Cerebral Oximetry During Pediatric In-Hospital Cardiac Arrest: A Multicenter Study of Survival and Neurologic Outcome

Abstract

OBJECTIVES:

To determine if near-infrared spectroscopy measuring cerebral regional oxygen saturation (crSo₂) during cardiopulmonary resuscitation is associated with return of spontaneous circulation (ROSC) and survival to hospital discharge (SHD) in children.

DESIGN:

Multicenter, observational study.

SETTING:

Three hospitals in the pediatric Resuscitation Quality (pediRES-Q) collaborative from 2015 to 2022.

PATIENTS:

Children younger than 18 years, gestational age 37 weeks old or older with in-hospital cardiac arrest (IHCA) receiving cardiopulmonary resuscitation greater than or equal to 1 minute and intra-arrest crSo₂ monitoring.

INTERVENTIONS:

None.

MEASUREMENTS AND MAIN RESULTS:

Primary outcome was ROSC greater than or equal to 20 minutes without extracorporeal membrane oxygenation. Secondary outcomes included SHD and favorable neurologic outcome (FNO) (Pediatric Cerebral Performance Category 1–2 or no change from prearrest). Among 3212 IHCA events (index and nonindex), 123 met inclusion criteria in 93 patients. Median age was 0.3 years (0.1–1.4 yr) and 31% (38/123) of the cardiopulmonary resuscitation events occurred in patients with cyanotic heart disease. Median cardiopulmonary resuscitation duration was 8 minutes (3–28 min) and ROSC was achieved in 65% (80/123). For index events, SHD was achieved in 59% (54/91) and FNO in 41% (37/91). We determined the association of median intra-arrest crSo₂ and percent of crSo₂ values above a priori thresholds during the: 1) entire cardiopulmonary resuscitation event, 2) first 5 minutes, and 3) last 5 minutes with ROSC, SHD, and FNO. Higher crSo₂ for the entire cardiopulmonary resuscitation event, first 5 minutes, and last 5 minutes were associated with higher likelihood of ROSC, SHD, and FNO. In multivariable analysis of the infant group (age < 1 yr), higher crSo₂ was associated with ROSC (odds ratio [OR], 1.06; 95% CI, 1.03–1.10), SHD (OR, 1.04; 95% CI, 1.01–1.07), and FNO (OR, 1.05; 95% CI, 1.02–1.08) after adjusting for presence of cyanotic heart disease.

CONCLUSIONS:

Higher crSo₂ during pediatric IHCA was associated with increased rate of ROSC, SHD, and FNO. Intra-arrest crSo₂ may have a role as a real-time, noninvasive predictor of ROSC.

Cardiac arrest occurs in 2–6% of children admitted to a PICU (1–4). Survival to hospital discharge (SHD) after pediatric in-hospital cardiac arrest (IHCA) in the United States remains less than 45% despite advances in resuscitation science over the last 20 years (5–7). Current guidelines recommend monitoring chest compression mechanics (e.g., depth, rate), invasive hemodynamics (diastolic blood pressure [DBP]), and end-tidal co₂ (ETCO₂) to guide resuscitation and ensure high-quality cardiopulmonary resuscitation (CPR) (8); however, none of these modalities directly monitors the oxygen delivery to the brain. Although ETCO₂ monitoring has been proposed as a marker of organ perfusion and return of spontaneous circulation (ROSC), its utility is limited by ventilation: perfusion mismatch and its poor correlation with cerebral blood flow.

Cerebral near-infrared spectroscopy (NIRS) non-invasively measures the regional difference between oxygenated and deoxygenated hemoglobin, an expression of the difference in oxygen supply and demand (regional oxygen saturation). When applied to the patient’s forehead, it allows the evaluation of cerebral regional oxygen saturation (crSo₂) from superficial brain structures, which are among the most susceptible areas to ischemic-hypoxic injury (9). Unlike pulse oximetry, NIRS does not require pulsatile flow and thus can measure crSo₂ during low-flow or no-flow states such as cardiac arrest. During adult IHCA and out-of-hospital cardiac arrest (OHCA), higher crSo₂ values during CPR are associated with higher rates of ROSC. Additionally, having crSo₂ of at least 50% was associated with increased odds of ROSC (10, 11) and favorable neurologic outcome (FNO) (12). ROSC rarely occurred when crSo₂ was less than 30% (10). There are limited data on whether cerebral oximetry can serve as a useful predictor of ROSC and SHD in pediatric cardiac arrest (13–16). The International Liaison Committee on Resuscitation (ILCOR) and the American Heart Association have identified the need for pediatric-specific data regarding whether NIRS-guided CPR improves outcomes (17). Thus, we aimed to investigate cerebral oximetry during pediatric IHCA and whether higher crSo₂ during CPR is associated with ROSC, SHD, and FNO.

