Hypoxemia is defined as low oxygen levels in the blood (Bhutta et al., 2022). It can occur in obstructive sleep apnea (OSA), in various pulmonary diseases such as COPD and COVID-19, in neuromuscular disorders like Guillain-Barré syndrome, Pompe’s disease or myasthenia gravis, and in central nervous system conditions like Alzheimer’s (Sivak et al., 1999; Orešič et al., 2011; Branson and Faarc, 2018; Chokesuwattanaskul et al., 2021; Rahman et al., 2021). Hypoxemia may also develop at high altitudes, where less oxygen is available due to the low atmospheric pressure (Mulroney et al., 2015).

Hypoxia research is divided into three arbitrary exposure designs: chronic, intermittent and acute hypoxia. Research on chronic hypoxia has shown the ability of the human body to adapt to prolonged exposure to low oxygen levels or pressures (Gilbert-Kawai et al., 2014). Hence, individuals residing at high altitudes or those who have had recent altitude exposure are frequently ineligible to participate in clinical trials investigating the impacts of hypoxia. In recent years, intermittent hypoxia (IH) has been studied, attempting to use the human body’s adaptability to IH as a therapeutic effect for various diseases (Navarrete-Opazo and Mitchell, 2014). This review will focus only on acute non-intermittent hypoxia.

Studying acute hypoxia is of particular importance in assessing the safety of occupational or recreational settings that involve acute exposure to high altitude, e.g., pilots, military personnel, mountaineers, etc. Hypoxia facilities can induce either hypobaric hypoxia by reducing barometric pressure or normobaric hypoxia by diluting oxygen fraction with nitrogen (DeHart and Davis, 2002). There has long been debate as to whether there is a difference in physiological response to isolated changes in barometric pressure, with normal levels of oxygen and carbon dioxide (Naughton et al., 2012). More recent studies suggest that there is indeed a significant impact of pressure alone (Savourey et al., 2003; Savourey et al., 2007; Millet et al., 2012; Coppel et al., 2015; McMorris et al., 2017).

One of the first symptoms of hypoxia exposure is impaired cognitive functioning. The level of oxygen saturation is considered to be the key predictor in determining the extend of cognitive impairment caused by hypoxia (McMorris et al., 2017). During high-altitude activities, cognitive performance plays a critical role in tasks that require attention, decision-making and memorization of protocols, as it can mitigate the risk of potential disasters. Over the years, numerous cognitive tests have been developed to evaluate cognitive performance. However, due to the wide range of cognitive tests used in studies examining cognitive performance during acute hypoxia exposure, it is challenging to compare results across different studies.

A useful test for the cognitive effects of acute hypoxia can be characterized as any response measure that shows a clear, consistent response to meaningful hypoxia levels, across studies from a sufficient number of different research groups. A hypoxia level-response relationship and a plausible relationship between the function test and the physiological response to hypoxia are indications that the test reflects physiological activity.

Previously, similar criteria were used to evaluate the usefulness of central nervous system (CNS)-tests for the effects of antipsychotic drugs (De Visser et al., 2001), benzodiazepines (De Visser et al., 2003), selective serotonin re-uptake inhibitors (SSRIs) (Dumont et al., 2005), 3,4-methyleendioxymethamfetamine (MDMA) (Dumont and Verkes, 2006), D9-tetrahydrocannabinol (THC) (Zuurman et al., 2009) and alcohol (Zoethout et al., 2011) in healthy subjects. In general, these systematic reviews showed that only a small number of tests actually display proper characteristics for a meaningful effect biomarker, and that these tests differ between drug classes (Van Gerven et al., 2019).

This review aims to provide an overview of the wide range of cognitive tests used to assess cognitive performance during acute hypoxia in healthy volunteers, to evaluate the differences in test sensitivity to different levels of hypoxia, to explore the role of barometric pressure, and to ultimately identify which tests best meets the criteria to serve as a meaningful test of the functional effects of hypoxia.

