Oxygen Saturation Targets and Neurologic Outcomes Following Cardiac Arrest

Abstract

Background

More than 600,000 adults in the United States experience an out-of-hospital or in-hospital cardiac arrest each year. Following resuscitation from cardiac arrest, most patients receive mechanical ventilation. The oxygenation target that optimizes neurologic outcomes following cardiac arrest is uncertain.

Research Question

Following cardiac arrest, does a lower oxygen saturation (Spo₂) target improve neurologic outcomes compared with a higher Spo₂ target?

Study Design and Methods

This study was a secondary analysis of patients who experienced a cardiac arrest prior to enrollment in the Pragmatic Investigation of Optimal Oxygen Targets (PILOT) trial. The PILOT trial assigned critically ill adults receiving mechanical ventilation to a lower (88%-92%), intermediate (92%-96%), or higher (96%-100%) Spo₂ target. This subgroup analysis compared patients randomized to a lower or intermediate Spo₂ target (88-96%) vs a higher Spo₂ target (96%-100%) regarding the primary outcome of survival with a favorable neurologic outcome at hospital discharge (Cerebral Performance Category 1 or 2).

Results

Of 2,987 patients in the PILOT trial, 339 (11.3%) experienced a cardiac arrest prior to enrollment: 221 were assigned to a lower or intermediate Spo₂ target, and 118 were assigned to a higher Spo₂ target. Overall, the median age was 60 years, 43.5% were female, 58.7% experienced an in-hospital cardiac arrest, and 10.2% had an initial shockable rhythm. Survival with a favorable neurologic outcome occurred in 50 patients (22.6%) assigned to a lower or intermediate Spo₂ target and 15 (12.7%) patients assigned to a higher Spo₂ target (absolute risk difference, 9.9 percentage points; 95% CI, 1.8-18.1; P = .03).

Interpretation

Among patients receiving mechanical ventilation following a cardiac arrest, use of a lower or intermediate Spo₂ target was associated with a higher incidence of a favorable neurologic outcome compared with a higher target. A randomized trial comparing these targets in the cardiac arrest population is needed to confirm these findings.

Graphical Abstract

FOR EDITORIAL COMMENT, SEE PAGE 1076

Take-Home Points.

Study Question: For patients following cardiac arrest, does use of a lower or intermediate oxygen saturation (Spo₂) target increase the incidence of a favorable neurologic outcome compared with a higher Spo₂ target?

Results: Among 339 patients receiving mechanical ventilation following a cardiac arrest, survival with a favorable neurologic outcome occurred in 50 patients (22.6%) assigned to a lower or intermediate Spo₂ target and 15 (12.7%) patients assigned to a higher Spo₂ target (P = .03).

Interpretation: Use of a lower or intermediate Spo₂ target was associated with a higher incidence of a favorable neurologic outcome compared with a higher target.

More than 600,000 adults in the United States experience an out-of-hospital or in-hospital cardiac arrest each year, of whom only 10% to 25% survive to hospital discharge.^{1, 2, 3} Among patients who survive cardiac arrest, brain injury is common, is the leading cause of death, and is a major contributor to long-term disability.^4,5 Following cardiac arrest, nearly all patients receive invasive mechanical ventilation. Clinicians titrate Fio₂ on the mechanical ventilator to achieve a target arterial oxygen saturation (Spo₂), which is commonly measured by pulse oximetry.³

Following cardiac arrest, patients may be particularly susceptible to ongoing brain injury from extremes of oxygenation. Hypoxemia may worsen the hypoxic-ischemic injury that commonly occurs during the cardiac arrest.⁶ Hyperoxemia may worsen brain injury by inducing cerebral vasoconstriction and creating additional reactive oxygen species.^6,7 Although preclinical and observational studies have suggested that avoiding hyperoxemia may improve neurologic outcomes in survivors of cardiac arrest,^{8, 9, 10, 11, 12, 13} clinical trials evaluating the use of lower oxygenation targets have shown conflicting results, with some suggesting a mortality benefit,^14,15 some suggesting no effect,^16,17 and some suggesting harm.¹⁸ Currently, the International Liaison Committee on Resuscitation guidelines recommend avoiding hyperoxemia with a low certainty of evidence.⁵ Whether use of a lower or intermediate Spo₂ target rather than a higher Spo₂ target to avoid hyperoxemia prevents further brain injury and increases survival following cardiac arrest remains uncertain.