MATERIALS AND METHODS

Study Population and Enrollment

We included cardiac arrest events from three sites (Medical City Children’s Hospital, n = 54; Children’s Hospital of Philadelphia, n = 5; and Children’s Hospital of Wisconsin, n = 64) in the pediatric Resuscitation Quality (pediRES-Q) Collaborative from 2015 to 2022. The pediRES-Q Collaborative (ClinicalTrials.gov, NCT02708134) is an international multicenter pediRES-Q improvement network, previously described in detail (Supplemental Appendix 1, http://links.lww.com/CCM/H484) (18). The collection, management, and analysis of data in this study (institutional review board [IRB] no. 15–12099) was approved by the Children’s Hospital of Philadelphia IRB (Federal Wide Assurance Identifier: FWA00000459) and determined that the study met criteria for a waiver of consent per Code of Federal Regulations (CFR) 45 CFR 46.116(d) and 45 CFR 46.408(a). Sites participating in the collaborative were approved by their local IRB or research ethics boards and a Data Use Agreement was obtained. Inclusion criteria were IHCA, age younger than 18 years, gestational age 37 weeks old or older, CPR lasting greater than or equal to 1 minute, and intra-arrest cerebral oximetry monitoring. Exclusion criteria included OHCA. Data use agreements were obtained for each institution and compliance with the Health Insurance Portability and Accountability Act of 1996 was maintained.

Study Definitions and Outcome Measures

Cardiac arrest was defined as having no pulse or a pulse with poor perfusion requiring chest compressions. Sustained ROSC was defined as return of a palpable pulse for at least 20 minutes without extracorporeal membrane oxygenation (ECMO). Events without ROSC included patients who died and patients who achieved return of circulation (ROC) with ECMO. ROC was defined as sustained ROSC or ROC with ECMO. Survival with FNO was prospectively defined as a Pediatric Cerebral Performance Category (PCPC) score of 1 or 2 at hospital discharge or no change from prearrest PCPC score (19). Unfavorable neurologic outcome was defined as PCPC 3–6. The primary outcome was ROSC. Secondary outcomes included SHD and survival with FNO.

Patient Characteristics

Data corresponding to potential confounders and effect modifiers for initial and sustained ROSC or survival and neurologic outcomes were collected. Prearrest data included patient demographics and prearrest characteristics. Intra-arrest data included chest compression quality metrics, timing of arrest, presence of invasive monitoring before the onset of arrest, initial rhythm, duration of chest compressions, medications administered, and use of ECMO. Postarrest data included arrest outcome, hospital survival outcomes, and selected postresuscitation care variables.

Use of Cerebral Oximetry

Cerebral oximetry was measured per institutional clinical practice. Two hospitals used INVOS 5100C (Medtronic, Minneapolis, MN) devices and placed cerebral oximetry probes on all patients at admission (CPR event n = 118) and one used the Equanox 7600 (Nonin Medical, Plymouth, MN) device and placed cerebral oximetry probes at the time of the cardiac arrest event (CPR event n = 5). A single sensor on either side of the forehead was considered sufficient to measure crSo₂ because measurements have been found to be equal on both sides of the forehead during CPR (11). The oximeters measured crSo₂ at intervals of one value every 4–60 seconds depending on the device used (INVOS q4 s; Equanox q60 s). Data were recorded and stored on the NIRS device or BedMasterEX (Excel Medical, Jupiter, FL). Data were de-identified and uploaded to the pediRES-Q secure research server. Data were subsequently cleaned to remove artifact and periods of low-signal quality. Study data were managed using Research Electronic Data Capture tools hosted at Children’s Hospital of Philadelphia, the pediRES-Q data coordinating center (20). An example of prearrest, intra-arrest, and postarrest crSo₂ monitoring as depicted on the NIRS device for study patients is available in Supplemental eFigure 1 (http://links.lww.com/CCM/H484).

