Oxygen Therapy in Critical Illness

Precise Control of Arterial Oxygenation and Permissive Hypoxemia

Daniel Stuart Martin, BSc, MBChB, PhD, FRCA, FFICM; Michael Patrick William Grocott, MBBS, MD, FRCA, FRCP, FFICM


Crit Care Med. 2013;41(2):423-432. 

In This Article

Novel Clinical Strategies for Oxygen Prescription

Moving away from targeting normal values as physiological goals[84] for oxygen therapy in critically ill patients is consistent with recent paradigm shifts in relation to other interventions. These include hemoglobin concentration ([Hb])[85] and arterial partial pressure of carbon dioxide.[86] For oxygen therapy, consideration of how the balance of benefit and harm alters over time may also be important. For example, specific oxygen delivery targets seem to be effective in the resuscitation of acutely injured patients,[87–90] whereas elevating systemic oxygen delivery above normal does not improve outcome in established critical illness and may cause harm.[91–93] The same may be true for maintenance of PaO2 as critical illness develops with time.

Precise Control of Arterial Oxygenation

PCAO involves targeting of PaO2 or SaO2 to individually specified values, and avoiding significant fluctuation outside of a tightly defined range, thereby minimizing the potential harms associated with hyperoxemia and hypoxemia (unnecessarily high or low PaO2 and/or PIO2). The traditional clinical approach to oxygen therapy has been to prioritize the avoidance of hypoxemia while being relatively tolerant of hyperoxemia. The possibility that "too much" oxygen may be as harmful as "not enough" should[94] lead to a pragmatic rethinking of the practice of oxygen administration. Observational data suggests that current practice tends to hyperoxemia,[59] 60). This may be an inevitable consequence of the widespread use of pulse oximetry that effectively detects hypoxemia but cannot be used to differentiate between normoxemia and hyperoxemia (ceiling effect). The result of PCAO should be oxygenation values that fall within a considerably narrower range than is currently common, the midpoint of which is appropriately targeted for a specific patient. Prescription of an achievable range (e.g., "60 to 75 mm Hg" or 8 to 10 kPa) might replace the more commonly observed prescription of "> 60 mm Hg" (8 kPa).

Founded upon physiological first principles, one can construct a theoretical schema that depicts a patient's target oxygenation zone (Fig. 1). The choice of values for a specific patient will depend upon their age, the clinical setting, underlying disease (and its chronicity), and other comorbidities. Agreed target values may be suitable for cohorts of patients, for example postoperatively or following a myocardial infarction. This approach to PCAO can be compared to other well-founded, evidence-based practices in critical care medicine; for example, the administration of intravenous fluids to optimize intravascular volume status[95] is now commonly guided by measurable end points such as stroke volume. In this instance, increased risks are encountered if physiological goals are ignored and fluid is administered in a uniformly prescriptive style (e.g., volume per hour or kilogram) rather than being tailored according to individual's requirements.[96] Tight control of blood glucose again endorses this approach, having a significant impact on mortality and morbidity in the critically ill.[97]

Figure 1.

Schematic diagram of the precise control of arterial oxygenation concept. A target arterial partial pressure of oxygen or arterial hemoglobin oxygen saturation is selected for each patient (thick dashed,arrowed line in the center of curve) around which tight boundaries are delineated that create the therapeutic target range for oxygenation (thindashed lines). Harm is possible if oxygenation strays outside of this selected range. The optimal range for individuals will be dependent upon their specific clinical situation.

Permissive Hypoxemia

The concept of target defined arterial oxygenation described above (PCAO) is echoed in a recently published guideline for acutely ill patients which suggests normal or near normal oxygenation as the goal for oxygen therapy rather than unrestricted administration of oxygen to hypoxemic patients.[8] However, while targeting normoxemia may be the best practice in acute situations, it may be neither achievable nor beneficial in critically ill patients exposed to subacute or sustained hypoxemia. In patients who have had sufficient time to be adapting or adapted to a subacute or sustained hypoxemia (Table 2), a strategy of PH may improve outcomes because of the marginal benefit (and potential increased harm) that arises from increasing arterial oxygenation to normal. In other words, the goal of PH is to reduce morbidity and mortality in selected hypoxemic patients who have had sufficient time to adapt this state, by targeting lower levels of arterial oxygenation than are currently acceptable. Conceptually, this can be presented as a shift to the left of the PCAO curve in Figure 1 (lower PaO2), with the consequence that the zone of "optimal outcome" lies within the sector conventionally described as hypoxemia: hence PH (Fig. 2). The corollary of this is that normoxemia may be associated with worse outcome in these individuals. Within the oxygenation target zone (left of center) in Figure 2, adaptation to hypoxemia may facilitate a reduction of risk, in a similar way to the acclimatization process that occurs on ascent to high altitude permits continued functioning even under conditions where inspired oxygen is profoundly reduced.[75] This rationale for PH is in part a reflection of the fact that humans posses a variety of effective adaptive mechanisms that support hypoxia tolerance, whereas hyperoxia seems to be consistently harmful due to the absence of known defensive adaptations.

Figure 2.

