The medical use of oxygen: a time for critical reappraisal

F. Sjöberg, M. Singer

Oxygen treatment has been a cornerstone of acute medical care for numerous pathological states. Initially, this was supported by the assumed need to avoid hypoxaemia and tissue hypoxia. Most acute treatment algorithms, therefore, recommended the liberal use of a high fraction of inspired oxygen, often without first confirming the presence of a hypoxic insult. However, recent physiological research has underlined the vasoconstrictor effects of hyperoxia on normal vasculature and, consequently, the risk of significant blood flow reduction to the at-risk tissue. Positive effects may be claimed simply by relief of an assumed local tissue hypoxia, such as in acute cardiovascular disease, brain ischaemia due to, for example, stroke or shock or carbon monoxide intoxication. However, in most situations, a generalized hypoxia is not the problem and a risk of negative hyperoxaemia-induced local vasoconstriction effects may instead be the reality. In preclinical studies, many important positive anti-inflammatory effects of both normobaric and hyperbaric oxygen have been repeatedly shown, often as surrogate end-points such as increases in gluthatione levels, reduced lipid peroxidation and neutrophil activation thus modifying ischaemia-reperfusion injury and also causing anti-apoptotic effects. However, in parallel, toxic effects of oxygen are also well known, including induced mucosal inflammation, pneumonitis and retrolental fibroplasia. Examining the available 'strong' clinical evidence, such as usually claimed for randomized controlled trials, few positive studies stand up to scrutiny and a number of trials have shown no effect or even been terminated early due to worse outcomes in the oxygen treatment arm. Recently, this has led to less aggressive approaches, even to not providing any supplemental oxygen, in several acute care settings, such as resuscitation of asphyxiated newborns, during acute myocardial infarction or after stroke or cardiac arrest. The safety of more advanced attempts to deliver increased oxygen levels to hypoxic or ischaemic tissues, such as with hyperbaric oxygen therapy, is therefore also being questioned. Here, we provide an overview of the present knowledge of the physiological effects of oxygen in relation to its therapeutic potential for different medical conditions, as well as considering the potential for harm. We conclude that the medical use of oxygen needs to be further examined in search of solid evidence of benefit in many of the current clinical settings in which it is routinely used.