What's new?

The brain, as the organ most sensitive to hypoxia during cardiopulmonary resuscitation, requires continuous and precise monitoring. A technology that measures regional cerebral oxygen saturation using near‑infrared spectroscopy is simple and noninvasive. In our study, this technology was used by a specialized resuscitation team during advanced resuscitation procedures and postresuscitation care, along with simultaneous monitoring of end‑tidal carbon dioxide and the entire cardiopulmonary resuscitation process. We determined regional cerebral oxygen saturation values and their relationship with return of spontaneous circulation and 30‑day survival, which had not been previously studied in such a large group of patients.

Introduction

In‑hospital cardiac arrest (IHCA) is an incident that can potentially affect any patient admitted to a hospital ward.¹ IHCA is associated with high mortality rates, and although the success rate of resuscitation and return of spontaneous circulation (ROSC) ranges from 36% to 54%, the probability of survival to hospital discharge or 30‑day survival is only 15% to 34%. In the longer term, unfavorable neurological outcomes (cerebral performance category [CPC] 1–2) occur in 10% to 50% of patients who survive CA.^2-5

According to the guidelines of the European Resuscitation Council (ERC) and the American Heart Association, to ensure high‑quality cardiopulmonary resuscitation (CPR), chest compression mechanics and end‑tidal carbon dioxide (ETCO₂) should be monitored. Brain oximetry has been highlighted as a technology that can provide noninvasive information on regional cerebral oxygen saturation (rSO₂) during CPR.^6,7 Near‑infrared spectroscopy (NIRS) uses electromagnetic waves with a length of 750–1000 nm. NIRS devices utilize the difference in near‑infrared light absorption by oxygenated and deoxygenated hemoglobin.^8-10 NIRS, in combination with other monitoring methods, such as capnography, can become a crucial part of advanced monitoring in cardiac arrest and postresuscitation care. The application of NIRS during CA focuses on the use of this technology to assess the quality of CPR, predict ROSC and favorable neurological outcomes after ROSC, and classify a patient as requiring extracorporeal cardiopulmonary resuscitation (ECPR).^10,11

In this study, we focused on using NIRS to predict ROSC and long‑term neurological outcomes. To date, there have been no studies showing the usefulness of using NIRS by rapid response teams (RRTs) in the context of monitoring rSO₂ during advanced life support (ALS). Our aim was to analyze the dynamics of rSO₂ during CA and to study the predictive value of rSO₂ in comparison with ETCO₂ monitoring in achieving ROSC and in the prognosis of long‑term neurological outcomes.

Patients and methods

The study was carried out from January 11, 2023 to January 31, 2024, at the University Hospital, Kraków, Poland. The study group consisted of patients who experienced IHCA. Ethical approval was obtained from the Kielce Bioethics Committee, Poland (11/2023) and from the University Hospital in Kraków (as of December 1, 2022). The RRT connected the rSO₂ and ETCO₂ monitoring devices to each patient during IHCA within no more than 5 minutes after recognizing CA. For ROSC patients, the measurement of rSO₂ was continued for 24 hours. Before starting the study, the RRT was trained on operating the device, properly applying electrodes for rSO₂, and correctly attaching the ETCO₂ sensor. The inclusion criteria comprised sudden CA in hospitalized patients, age over 18 years, and normothermia. The exclusion criteria were no connection of the rSO₂ and ETCO₂ sensor or its malfunction (electrode or device) during CPR, hospital admission during CPR, CA occurring in patients hospitalized in the intensive care unit, CA occurring in the operating theater, placement of the rSO₂ monitoring sensor later than within 5 minutes since the occurrence of CA, suspicion of internal or external bleeding, or palliative care.

Cardiopulmonary resuscitation and near‑infrared spectroscopy

The CPR procedure was carried out according to the ERC 2021 guidelines.² After the RRT received the call and arrived at the scene where the patient experienced IHCA, they took over CPR from the ward’s nursing‑medical team. The electrodes for rSO₂ monitoring were placed according to the manufacturer’s recommendations by a designated trained team member. rSO₂ and ETCO₂ values were recorded at 2‑second intervals using the MOC‑9 module. The rSO₂ monitoring lasted from the sensor connection (10 seconds were allocated for the device calibration) until CPR cessation or ROSC, and in the case of ROSC also for 24 hours (postresuscitation care). The rSO₂ reading was visible on the monitor but did not influence IHCA management. Ventilation was performed asynchronously after intubation or placement of a supraglottic device using a self‑inflating bag or a ventilator with a CPR function. The ETCO₂ sensor was placed behind the breathing filter. The entire course of in‑hospital CPR was documented according to the Utstein Resuscitation Registry Template for IHCA.¹²

To assess the change in the mean rSO₂ value during CPR, 3 time points were evaluated: 1) M^a indicating the mean value of rSO₂ from the first 2 minutes of resuscitation after connecting the sensor, 2) M^b indicating the mean value of rSO₂ from the middle 2 minutes of resuscitation, and 3) M^c indicating the mean value of rSO₂ of the last 2 minutes of resuscitation.