Statistical Analysis

Continuous data are presented as median (interquartile ranges) due to nonparametric distributions; categorical data are presented as counts and percentages. Categorical values were compared using Fisher exact test or chi-square test; Wilcoxon rank-sum tests compared continuous variables.

For analyses evaluating associations between crSo₂ and event survival ROSC outcomes, all CPR events were analyzed. For analyses evaluating associations between NIRS and SHD or discharge with FNO, only the index event for each patient was analyzed. We compared median intra-arrest crSo₂ and percent of crSo₂ measurements above a priori thresholds (20%, 30%, 40%, 50%, 60%) during the: 1) entire CPR event, 2) first 5 minutes, and 3) final 5 minutes with ROSC, SHD, and FNO.

We performed additional analyses to assess median crSo₂’s ability to predict sustained ROSC. Receiver operating curves are presented and Youden’s index criteria were used to determine optimal cutoffs where sensitivity and specificity are maximized. These are shown for the overall cohort and stratified by age and cyanotic heart disease. Bootstrap model validation method was employed to validate the prediction ability of ROSC using crSo₂ median with area under the curve (AUC). Lastly, multivariable logistic regression analysis was performed for infants (age < 1 yr) to assess the relationship between median crSo₂ and all outcomes controlling for the presence of cyanotic heart disease as this patient population would have baseline arterial oxygen saturations below noncyanotic patients and thus lower baseline cerebral oxygen saturations than noncyanotic patients.

A p value of less than 0.05 was considered statistically significant. Due to the exploratory nature of this analysis, p values are not adjusted for multiple comparisons. Statistical analysis was performed with R, R stats, and pROC package (Version 4.2.2; R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Demographics and Clinical Characteristics

There were 123 IHCA events in 93 patients eligible for inclusion (Fig. 1). Median age was 0.3 years (0.1–1.4 yr), 69 (56%) were male, and 38 (31%) had cyanotic heart disease. Patient event demographic and prearrest characteristics are summarized in Table 1. Event-level data are summarized in Supplemental eTable 1 (http://links.lww.com/CCM/H484).

Figure 1.

Consort diagram of enrolled study patients. CA = cardiac arrest, CPR = cardiopulmonary resuscitation, NIRS = near-infrared spectroscopy, OHCA = out-of-hospital cardiac arrest, PediRESQ = Pediatric Resuscitation Quality, ROSC = return of spontaneous circulation, SHD = survival to hospital discharge.

TABLE 1.

Patient Demographics and Prearrest Characteristics of All Cardiac Arrest Events by Return of Spontaneous Circulation Status