The precise control of arterial oxygenation concept demonstrating the potential risk reduction presented by permissive hypoxemia. Shifting the therapeutic target range for oxygenation (area between thindashed lines) to the left on this conceptogram could potentially reduce harm to selected patients by tolerating increasing degrees of hypoxemia and avoiding interventions that pursue normoxemia or lead to hyperoxemia. Cellular and organ "acclimatization" may occur during subacute and sustained hypoxemia that facilitates survival without increased harm, which occurs during prolonged ascent to high altitude.

The two proposed strategies (PH and PCAO) could be used in combination; the application of PH without PCAO risks the unintended consequence of unacceptably low PaO2.

Individualizing Oxygen Therapy in Critical Illness

There is wide variability in human responses to a hypoxic stimulus, strikingly demonstrated by dramatic interindividual differences in performance at high altitude.[98] Prediction of an individual's response to hypoxemia is very poor, and neither isolated variables relating to an ability to improve oxygen transport (e.g., hypoxic ventilatory response), nor measures of physiological reserve relating to overall oxygen flux (e.g., peak oxygen consumption), are predictive of subsequent hypoxia tolerance.[21] That said, predictor variables for acute mountain sickness were recently identified in a large cohort of subjects ascending to high altitude, and these consisted of marked desaturation and low ventilatory response to hypoxia during exercise.[99] Comparable predictor variables in critically ill patients are largely unknown, and although individual risk stratification according to exercise capacity perioperatively is now commonplace,[100] a "one-size fits all" approach is still commonly adopted.

The etiology and time-course of hypoxemia will affect the optimal level of arterial oxygenation for an individual patient (Table 2). Our current understanding of the transition from acute response to a more adapted phenotype is limited in critically ill patients and there is likely to be substantial interindividual variation in the magnitude and time-course of these processes. For example, the targets for arterial oxygenation, [Hb] and blood pressure in a patient with a traumatic chest injury may well be different from those in a patient with long-standing multiple organ failure secondary to sepsis. Clinical decisions should be driven by a pragmatic approach to individual patients guided by the best available clinical evidence.

Patient Selection for PCAO and PH

Although most patients are likely to benefit from PCAO, not all patients will tolerate profound hypoxemia. Hypoxemia is currently contraindicated in patients with severe brain injuries and evidence exists that increasing systemic oxygenation with supplemental oxygen following major surgery reduces postoperative wound infection rates.[101] Given that even in these patients excessive oxygenation is likely to be harmful, and PCAO is likely to be beneficial, the identification of "susceptibility" biomarkers for hypoxia tolerance or intolerance is imperative and would facilitate individualized implementation of PCAO and PH to patients. Such biomarkers might include physiological variables, biochemical signals in plasma or other compartments, genetic loci, and epigenetic modifications.[102] Rapid diagnostic technologies currently being developed might permit bedside characterization of the likely rate and degree of adaptation to hypoxemia and thereby guide management.

Patient Responses and Targeting of Treatments

Should a distinct group of "response" biomarkers be identified that indicate hyperoxic or hypoxic tissue damage, they could also be used to monitor treatment and modify therapeutic strategies for individual patients. Markers of cellular damage (e.g., S100),[103] beneficial adaptation (nitrogen oxides),[104] or oxidative stress[105] may be useful for this purpose. In addition, continuous monitoring of tissue oxygenation to provide a real-time readout of the balance between oxygen delivery and consumption is also likely to be imperative to successful PH. Technologies such as near infrared spectroscopy,[106] real-time in vivo speckle laser,[107] Clarke electrodes, microdialysis, and fluorescence quenching[108] offer the potential of monitoring oxygenation in a variety of tissues and the potential for generating organ-specific oxygen toxicity profiles. It is difficult to pinpoint the threshold for oxygen-related tissue damage in humans; the precise FIO2 is likely to differ between individuals and depend upon their degree of underlying lung injury. The consequences of ROS-mediated oxidative damage due to high FIO2 in the lung may be identified through analysis of pulmonary surfactant with characterization of phospholipid oxidative damage using high precision modern diagnostics such as bedside mass spectrometry. The identification of such markers is an unmet research need with the potential to improve the safety and efficacy of oxygen therapy.

Taken further, perhaps the ultimate goal, as with other contexts in medicine where precise control of a monitored variable is required, would be the introduction of "servo control" systems to permit the automated control of arterial oxygenation. With appropriate monitoring and safety structures in place, systems based on high quality input variables, such as that derived from a reliable pulse oximetry source or continuous intra-arterial oxygen tension monitoring,[109] could be linked to variable oxygen administration systems, to allow real-time management of hypoxemia.

Oxygen Delivery and PH

When implementing PH, it may necessary to manipulate [Hb] and cardiac output in subgroups of patients to ensure adequate convective oxygen delivery to tissues.[5] Patients with reduced oxygen delivery due to low [Hb] (e.g., Jehovah's Witness post surgery) or low cardiac output (e.g., end-stage heart failure) may therefore not be suitable candidates for PH but may still benefit from modified target values for PaO2. Determination of the precise thresholds for [Hb] and cardiac output in the context of a selected oxygenation target will depend on basal metabolic oxygen requirements and should be guided by the use of biomarkers and monitors of tissue oxygenation/hypoxia. While it may be necessary to reduce metabolic requirements in patients in whom hypoxemia is severe and oxygenation targets are low, for example through the use of sedatives, muscle relaxants, or therapeutic hypothermia, tissue oxygen extraction might also be amenable to manipulation via matching of microvascular blood flow to local tissue demands.[110]