To address this gap in knowledge, we conducted a secondary analysis of a recent randomized trial that evaluated Spo₂ targets among critically ill adults receiving invasive mechanical ventilation. The goal was to examine the effect of using a lower Spo₂ target of 88% to 92% or an intermediate Spo₂ target of 92% to 96% compared with a higher Spo₂ target of 96% to 100% on outcomes in patients following cardiac arrest. We hypothesized that avoiding hyperoxemia by using a lower or intermediate Spo₂ target would increase the incidence of a favorable neurologic outcome at hospital discharge.

Study Design and Methods

Study Design

This was a pre-planned subgroup analysis of the Pragmatic Investigation of Optimal Oxygen Targets (PILOT) trial, which was a cluster-randomized, cluster-crossover trial in adult patients receiving invasive mechanical ventilation in the medical ICU or the emergency department at a single academic medical center.^19,20 The PILOT trial compared the effects of lower, intermediate, and higher Spo₂ targets on ventilator-free days and in-hospital mortality among critically ill adults receiving mechanical ventilation and has been published previously.¹⁹ Patients were enrolled at the first receipt of mechanical ventilation in a study unit between July 1, 2018, and August 31, 2021. The PILOT trial was approved by the institutional review board at Vanderbilt University Medical Center (IRB#171272).

Population

All patients enrolled in the PILOT trial were screened for inclusion. To maximize the sample size, we also included patients enrolled during the prespecified analytic washout periods in which patients received the assigned Spo₂ target but were not included in the primary analysis of the trial. Patients who experienced a cardiac arrest (defined as the receipt of chest compressions or defibrillation) with sustained return of spontaneous circulation (defined as spontaneous and sustained return of a pulse for ≥ 20 consecutive minutes) prior to enrollment were included in the analysis.²¹ The PILOT trial prespecified cardiac arrest as a subgroup that was expected to modify the effect of the intervention on outcomes.²⁰

Study Interventions

The trial protocol instructed respiratory therapists to adjust the Fio₂ administered (between 21% and 100%) to achieve an Spo₂ within the assigned target range beginning within 15 minutes following initiation of mechanical ventilation and ending at discontinuation of mechanical ventilation or transfer out a participating unit, or at the end of the 2-month study period, whichever occurred first.¹⁹ Pao₂ was used to guide titration of Fio₂ for participants without functioning pulse oximetry monitoring.²⁰ For the current analysis, we reduced the 3 trial groups in PILOT (Spo₂ of 88%-92%, 92%-96%, and 96%-100%) to 2 groups to compare the lower or intermediate Spo₂ target group (88%-96%) vs the higher Spo₂ target group (96%-100%). We hypothesized that, for patients following cardiac arrest, hyperoxemia would have harmful effects through hyperoxia-induced cerebral vasoconstriction and the formation of reactive oxygen species. This study was designed to test whether a strategy to avoid hyperoxemia (which was the goal of both the lower and intermediate groups) would improve outcomes compared with a strategy not designed to avoid hyperoxemia (the higher Spo₂ group).