Patient Characteristics	Overall (n = 123)a	ROSC (n = 80)a	No ROSC (n = 43)a	p b
Age (yr)	0.3 (0.1–1.4)	0.3 (0.1–0.6)	0.7 (0.2–4.2)	0.008
Infant (age < 1 yo)				0.001
Infant	90 (73)	66 (82)	24 (56)
Noninfant	33 (27)	14 (18)	19 (44)
Male	69 (56)	44 (55)	25 (58)	0.7
Race				0.2
White	66 (54)	38 (48)	28 (65)
African American	32 (26)	25 (31)	7 (16)
Other	12 (9.8)	9 (11)	3 (7)
Unknown/not reported	12 (9.8)	7 (8.8)	5 (12)
Ethnicity				> 0.9
Hispanic	33 (27)	21 (26)	12 (28)
Not Hispanic	78 (63)	51 (64)	27 (63)
Weight (kg)	5.0 (3.5–9.7)	4.8 (3.3–7.0)	6.4 (4.1–14.1)	0.033
Recurrent cardiopulmonary resuscitation event	30 (24)	16 (20)	14 (33)	0.12
Index event (yes)	93 (76)	64 (80)	29 (67)	0.13
Prearrest conditions
Cardiac malformation	76 (62)	51 (64)	25 (58)	0.5
Single ventricle heart disease	34 (28)	23 (29)	15 (35)	0.5
Cyanotic heart disease	38 (31)	23 (29)	15 (35)	0.5
Congenital malformation (noncardiac)	17 (14)	10 (12)	7 (16)	0.6
Respiratory insufficiency	104 (85)	65 (81)	39 (91)	0.2
Congestive heart failure	59 (48)	39 (49)	20 (47)	0.8
Renal insufficiency	22 (18)	13 (16)	9 (21)	0.5
Chromosome genetic abnormality	22 (18)	9 (11)	13 (30)	0.009
Illness category				0.8
Surgical cardiac	45 (37)	32 (40)	13 (30)
Medical cardiac	36 (29)	21 (26)	15 (35)
Surgical noncardiac	3 (2)	1 (1)	1 (2)
Medical noncardiac	37 (30)	24 (30)	13 (30)
Trauma	2 (2)	1 (1)	1 (2)
Prearrest PCPC				0.3
PCPC 1, normal	78 (64)	53 (66)	25 (60)
PCPC 2, mild disability	13 (11)	8 (10)	5 (12)
PCPC 3, moderate disability	6 (5)	2 (3)	4 (10)
PCPC 4, severe disability	1 (1)	0 (0)	1 (2)
PCPC 5, coma/vegetative state	2 (2)	1 (1)	1 (2)
Unable to determine	22 (18)	16 (20)	6 (14)

PCPC = Pediatric Cerebral Performance Category, ROSC = return of spontaneous circulation.

^an (%) or median (interquartile range).

^bFisher exact test; Pearson’s χ² test; and Wilcoxon rank-sum test.

ROSC, Survival, and Neurologic Outcomes

Overall, 80 of 123 events (65%) achieved sustained ROSC, 26 events (21%) achieved ROC with ECMO, and 17 (14%) did not survive the CPR event. Of 93 index events, 54 patients (58%) survived to hospital discharge, 37 patients (40%) did not survive to hospital discharge, and two patients (2%) were still hospitalized at the time of data analysis. Of 54 survivors, 36 (67%) had FNO at hospital discharge (Table 2).

TABLE 2.

Survival Outcomes

Survival Outcome	Overalla
Immediate event outcome	(Total = 123)
ROC	106 (86)
Return of spontaneous circulation ≥ 20 min	80 (65)
ROC with extracorporeal cardiopulmonary resuscitation	26 (21)
Died	17 (14)
Hospital discharge outcomes	(Total = 93)
24-hr survival	76 (82)
Survival to ICU discharge	55 (59)
Survival to hospital discharge	54 (58)
Unknown survival status (still hospitalized)	2 (2)
Declared do-not-attempt resuscitation during admission (yes)	11 (12)
Neurologic outcomes	(Total = 79)b
Favorable outcomes (PCPC 1–2 or no change)	36 (46)
Unfavorable outcomes (PCPC 3–6)	43 (54)
Discharge Pediatric Cerebral Performance Category
1 = normal	32 (41)
2 = mild disability	4 (5)
3 = moderate disability	3 (4)
4 = severe disability	2 (3)
5 = coma/vegetative state	1 (1)
6 = death	37 (47)
Unable to determine	14 (13)

ROC = return of circulation, PCPC = Pediatric Cerebral Performance Category.

^an (%).

^bDischarge PCPC score was unable to be determined in 14 patients due to paralysis/sedation at transfer to an outside hospital or not documented.

Table 3 depicts the comparison of the median crSo₂ and percent of measurements above crSo₂ thresholds among all cardiac arrest events by ROSC, SHD, and FNO. The median crSo₂ value among all 123 events was 38% (22–53%). The median crSo₂ was significantly higher for events with ROSC compared without ROSC for the entire CPR event (44% [30–60%] vs. 26% [15–40%]; p < 0.001), first 5 minutes (42% [28–58%] vs. 29% [17–42%]; p = 0.002), and final 5 minutes (44% [32–62%] vs. 30% [16–45%]; p = 0.002). Patients who achieved ROSC had a higher percent of crSo₂ measurements above all a priori thresholds for the entire CPR event. All patients who achieved ROSC had crSo₂ above 30% throughout the CPR event.