A large number of tests were used in the literature to measure the acute CNS effects of hypoxia in healthy adults. As with similar reviews for drug classes (De Visser et al., 2001; De Visser et al., 2003; Dumont et al., 2005; Dumont and Verkes, 2006; Zuurman et al., 2009; Zoethout et al., 2011), there were more tests than studies: 86 in 38 studies. Only six individual tests, including their variants, (7%) were used five times or more and consequently provided sufficient data points for our individual test analysis. This wide variety of tests limits cross-study comparison. This limitation not only hampers the current review, but a virtually unrestricted use of a large number of distinct neurocognitive tests also thwarts the field of hypoxia research as a whole. Although it may be useful to study different distinct areas of cognitive performance, interpretations and insights would strongly benefit from standardization within cognitive domains and test clusters. Moreover, standardization of exposure protocols would also improve comparisons between studies. Given this suboptimal context, we grouped tests into test clusters and functional domains, to obtain more insights into potentially meaningful hypoxia CNS-effects and protocols. Prior reviews demonstrated that this approach is useful for assessing function tests or biomarkers for drug effects (De Visser et al., 2001; De Visser et al., 2003; Dumont et al., 2005; Dumont and Verkes, 2006; Zuurman et al., 2009; Zoethout et al., 2011). While this methodology inevitably results in the loss of some information, it ultimately yields a structured and comprehensive overview of the chance of measuring significant CNS effects of acute hypoxia in studies with healthy adults.

Attention is an important functional domain that underlies most of the other CNS-performance tasks. Hypoxia has an effect on attention, particularly, at levels below 80%, which caused significant abnormalities of divided or sustained attention in 36% and 47% of studies, respectively. This effect of hypoxia on more demanding (particularly, divided) attention tests may also have influenced other complex or multifunctional tasks, particularly those in the executive domain. These clusters represent aspects of cognition with a direct effect on safety for military personnel and pilots. Especially, impairments in sustained attention, which were already frequently observed at an oxygen saturation between 89% and 80% (60%), could lead to serious safety risks when military personnel have to stand guard at altitude (2438 m–4572 m) on hostile territory. Therefore, this aspect of cognition should be carefully considered during training and preparation for a mission or flight. The effect of hypoxia on reaction time seems limited, even under more severe hypoxic conditions (SpO₂ < 80%) a low-test sensitivity is observed (33%). However, it should be noted that in some cognitive tests, secondary effects on reaction time were observed, which were not included in the analysis to prevent bias.

In the executive domain, the different test clusters rarely showed an effect of mild hypoxia (SpO₂ ≥ 80%). However, when oxygen saturation fell below 80%, the test sensitivity increased. The level-response relationship between oxygen saturation and test sensitivity could indicate that cognitive testing of the executive domain gives the most accurate representation of the physiological response of hypoxia. Although the mechanism causing the impairment of executive function by hypoxia is not clearly understood, this could be related to impaired concentration or (divided) attention, which are required during performance of any executive task. It is also possible that hypoxia-induced executive impairment is related more directly to decreased neural activity of the prefrontal cortex (Bishop et al., 2010; Shaked et al., 2018; Rosen et al., 2019). A review by Beebe and Gozal linked executive impairment of OSA patients to prefrontal cortex dysfunction (Beebe and Gozal, 2002). To our knowledge, however, only one study has examined the role of the prefrontal cortex in reducing executive function after hypoxia exposure in healthy individuals (Ochi et al., 2018). Although this study by Ochi et al. found a statistically significant relationship between impaired Stroop test performance and lower dorsolateral prefrontal cortex activation after a combined intervention of exercise and hypoxia, this relationship was not significant in hypoxia at rest. However, the oxygen saturation was also lower in the combined exercise and hypoxia intervention than in the hypoxia at rest intervention, which is expected as, exercise acutely decreases the oxygen saturation both combined with and independent of hypoxia (Woorons and Richalet, 2021). Therefore, a similar effect on the lower dorsolateral prefrontal cortex might be found for more severe hypoxia. Further research into the interaction between hypoxia, prefrontal cortex activation and executive function could help to get a better understanding of the physiological response to hypoxia.

Within the inhibition cluster, the Stroop test more often showed impairments (42%) than when other inhibition tests were used (mean of 26%). This may reflect the dependence of the performance of the Stroop test on accurate colour discrimination. Although this systematic review did not specifically search for visual function studies, the retina has been reported to be highly sensitive to hypoxia (Selvam et al., 2018). Based on this observation we recommend choosing the Stroop test over other inhibition tasks.