Data Collection

Trial personnel collected data on baseline characteristics and in-hospital outcomes from the electronic medical record. These included data on the initial presenting rhythm (shockable or non-shockable), witnessed (bystander, medical team, or unwitnessed), and cause of arrest (defined as the most likely primary cause of the arrest as determined by physician review and classified according to the Utstein guidelines).²² Cardiac arrests from a medical cause were categorized as being due to respiratory failure, cardiac, shock, or other medical causes. Cardiac arrests not from a medical cause included cardiac arrests due to trauma, drug overdose, drowning, or asphyxia.²² Data on Spo₂ and Fio₂ following enrollment were automatically extracted from the bedside monitor at a frequency of every 1 minute.²³

Study Outcomes

The primary outcome was a favorable neurologic outcome at hospital discharge, defined as a Cerebral Performance Category of 1 (good cerebral performance) or 2 (moderate cerebral disability) at the time of hospital discharge.^{24, 25, 26, 27} The Cerebral Performance Category scale is a widely used and validated measure for assessing the functional status of patients following in-hospital and out-of-hospital cardiac arrest, and is predictive of long-term outcomes.^24,25,28 The scale consists of 5 mutually exclusive categories: 1, good cerebral performance; 2, moderate cerebral disability; 3, severe cerebral disability; 4, coma or vegetative state; and 5, death. Data regarding patients’ Cerebral Performance Category at hospital discharge were recorded by 2 independent study physicians anonymized to trial group assignment following independent reviews of the entire electronic medical record using a standardized case report form. A third anonymized physician resolved cases with a disagreement between the initial 2 reviewers. Three-way disagreements were resolved by discussion among the 3 reviewers (K. P. S., S. C. D., and A. L. M.) until a consensus was reached. This method of electronic medical record review and Cerebral Performance Category determination has been previously reported and validated.²⁴

The secondary outcome was in-hospital death of any cause. Exploratory outcomes included the number of days alive and free of invasive mechanical ventilation to 28 days (ventilator-free days)²⁹ and the location of discharge, dichotomized as discharge to home (home with or without skilled care) vs all other discharge destinations, including death.

Power Calculation

This study used a fixed sample size of 339 patients (221 in the lower or intermediate Spo₂ target group and 118 in the higher target group) and assumed an absolute difference of 12.0 percentage points between groups in the incidence of favorable neurologic outcome based on prior data (23% in the lower or intermediate target group and 11% in the higher target group). Using these factors, we calculated that we would have 80% statistical power at a 2-sided alpha level of .05 to be able to reject the null hypothesis.^30,31

Statistical Analysis

The primary analysis was an unadjusted intention-to-treat comparison of the incidence of a favorable neurologic outcome (primary outcome) between patients assigned to a lower or intermediate Spo₂ target vs patients assigned to a higher Spo₂ target using a χ² test.

Two additional analyses of the Cerebral Performance Category were performed. In the first, a multivariable logistic regression model was fit for the primary outcome with independent variables of trial group assignment and the following prespecified baseline covariates: age, sex, initial shockable rhythm, witnessed (bystander, medical team, or unwitnessed), cause of arrest, and the date of enrollment in the trial. In the second, the Cerebral Performance Category was analyzed as an ordinal outcome ranging from 1 (good cerebral performance) to 5 (death) using a proportional odds model. An OR > 1.0 was used to indicate a more favorable outcome with a lower or intermediate Spo₂ target compared with a higher Spo₂ target.

The secondary outcome (in-hospital death) and exploratory outcomes (discharge to home; ventilator-free days) were compared between the 2 study groups by using a χ² test for dichotomous outcomes and a proportional odds model for ordinal outcomes.

Analyses were performed by using STATA version 18.0 (StataCorp) and R version 4.1.0 (R Foundation for Statistical Computing).

Results

Of the 2,987 patients enrolled in the PILOT trial, 378 patients had experienced a cardiac arrest when assessed at enrollment in the trial. Of these, 27 were excluded because the onset of the cardiac arrest was not prior to trial enrollment, 7 were excluded because they did not receive chest compressions or defibrillation, and 5 were excluded because return of spontaneous circulation was not sustained (Fig 1).²¹ The remaining 339 patients were included in this analysis, 221 of whom were assigned to the lower or intermediate Spo₂ targets and 118 of whom were assigned to the higher Spo₂ target. Overall, the median age was 60 years, 43.5% were female, 58.7% experienced an in-hospital cardiac arrest, and 10.2% had an initial shockable rhythm. The median time from cardiac arrest to enrollment in the trial was 105 minutes. The trial groups had similar characteristics at baseline (Table 1).