TABLE 3.

Comparison of Median Cerebral Oxygen Saturation and Percent of Measurements Above Cerebral Oxygen Saturation Thresholds Among All Cardiac Arrest Events by Return of Spontaneous Circulation, Survival to Hospital Discharge, and Favorable Neurologic Outcome

Cerebral Oxygen Saturation	All Events (n = 123)a	ROSC (n = 80)a	No ROSC (n = 43)a	p
crSo2 entire event	38% (22–55%) (n = 123)	44% (30–60%)	26% (15–40%)	< 0.001
crSo2 first 5 min	38% (23–55%) (n = 122)	42% (28–58%)	29% (17–42%)	0.002
crSo2 last 5 min	41% (26–56%) (n = 122)	44% (32–62%)	30% (16–45%)	0.002
Percent of measurements with crSo2 above a priori thresholds for entire event:
≥ 20%	100% (55–100%) (n = 123)	100% (93–100%)	83% (23–100%)	< 0.001
≥ 30%	94% (25–100%) (n = 123)	100% (50–100%)	27% (4–89%)	< 0.001
≥ 40%	45% (0–100%) (n = 123)	68% (0–100%)	16% (0–55%)	0.002
≥ 50%	8% (0–91%) (n = 123)	25% (0–100%)	0% (0–15%)	0.018
≥ 60%	0% (0–23%) (n = 123)	0% (0–56%)	0% (0–1%)	0.010
Cerebral Oxygen Saturation	All Patients (n = 91)	SHD (n = 54)	No SHD (n = 37)	p
crSo2 entire event	38% (23–54%) (n = 91)	44% (33–59%)	26% (18–43%)	< 0.001
crSo2 first 5 min	37% (25–54%) (n = 90)	44% (31–57%)	30% (18–42%)	0.003
crSo2 last 5 min	41% (26–56%) (n = 90)	49% (34–61%)	30% (18–43%)	0.002
Percent of measurements with crSo2 above a priori thresholds for entire event:
≥ 20%	100% (59–100%) (n = 91)	100% (92–100%)	88% (47–100%)	0.016
≥ 30%	89% (26–100%) (n = 91)	100% (71–100%)	36% (2–100%)	< 0.001
≥ 40%	50% (0–100%) (n = 91)	71% (17–100%)	15% (0–62%)	0.002
≥ 50%	0% (0–78%) (n = 91)	25% (0–96%)	0% (0–13%)	0.006
≥ 60%	0% (0–20%) (n = 91)	2% (0–49%)	0% (0–0%)	0.015
Cerebral Oxygen Saturation	All Patients Neurologic Outcome (n = 79)	Favorable Neurologic Outcome (n = 36)	No Favorable Neurologic Outcome (n = 43)	p
crSo2 entire event	40% (22–54%) (n = 79)	50% (36–60%)	26% (18–46%)	< 0.001
crSo2 first 5 min	38% (25–54%) (n = 78)	46% (33–58%)	30% (18–45%)	0.002
crSo2 last 5 min	41% (26–58%) (n = 78)	53% (39–63%)	32% (18–48%)	0.004
Percent of measurements with crSo2 above a priori thresholds for entire event:
≥ 20%	100% (58–100%) (n = 79)	100% (100–100%)	91% (45–100%)	0.002
≥ 30%	89% (25–100%) (n = 79)	100% (83–100%)	45% (3–100%)	< 0.001
≥ 40%	50% (0–100%) (n = 79)	85% (25–100%)	16% (0–82%)	0.002
≥ 50%	3% (0–78%) (n = 79)	38% (0–94%)	0% (0–31%)	0.028
≥ 60%	0% (0–23%) (n = 79)	1% (0–54%)	0% (0–8%)	0.068

crSo₂ = cerebral regional oxygen saturation, ROSC = return of spontaneous circulation, SHD = survival to hospital discharge.

^aMedian (interquartile range).