In the memory domain, delayed recognition was the most sensitive to hypoxia. Delayed auditory/verbal recognition tests showed the highest sensitivity among all clusters (83%). Delayed recognition is an aspect that can be highly relevant while navigating through new terrain. Therefore, this should be considered especially for exploratory expeditions at altitude.

Although the number of studies using neurophysiological tests during acute hypoxia was limited, our data suggest that tests within the neurophysiological domain could be a particularly sensitive tool to assess CNS functions during hypoxia. At an oxygen saturation between 89% and 80%, the evoked potential cluster showed higher sensitivity than any other cluster. This is in line with previous research by Tsarouchas et al. which concluded that evoked brain responses can be used for early detection of cognitive alterations during exposure to moderate hypobaric hypoxia (Tsarouchas et al., 2008).

The impact of barometric pressure on the severity of hypoxia is a subject of debate (Savourey et al., 2003; Savourey et al., 2007; Millet et al., 2012; Naughton et al., 2012; Coppel et al., 2015; McMorris et al., 2017). In light of this, a secondary objective of this review was to explore the effects of barometric pressure on cognitive function. No significant difference was found in test sensitivity due to barometric pressure in the overall analysis at all oxygen saturation levels. Similarly, when only the (complex) attention or executive function domains were considered, no significant difference was found. Although the number of studies was too small to be conclusive, this review does not provide indications that barometric pressure has a large impact on hypoxia sensitivity. This supports the traditional consensus that normobaric and hypobaric hypoxia can be used interchangeably (Naughton et al., 2012). However, more recent findings of several studies indicate that normobaric and hypobaric hypoxia have different physiological effects (Savourey et al., 2003; Savourey et al., 2007; Millet et al., 2012; Coppel et al., 2015; McMorris et al., 2017). Coppel et al. suggested that the traditional view might be based on the barometric pressure only showing effects for longer exposure times (>3 h) (Coppel et al., 2015). This could explain why this review of acute hypoxia showed no effect of barometric pressure, as only 8% of our studies used exposure times of more than 3 h, and only one of these tested the executive domain.

One limitation of this study was its reliance on characterizing the literature based on inspired oxygen fraction or altitude levels. This approach is probably hiding relatively large interindividual SpO₂ variability and therefore mixing within single studies participants with very different hypoxemic levels. This limitation is inherent to experiments where oxygen saturations are not individually controlled, which constituted the vast majority of the studies in this review. To address this limitation and provide a clearer understanding of the results, we expressed the variability in Table 1 within each study by providing measures such as standard deviation (SD), standard error of the mean (SEM), or 95% confidence interval (95% CI).

An associated limitation of this review may be the variation in the distribution and heterogeneity of SpO₂ levels between different tests or clusters, which were derived from a heterogeneity of different studies. This disparity in SpO₂ levels may account for the observed differences in sensitivity to hypoxia among these clusters. For instance, clusters with studies encompassing a wide range of hypoxic levels may have an increased likelihood of demonstrating higher sensitivity to hypoxia severity, whereas clusters with studies investigating a narrow range of hypoxic levels may have reduced chances of showing hypoxia sensitivity. Consequently, this effect could introduce bias in the analysis of the hypoxic-dose response.

A third limitation of this study is that the majority of included studies were not blinded. This lack of blinding may introduce bias, as participants and researchers being aware of the hypoxic or normoxic exposure could inadvertently influence cognitive performance assessments.

All these factors reflect variabilities and disparities between hypoxia studies, which limits the conclusions that can be reached from the literature, and the selection of the most sensitive and reliable tests of cognitive effects of acute hypoxia. This emphasizes the need for more standardisation of methodologies, to improve the comparability and generalizability of hypoxia experiments.

A large variety of cognitive tests were used in the literature to assess the effects of acute hypoxia on cognition in healthy adults. This huge methodological diversity is a major detriment to the investigation of the CNS effects of hypoxia. With these limitations, some suggestions can be made. The Stroop test as well as the clusters of sustained attention, auditory/verbal: delayed recognition, visual/spatial: delayed recognition, and evoked potential showed higher sensitivity than other tests and clusters. All clusters within the executive domain with more than 10 data points showed a clear level-response relationship or trend, with more frequent impairments at more severe levels of hypoxia. The data in our review showed no different physiological effect between hypobaric hypoxia and normobaric hypoxia. To further improve our understanding of the effects of acute hypoxia on cognition, standardization of exposure protocols and cognitive testing is crucial.