Figure 1.

Number of patients screened, excluded, and included in the analysis.

Table 1.

Baseline Characteristics

Characteristic	Lower or Intermediate Spo2 Target (n = 221)	Higher Spo2 Target (n = 118)
Age, y	59 (45, 69)	60 (46, 70)
Female sexa	95 (43.2)	52 (44.1)
Race and ethnicityb
Non-Hispanic Black	40 (18.1)	22 (18.6)
Non-Hispanic White	152 (68.8)	90 (76.3)
Hispanic	5 (2.3)	0 (0.0)
Otherc	4 (1.8)	0 (0.0)
Unknown	20 (9.0)	6 (5.1)
Medical history
Coronary artery disease	51 (23.1)	36 (30.5)
Heart failure with reduced ejection fraction	11 (5.0)	12 (10.2)
End-stage renal disease	16 (7.2)	8 (6.8)
Location of cardiac arrest
Public place	30 (13.6)	11 (9.3)
Home/residence	54 (24.4)	34 (28.8)
Assisted livingd	10 (4.5)	1 (0.8)
Study hospital	66 (29.9)	45 (38.1)
Study emergency department	20 (9.0)	5 (4.2)
Transfer hospitale	41 (18.6)	22 (18.6)
Cardiac arrest characteristics
Witnessedf
Bystander	40 (18.7)	21 (18.3)
Medical team	132 (61.7)	83 (72.2)
Unwitnessed	42 (19.6)	11 (9.6)
Initial shockable rhythmg	21 (11.0)	9 (8.7)
Received defibrillation during the arresth	36 (19.9)	22 (22.4)
Time from confirmed pulselessness to ROSC, mini	10 (4, 22)	10 (6, 20)
Cause of cardiac arrestj
Respiratory failure	56 (25.3)	42 (35.6)
Shock	41 (18.6)	13 (11.0)
Cardiac	16 (7.2)	7 (5.9)
Other medical	76 (34.4)	42 (35.6)
Non-medical	32 (14.5)	14 (11.9)
Time from arrest to enrollment, min	109 (29, 293)	105 (27, 341)
Acute conditions at time of enrollment
Myocardial infarction	77 (34.8)	38 (32.2)
Sepsis or septic shock	63 (28.5)	41 (34.7)
Glasgow Coma Scale at time of enrollmentk	3 (3, 7)	4 (3, 8)

Data are presented as median (quartile 1, quartile 3) or No. (%). ROSC = return of spontaneous circulation; Spo₂ = oxygen saturation.

^aData unavailable for 1 patient in the lower or intermediate target group.

^bRace and ethnic group were reported by patients or their surrogates as part of clinical care and obtained from the electronic medical record and grouped into fixed categories.

^cAsian, Alaskan/Indian, or Pacific Islander.

^dComprises assisted living facility, skilled nurse facility, and in-patient rehabilitation.

^eA hospital other than the study hospital, in which a patient experienced a cardiac arrest.

^fData were unavailable for 7 patients in the lower or intermediate Spo₂ target group and 3 patients in the higher Spo₂ target group. Witnessed categorized as either bystander (lay person) or medical team (first responder, emergency medical service, or organized medical response team).

^gPatients were considered to have an initial shockable rhythm if clinicians documented the initial rhythm as ventricular fibrillation, ventricular tachycardia, shockable, or the patient received immediate defibrillation. Data were unavailable for 30 patients in the lower or intermediate Spo₂ target group and 14 in the higher Spo₂ target group.

^hData were unavailable for 40 patients in the lower or intermediate Spo₂ target group and 20 patients in the higher Spo₂ target group.

ⁱData were unavailable for 26 patients in the lower or intermediate Spo₂ target group and 18 patients in the higher Spo₂ target group.