Receiver operator curves of median crSo₂ predicting ROSC are shown in Supplemental eFigure 2 (http://links.lww.com/CCM/H484) in all patients, infants only, noninfants only, and infants with and without cyanotic heart disease. Analyses revealed an optimal cutoff for median crSo₂ to predict ROSC of 30% for all events (AUC, 0.72 [95% CI, 0.63–0.81]; sensitivity, 0.74; specificity, 0.71), 29.5% for all infants (AUC, 0.75 [95% CI, 0.64–0.87]; sensitivity, 0.71; specificity, 0.73), and 47.8% for noninfants (AUC, 0.78 [95% CI, 0.62–0.95]; sensitivity, 0.79, specificity, 0.71). Among infants with cyanotic heart disease, the optimal cutoff for median crSo₂ to predict ROSC was 25.5% (AUC, 0.85 [95% CI, 0.71–0.99]; sensitivity, 0.77; specificity, 0.91) compared with an optimal cutoff of 37.3% (AUC, 0.69 [95% CI, 0.52–0.85]; sensitivity, 0.82; specificity, 0.56) in infants without cyanotic heart disease.

The median crSo₂ during the 91 index CPR events was 38% (23–54%). The median crSo₂ was higher in survivors vs. nonsurvivors for the entire CPR event (44% [33–59%] vs. 26% [18–43%]; p ≤ 0.001), the first 5 minutes (44% [31–57%] vs. 30% [18–42%]; p = 0.003), and the final 5 minutes (49% [34–61%] vs. 30% [18–43%]; p = 0.002). Patients with SHD had higher percent crSo₂ measurements above all a priori thresholds for the entire event (Table 4). All patients who survived to hospital discharge had a crSo₂ above 30% throughout the CPR event.

TABLE 4.

Multivariable Logistic Regression Analysis of Survival Outcomes in Infants (< 1 Yr Old) Controlling for Cyanotic Heart Disease

Survival Outcomes	Adjusted OR (95% CI)	p
Return of spontaneous circulation (event level analysis; n = 90)
crSo2 median	1.06 (1.03–1.10)	< 0.001
Cyanotic heart disease	3.25 (1.13–9.93)	0.032
Return of circulation (event level analysis; n = 90)
crSo2 median	1.03 (0.99–1.08)	0.2
Cyanotic heart disease	1.01 (0.19–4.49)	> 0.9
SHD (index event level analysis; n = 69)
crSo2 median	1.04 (1.01–1.07)	0.010
Cyanotic heart disease	0.74 (0.24–2.17)	0.6
SHD with favorable neurologic outcome (index event level analysis; n = 69)
crSo2 median	1.05 (1.02–1.08)	0.003
Cyanotic heart disease	0.40 (0.12–1.26)	0.12

crSo₂ = cerebral regional oxygen saturation, OR = odds ratio, SHD = survival to hospital discharge.

The median crSo₂ was higher in patients with FNO vs. patients with non-FNO (PCPC 6 = death included) for the 79 patients with known neurologic outcome during the entire event (50% [36–60%] vs. 26% [18–46%]; p < 0.001), the first 5 minutes (46% [33–58%] vs. 30% [18–45%]; p = 0.002), and the final 5 minutes (53% [39–63%] vs. 32% [18–48%]; p = 0.004). Patients with FNO had a higher percent of crSo₂ measurements above 20%, 30%, 40%, and 50%. Patients with FNO had all crSo₂ measurements above 30% throughout the CPR event (Table 3).

Multivariable logistic regression analysis was performed for all infants under 1 year old (n = 90) to assess the relationship between median crSo₂ and all outcomes controlling for the presence of cyanotic heart disease (Table 4). Median crSo₂ in infants was associated with ROSC (odds ratio [OR], 1.06; 95% CI, 1.03–1.10; p < 0.001), SHD (OR, 1.04; 95% CI, 1.01–1.07; p = 0.01), and FNO (OR, 1.05; 95% CI, 1.02–1.08; p = 0.003) after controlling for cyanotic heart disease.

DISCUSSION

This study of pediatric IHCA investigated the relationship between intra-arrest cerebral oximetry and ROSC, SHD, and survival with FNO. Higher median crSo₂ measured by cerebral NIRS for the entire event, and the first and last 5 minutes of the event was associated with increased rates of ROSC, SHD, and survival with FNO. Patients with ROSC and SHD had a greater proportion of the arrest with cerebral oximetry above 20%, 30%, 40%, 50%, and 60%. Patients with FNO had a greater proportion of the arrest with cerebral oximetry above 20%, 30%, 40%, and 50%. Additionally, all patients who had ROSC, SHD, and survival with FNO had crSo₂ over 30% for the entire CPR event.

