracorporeal membrane oxygenation for 2009 influenza A (H1N1) acute respiratory distress syndrome. JAMA 2009; 302:1888–1895.
4. Martin DS, Grocott MP III. Br J Anaesth 2013; 111:867–871.
5. Pannu SR, Moreno Franco P, Li G, et al. Development and validation of severe hypoxemia associated risk prediction model in 1 000 mechanically ventilated patients*. Crit Care Med 2015; 43:308–317.
6. Mikkelsen ME, Christie JD, Lanken PN, et al. The adult respiratory distress syndrome cognitive outcomes study: long-term neuropsychological function in survivors of acute lung injury. Am J Respir Crit Care Med 2012; 185:1307–1315.
7. ELSO Guidelines. Patient care practice guidelines.
8. Risnes I, Wagner K, Nome T, et al. Cerebral outcome in adult patients treated with extracorporeal membrane oxygenation. Ann Thorac Surg 2006; 81:1401–1406.
9. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342:1301–1308.
10. Rixen D, Raum M, Holzgraefe B, et al. A pig hemorrhagic shock model: oxygen debt and metabolic acidemia as indicators of severity. Shock 2001; 16:239–244.
11. Lee YK, Hwang SY, Shin TG, et al. Prognostic value of lactate and central venous oxygen saturation after early resuscitation in sepsis patients. PLoS One 2016; 11:e0153305.
12. Boston US, Slater JM, Orszulak TA, Cook DJ. Hierarchy of regional oxygen delivery during cardiopulmonary bypass. Ann Thorac Surg 2001; 71:260–264.
13. Holzgraefe B. Extracorporeal membrane oxygenation for pandemic H1N1 2009 respiratory failure. Minerva Anestesiol 2010; 76:1043–1051.
14. Wilking AN, Elliott E, Garcia MN, et al. Central nervous system manifestations in pediatric patients with influenza a h1n1 infection during the 2009 pandemic. Pediatr Neurol 2014; 51:370–376.
15. Okumura A, Nakagawa S, Kawashima H, et al. Severe form of encephalopathy associated with 2009 pandemic influenza A (H1N1) in Japan. J Clin Virol 2013; 56:25–30.
16. Ranieri VM, Rubenfeld GD, Thompson BT, et al. ARDS Definition Task Force. Acute respiratory distress syndrome: the Berlin definition. JAMA 2012; 307:2526–2533.
17. Martin DS, Grocott MPW. Oxygen therapy in critical illness: precise control of arterial oxygenation and permissive hypoxemia. Crit Care Med 2013; 41:423–432.
18. Kratz A, Lewandrowski KB. Case records of the Massachusetts General Hospital. Weekly clinicopathological exercises. Normal reference laboratory values. N Engl J Med 1998; 339:1063–1072.
19. Severinghaus JW. Simple, accurate equations for human blood O2 dissociation computations. J Appl Physiol 1979; 46:599–602.
20. Ortiz RM, Cilley RE, Bartlett RH. Extracorporeal membrane oxygenation in pediatric respiratory failure. Pediatr Clin North Am 1987; 34:39–46.
21. Wechsler D. Wechsler Adult Intelligence Scale-Fourth Edition (WAIS-IV): administration and scoring manual. San Antonio, TX: Psychological Corporation; 2008.
22. Schmidt M. Rey Auditory Verbal Learning Test: a handbook. Los Angeles, CA: Psychological Corporation; 1996.
23. Meyers JE, Meyers KR. Rey complex figure test and recogntition trial (RCFT), professional manuel. Odessa, TX: Psychological Assessment Resoruces; 1995.
24. Wechsler D. Wechsler memory scale-Third edition manual. San Antonio, TX: The Psychological Corporation; 1997.
25. Marks DF. IQ variations across time, race, and nationality: an artifact of differences in literacy skills. Psychol Rep 2010; 106:643–664.
26. Jones C, Griffiths RD, Slater T, et al. Significant cognitive dysfunction in nondelirious patients identified during and persisting following critical illness. Intensive Care Med 2006; 32:923–926.
27. Hoiland RL, Bain AR, Rieger MG, et al. Hypoxemia, oxygen content, and the regulation of cerebral blood flow. Am J Physiol Regul Integr Comp Physiol 2016; 310:R398–R413.
28. Murphy GJ, Pike K, Rogers CA, et al. Liberal or restrictive transfusion after cardiac surgery. N Engl J Med 2015; 372:997–1008.
29. Holst LB, Haase N, Wetterslev J, et al. Lower versus higher hemoglobin threshold for transfusion in septic shock. N Engl J Med 2014; 371:1381–1391.
30. Vincent J-L. Which carries the biggest risk: anaemia or blood transfusion? Transfus Clin Biol 2015; 22:148–150.
31. Hovaguimian F, Myles PS. Restrictive versus liberal transfusion strategy in the perioperative and acute care settings: a context-specific systematic review and meta-analysis of randomized controlled trials. Anesthesiology 2016; 125:46–61.
32. Biagas K. Hypoxic-ischemic brain injury: advancements in the understanding of mechanisms and potential avenues for therapy. Curr Opin Pediatr 1999; 11:223–228.
33. Grocott MPW, Martin DS, Levett DZH, et al. Arterial blood gases and oxygen content in climbers on Mount Everest. N Engl J Med 2009; 360:140–149.
34. Takahashi S, Higano S, Ishii K, et al. Hypoxic brain damage: cortical laminar necrosis and delayed changes in white matter at sequential MR imaging. Radiology 1993; 189:449–456.
35. Grégoire J, Coalson DL, Jianjun Zhu. Analysis of WAIS-IV index score scatter using significant deviation from the mean index score. Assessment 2011; 18:168–177.
© 2017 European Society of Anaesthesiology