^jIndicates most likely primary cause of cardiac arrest. These were categorized as medical (respiratory failure, shock, and cardiac), other medical (eg, metabolic derangement), and non-medical (traumatic, drug overdose, drowning, and asphyxia) as defined by the Ustein variable.²⁰

^kData were unavailable for 2 patients in the lower or intermediate Spo₂ target group and 2 in the higher Spo₂ target group.

A total of 864,886 Spo₂ values were measured between enrollment and cessation of mechanical ventilation among the 339 patients, with a median interval between Spo₂ measurements of 1 minute. Oxygen saturation and Pao₂ values were lower among patients assigned to the lower or intermediate Spo₂ targets than those assigned to the higher Spo₂ target (e-Table 1, Fig 2). A single mean Spo₂ value was calculated for each patient; the median of these values was 95% (quartile 1, quartile 3, 93%, 97%) in the combined lower or intermediate target group and 97% (quartile 1, quartile 3, 96%, 99%) in the higher target group.

Figure 2.

Spo₂ and Fio₂ values in each group. Shown are the mean values (colored lines) and 95% CIs (gray shading) for the hourly mean Spo₂ as measured by pulse oximetry (A) and the Fio₂ (B) from enrollment to study day 7. The data were censored at the time that invasive mechanical ventilation was discontinued. Spo₂ and Fio₂ values were obtained approximately every 1 minute, and hourly means were calculated by averaging all measurements obtained during the hour. The number of patients who were alive and receiving invasive mechanical ventilation in each group on each day is shown on the bottom panel. Spo₂ = oxygen saturation.

Similarly, values for Fio₂ administered were lower in patients assigned to the lower or intermediate Spo₂ targets than those assigned to the higher Spo₂ target (Fig 2). The incidence of hypoxemia, defined as Spo₂ < 85%, was similar in both groups (e-Table 2).

A favorable neurologic outcome at hospital discharge occurred in 50 patients (22.6%) assigned to the lower or intermediate Spo₂ targets and 15 patients (12.7%) assigned to the higher Spo₂ target (absolute risk difference, 9.9 percentage points; 95% CI, 1.8-18.1; P = .03) (Fig 3). Results were similar in the adjusted analysis (adjusted OR, 2.24; 95% CI, 1.11-4.53; P = .02) (e-Table 3) and in the analysis treating Cerebral Performance Category scale as an ordinal outcome ranging from 1 (good cerebral performance) to 5 (death) (OR, 1.63; 95% CI, 0.99-2.68; P = .05) (Table 2).

Figure 3.

Results of the primary outcome (favorable neurologic outcome) and secondary outcome (in-hospital death). A favorable neurologic outcome at hospital discharge was defined as a Cerebral Performance Category of 1 (good cerebral performance) or 2 (moderate cerebral disability). Alive with poor neurologic outcome was defined as Cerebral Performance Category of 3 (severe cerebral disability) or 4 (coma or vegetative state). Category 5 was in-hospital death. A favorable neurologic outcome occurred in 50 patients (22.6%) assigned to a lower or intermediate Spo₂ target and 15 patients (12.7%) assigned to a higher Spo₂ target (P = .03). Spo₂ = oxygen saturation.

Table 2.

Outcomes

Outcome	Lower or intermediate Spo2 Target (n = 221)	Higher Spo2 Target (n = 118)	RD, OR, or Adjusted OR (95% CI)	P Value
Primary outcome
Favorable neurologic outcomea	50 (22.6)	15 (12.7)	RD, 9.9 (1.8 to 18.1) OR, 2.00 (1.07 to 3.76)	.03
Additional analyses of neurologic outcome
Favorable neurologic outcome, adjusted comparisonb	50 (22.6)	15 (12.7)	aOR, 2.24 (1.11 to 4.53)	.02
Cerebral Performance Categoryc			OR, 1.63 (0.99 to 2.68)	.05
Category 1	24 (10.9)	8 (6.8)
Category 2	26 (11.8)	7 (5.9)
Category 3	20 (9.1)	11 (9.3)
Category 4	5 (2.3)	3 (2.5)
Secondary outcome
In-hospital death	146 (66.1)	89 (75.4)	RD, –9.4 (–19.3 to 0.6)	.08
Exploratory outcomes
Discharge homed	44 (19.9)	14 (11.9)	RD, 8.0 (0.18 to 15.9)	.06
Ventilator-free dayse	0 (0,22)	0 (0,0)	OR, 1.62 (0.98 to 2.68)	.06
Mean ± SD	7.6 ± 11.2	5.3 ± 9.9

Data are presented as No. (%) or median (quartile 1, quartile 3) unless otherwise indicated. RD = risk difference; Spo₂ = oxygen saturation.