In 2021, ILCOR conducted a scoping review to evaluate the use of NIRS during CPR in children, which included one case series and one cohort study (15, 16, 18). The cohort study of 10 pediatric OHCA patients, suggested minimum cerebral crSo₂ values during CPR were significantly higher in the ROSC group (mean % ± SD: 30.0 ± 1.0 in ROSC vs. 20.7 ± 5.7 in non-ROSC population; p = 0.02) (16). Since the ILCOR scoping review, one additional prospective single-center study of 21 children described the use of cerebral oximetry during pediatric cardiac arrest (17). Similar to our results, they noted that patients who achieved ROSC had higher median intra-arrest crSo₂ for the overall CPR event (62% [56–70%] vs. 45% [35–51%]; p = 0.025) and during the final 5 minutes of the event (66% [55–72%] vs. 43% [35–44%]; p = 0.01) than those who did not achieve ROSC (17). These results and our results are consistent with several adult studies despite differences in etiology of cardiac arrest, pathophysiology, and anatomy between children and adults. Two prior systematic reviews on NIRS monitoring during adult cardiac arrest showed that a higher crSo₂ was associated with a greater chance of ROSC and survival, whereas a lower crSo₂ was linked with increased mortality (21, 22). Persistently low crSo₂ could mean that CPR is either futile or inadequate. However, there seems to be no consensus on specific thresholds of crSo₂ at which a prediction can be made regarding outcomes such as ROSC, survival, or neurologic outcome (21). Furthermore, there was a wide overlap of mean or median crSo₂ values between patients with ROSC and patients where ROSC was not achieved, a finding that was also reflected in individual adult studies (23–26).

To our knowledge, this is the first pediatric study to report a significant association between cerebral NIRS and SHD, as well as survival with FNO. The only prior pediatric study to evaluate SHD found no association between crSo₂ and survival, and they speculated that the failure to detect such an association was potentially related to low statistical power (23 CA events in 21 patients) and only having five patients surviving to hospital discharge (17). Although the prior study did not find a significant association of crSo₂ with SHD, the median crSo₂ values were 14% higher in those who survived and showed a strong trend toward benefit with a p = 0.09 (17). Additionally, they noted patients who survived to discharge had 68% of crSo₂ measurements above 60% compared with 0% of measurements reaching this threshold in nonsurvivors (17). These findings are consistent with both our data and several larger adult studies that demonstrated higher crSo₂ values in patients who survived to discharge or survived with FNO (11, 12, 27).

Recent publications have strengthened the link between ETCO₂ and DBP and CPR quality, as well as associations with ROSC and SHD after pediatric IHCA. Morgan et al (28) recently published a multicenter study of children receiving CPR in ICUs demonstrating an average ETCO₂ of at least 20 mm Hg was associated with attaining ROSC and surviving to discharge (29). Prior adult studies showed that both crSo₂ and ETCO₂ are good predictors of ROSC, but crSo₂ performed better than ETCO₂(26, 30–33). This may be partially explained by the dependency of ETCO₂ on both effective minute ventilation and administration of vasoactive drugs. DBP trends are also instrumental, as maintaining a minimum DBP during CPR is associated with improved coronary perfusion pressure, vital for successful resuscitation.

Sutton et al (29) in a recent multicenter prospective observational study validated prior findings that attaining mean DBP greater than or equal to 25 mm Hg during CPR in infants less than 1 year old and greater than or equal to 30 mm Hg in older children is associated with a higher likelihood of SHD (34). Monitoring of accurate ETCO₂ and DBP, however, requires endotracheal intubation and an indwelling arterial line, which may or may not be present during a pediatric IHCA.

Cerebral oximetry may have the potential to guide treatment and decision-making during CPR, possibly in combination with ETCO₂ capnography and DPB. Most PICUs routinely use ETCO₂; however, the application of cerebral NIRS is only routinely used in pediatric cardiac ICUs to date. NIRS monitoring can be rapidly applied to the forehead and values are readable in less than 20 seconds after powering the device on, which may lend the application of NIRS at the start of a cardiac arrest event in those populations where it is not routinely used. Our data support maintaining crSo₂ above 30% as a possible target, as all patients who had ROSC, SHD, and survival with FNO had crSo₂ values above 30% for the entire CPR event. Additionally, potential cutpoints for the entire cohort and particularly the infant population predicted ROSC when median crSo₂ was also greater than 30%. Among infants with cyanotic heart disease, who have lower baseline arterial oxygen saturation, a slightly lower cutpoint of crSo₂ greater than 25% may be a possible target to predict ROSC. It is likely that the integration of all these parameters of CPR quality (ETCO₂, DBP, and crSo₂) into resuscitation protocols will provide aid in monitoring the effectiveness of interventions, guiding decision-making, and potentially improve outcomes for pediatric patients experiencing cardiopulmonary arrest.

Future studies should evaluate which resuscitative actions (e.g., medications, CPR technique, or an improvement in its quality) best improve NIRS saturation values, and how all these markers of CPR quality in combination could contribute to improving patient outcomes. Randomized clinical trials should follow to evaluate how NIRS-guided resuscitation may result in observable benefits both on survival and on neurologic outcomes. Incorporating cerebral oximetry in routine cardiac arrest care, in concert with ETCO₂ and DBP, may enable clinicians to take steps to improve cerebral and other vital organ perfusion during cardiac arrest and thus potentially improve survival, neurologic as well as functional outcomes.

Limitations of this study include the small number of patients, although it does represent the largest pediatric study to date. Sites used different NIRS devices (INVOS 5100C and Equanox 7600), although 96% of the events used the INVOS 5100C monitor, which potentially introduces a monitoring bias due to differences in characteristics in which the two devices differ (e.g., the number of light emitters and detectors, data acquisition frequency, the different distance between them and, consequently, a different tissue penetration of light, the number of wavelengths adopted, and a different built-in proprietary algorithm to assess oxygen saturation) (35). Accordingly, device-specific thresholds will probably need to be developed for ongoing research and clinical care. Since direct comparison of NIRS devices during pediatric CPR is likely not feasible, large investigations on each device are needed to identify specific thresholds. We acknowledge that 31% of arrests occurred in patients with cyanotic heart disease, although a subanalysis using multivariable regression in the infant cohort controlling for cyanotic heart disease demonstrated similar results as the primary analysis. Additionally, we included ROC with ECMO events in the analysis of the non-ROSC category but given the small numbers, we were unable to do a subanalysis of this group. A final limitation is that while there were 56 hospitals contributing data to the pediRES-Q collaborative, all the cardiac arrest events reported in this study were from three centers in the collaborative, limiting generalizability. We acknowledge that hospitals who contribute to the collaborative may be a skewed representation of centers who are dedicated to improving CPR quality in children, and thus may not represent all children’s hospitals.

CONCLUSIONS

In this multicenter international pediatric resuscitation collaborative, children with greater crSo₂ measured by cerebral NIRS during pediatric IHCA was associated with increased rates of ROSC, SHD, and survival with FNO. Further large-scale pediatric studies are needed to examine the relationship between cerebral perfusion pressure and crSo₂, as well as the relationship between cerebral NIRS and longer-term survival and neurologic outcomes.

KEY POINTS.

Question:

During pediatric in-hospital cardiac arrest, is cerebral oximetry measured by near-infrared spectroscopy associated with return of spontaneous circulation, survival to hospital discharge, and favorable neurologic outcome?

Findings:

In this multicenter observational study of pediatric in-hospital cardiac arrest, higher median cerebral oxygenation during cardiopulmonary resuscitation and a having a greater proportion of time with cerebral oximetry above 20%, 30%, 40%, and 50% during cardiac arrest are all associated with return of spontaneous circulation and neurologically favorable survival to hospital discharge.

Meaning:

These data support that higher cerebral oxygenation during cardiopulmonary resuscitation may be a metric that can be measured and targeted to improve return of spontaneous circulation and favorable outcomes.