^aThe primary outcome, favorable neurologic outcome, was defined as a Cerebral Performance Category of 1 (good cerebral performance) or 2 (moderate cerebral disability) at the time of hospital discharge.²⁶ The interrater agreement of favorable vs unfavorable outcome between the first two independent reviewers resulted in a kappa of 0.66. A χ² test was used for between group testing of the primary outcome.

^bThe adjusted analysis of the primary outcome used a logistic regression model that included the following prespecified co-variables: age, sex, initial shockable rhythm, witnessed arrest, cause of arrest, and time from study start (in days). An OR > 1.0 indicates higher odds of survival with a favorable neurologic outcome if there is a lower or intermediate Spo₂ target than a higher Spo₂ target.

^cAnalysis of the ordinal Cerebral Performance Category, which ranged from 1 (good cerebral performance) to 5 (death), was performed using an unadjusted, proportional odds logistic regression model. An OR > 1.0 indicates a more favorable outcome with a lower or intermediate Spo₂ target than with a higher Spo₂ target.

^dDischarge home was analyzed as a binary outcome (home with or without skilled care vs all other discharge destinations, including death).

^eVentilator-free days refer to the number of days in which a patient was alive and free from the ventilator in the first 28 days following enrollment. Patients were assigned a value of 0 ventilator-free days if they died prior to day 28, continued to receive the invasive mechanical ventilation beyond day 28 or at the time of hospital discharge, whichever occurred first. An OR > 1.0 indicates a better outcome (eg, more days alive and free from the supportive therapy) with a lower or intermediate Spo₂ target than a higher Spo₂ target.

At 28 days following enrollment, 146 patients (66.1%) assigned to the lower or intermediate Spo₂ targets and 89 patients (75.4%) assigned to the higher Spo₂ target died prior to hospital discharge (absolute risk difference, –9.4 percentage points; 95% CI, –19.3 to 0.6; P = .08). Additional outcomes are presented in Table 2 and e-Tables 4 to 6.

Discussion

Among adult patients receiving invasive mechanical ventilation following cardiac arrest in this secondary analysis of a randomized trial, use of a lower or intermediate Spo₂ target (88%-96%) was associated with a higher incidence of a favorable neurologic outcome at hospital discharge compared with use of a higher Spo₂ target (96%-100%).

Preclinical data suggest that a lower oxygenation target may be associated with improved neurologic outcomes in survivors of cardiac arrest, but some clinical research has generated conflicting evidence.^{8, 9, 10, 11,32} The Reduction of Oxygen After Cardiac Arrest (EXACT) trial compared oxygen titration to achieve an Spo₂ of 90% to 94% vs an Spo₂ of 98% to 100% among patients in the prehospital or emergency department setting who had experienced an out-of-hospital cardiac arrest and found lower survival in the 90% to 94% group (absolute risk difference, –9.6%; 95% CI, –18.9 to –0.2).¹⁸ This trial suggests that early oxygen titration, often prior to Spo₂, can be reliably measured and may lead to hypoxemia and worse clinical outcomes. The Blood Pressure and Oxygenation Targets in Post-Resuscitation Care (BOX) trial compared a restrictive oxygenation target (Pao₂ of 68 to 75 mm Hg) vs a liberal oxygenation target (Pao₂ of 98 to 105 mm Hg) in comatose survivors of out-of-hospital cardiac arrest.¹⁶ The investigators found no difference in a composite of death or hospital discharge with severe disability or coma (hazard ratio, 0.95; 95% CI, 0.75 to 1.21; P = .69). Notably, both the restrictive and liberal Pao₂ targets in the BOX trial correlated with Spo₂ values within the low or intermediate target ranges of the PILOT trial; few patients experienced severe hyperoxemia. In the Intensive Care Unit Randomized Trial Comparing Two Approaches to Oxygen Therapy (ICU-ROX) trial, the suspected hypoxic-ischemic encephalopathy subgroup had significantly lower death in the conservative-oxygen group (Spo₂ target < 97%) than the usual care group (relative risk, 0.73; 95% CI, 0.54 to 0.99), suggesting a potential benefit in this population.³¹

Most prior trials of oxygenation targets following cardiac arrest are limited to patients outside of the United States and those with an out-of-hospital cardiac arrest and an initial shockable rhythm. Little evidence is available to inform care for the 300,000 patients each year who experience an in-hospital cardiac arrest and patients with a non-shockable initial rhythm, which are the patient populations largely represented in the current study.^16,18 Our results support the theory that use of a lower Spo₂ target to avoid hyperoxemia in patients following cardiac arrest may be beneficial.^14,15 Multiple ongoing randomized trials may help further address oxygenation targets in this population.^{33, 34, 35}

This secondary analysis has several strengths. The PILOT trial collected Spo₂ and Fio₂ values every minute, which allows for a highly granular assessment of the oxygenation and oxygen therapy experienced by patients in this analysis. Furthermore, this trial enrolled patients at the first receipt of mechanical ventilation in the study emergency department or ICU, capturing the early period in which the harmful effects of hyperoxia are thought to be most pronounced.³⁶ Most patients (58.7%) in this analysis experienced an in-hospital cardiac arrest, which is a patient group not well represented by most previous trials evaluating oxygenation targets following cardiac arrest. Finally, this analysis assessed functional outcomes such as the Cerebral Performance Category scale (analyzed as both a binary and ordinal outcome) and discharge destination.

This study also has several limitations. The pragmatic trial and this secondary analysis used only data available from the electronic medical record, which limits the data available for assessment of baseline characteristics such as measures of CPR quality and of outcomes for analysis. The Spo₂ and supplemental oxygen concentration experienced by patients in the lower or intermediate target group were lower than those experienced by patients in the higher target group, but whether different oxygenation targets or oxygen titration strategy would have resulted in different effects is not known. The trial enrolled patients from a medical ICU, resulting in a high proportion of patients with in-hospital cardiac arrest and without shockable rhythms. A study of this population fills an important knowledge gap but also limits our ability to compare these findings vs those of prior trials. The cerebral performance category outcome was assessed by electronic health record review as previously reported and validated, which may have limitations when compared with research staff assessments. Sample size limited our ability to definitively evaluate mortality as an outcome. Finally, the primary outcome assessment occurred at the time of hospital discharge, which may differ from neurologic outcome assessments at later timepoints following discharge.³⁷

Interpretation

Among adults receiving mechanical ventilation following cardiac arrest in a randomized clinical trial of Spo₂ targets, survival with a favorable neurologic outcome occurred more often with use of a lower or intermediate Spo₂ target (88%-96%) compared with the use of a higher Spo₂ target (96%-100%). Additional randomized clinical trials are needed to confirm these findings, particularly among patients following cardiac arrest without an initial shockable rhythm.

Funding/Support

This work was supported in part by grants from the National Institutes of Health (NIH). The following authors were supported in part by the NIH: J. D. C. [K23HL153584], S. C. D [5T32GM108554], K. P. S. [5T32HL087738], M. W. S. [K23HL143053], T. W. R. [UL1TR002243], and W. H. S. [UL1TR002243]. N. J. J. receives funding from the NIH, the Centers for Disease Control and Prevention, and the American Heart Association; and serves on guidelines committees for the American Heart Association and International Liaison Committee on Resuscitation.

Financial/Nonfinancial Disclosures

None declared.