To assess the dynamics of rSO₂ changes and assume that ROSC occurred in a 2‑minute CPR loop, ∆rSO₂ was calculated. During rhythm evaluation (at the beginning of the 2‑minute loop), the rSO₂ value was read, and the next value was read at the end of that loop. The difference between these values was defined as ∆rSO₂.

Results

The total number of IHCAs during the study period was 152, of which simultaneous monitoring of ETCO₂ and rSO₂ was achieved in 121 cases. Seventeen patients were excluded from the study, with a total of 104 patients finally included. ROSC was achieved in 54 patients (51.9%), with 30‑day survival, and CPC 1 or 2 was reached in 9 patients (8.7%). The data are presented in Figure 1. Men constituted 63.5% of the study cohort, and its mean (SD) age was 68 (12) years. Upon hospital admission, 85.6% of the patients had a neurological status assessed as good or as moderate disability (CPC 1–2). The median (IQR) AACCI was 4 (2–4), indicating a moderate burden of the disease but increased risk of mortality. The median (IQR) duration of the hospital stay before the occurrence of CA was 3 (1–8) days, and 41 patients (39.4%) experienced CA within the first 48 hours. Data broken down by groups are presented in Table 1. The most common initial IHCA rhythms were nonshockable rhythms: pulseless electrical activity (n = 54; 51.9%) and asystole (n = 33; 31.7%). Ventricular fibrillation and pulseless ventricular tachycardia constituted 11.5% and 4.8%, respectively. The origin of CA was identified as cardiac (n = 36; 34.5%), respiratory (n = 24; 23.1%), and other (n = 44; 42.3%).

The mean (SD) value of rSO₂ during CPR for the patients with ROSC was 63.8% (7.4%), while for the patients without ROSC, it was 35.6% (5.6%). The ROSC group had higher parameter values than the non‑ROSC group (P <⁠0.001; Figure 2). The mean (SD) value of ETCO₂ was higher in the ROSC than in the non‑ROSC group (26 [4.9] mm Hg vs 17 [4.8] mm Hg; P <⁠0.001).

In the non‑ROSC group, the mean rSO₂ values remained low throughout the resuscitation period. The initial mean (SD) value (M^a) was 35.8% (7.4%), the intermediate value (M^b) was 35.7% (6%), and the final value (M^c) reached 34% (5.2%). No significant changes in the measured parameter were observed at the time points analyzed (M^a–M^b, P = 1; M^a–M^c, P = 0.13; and M^b–M^c, P = 0.054).

In the ROSC group, mean (SD) rSO₂ values increased throughout the course of resuscitation. M^a of rSO₂ was 60.9% (9.4%), with M^b increasing to 62% (8.5%), and M^c of 69.2% (5.9%) in the last 2 minutes. Statistical analysis showed an increase between the 2 time points (M^b–M^c, P <⁠0.001; M^a–M^c, P <⁠0.001; and M^a–M^b, P = 0.23). In the ROSC group, rSO₂ increased during resuscitation, while in the non‑ROSC group, despite ongoing resuscitation efforts, no significant change in rSO₂ values was observed. Differences between the ROSC and non‑ROSC groups were also demonstrated at the studied time points (P for interaction <⁠0.001; Figure 3).

The mean values of rSO₂ during the first 2 minutes of monitoring during CPR and throughout the entire resuscitation process were higher in the group of patients with an initially recorded shockable rhythm than in those with a nonshockable rhythm (P = 0.04). These differences were not observed for the mean values of ETCO₂ during the first 2 minutes and throughout resuscitation (P = 0.13 and P = 0.14, respectively). Our study demonstrated that in shockable rhythms, the mean rSO₂ throughout all CPR (P = 0.04) and from the initial 2 minutes of CPR (P = 0.04) was markedly higher. These differences were not demonstrated for ETCO₂ (P = 0.13 and P = 0.14). A shockable rhythm at the time of the first assessment was recorded in 17 patients. Although rSO₂ was significantly higher during CPR in these patients, the initial rhythm was not significant in predicting ROSC in the logistic regression analysis (Supplementary material, Table S1).

Association with return of spontaneous circulation

For rSO₂, an area under the curve (AUC) of 0.978 (95% CI, 0.956–0.999; P <⁠0.001) indicates a very good predictive capacity of rSO₂ associated with ROSC. The optimal cutoff point was determined to be 47.6% (sensitivity 94%, specificity 92%). For ETCO₂, the AUC of 0.815 (95% CI, 0.737–0.897; P <⁠0.001) indicates a good capacity of ETCO₂ to predict ROSC. The optimal cutoff point was determined to be 19.9 mm Hg (sensitivity 81.5%, specificity 70%; Supplementary material, Figure S1). The difference between the AUCs for the ROC curve (rSO₂-ETCO₂) was 0.162 (95% CI, 0.079–0.246; P <⁠0.001).

The dynamics of the increase in rSO₂ during the 2‑minute loop was evaluated by ΔrSO₂. It was shown that an increase in rSO₂ by 4 percentage points during the CPR loop may indicate ROSC (sensitivity 80.4%, specificity 83.2%; AUC for rSO₂, 0.875; 95% CI, 0.837–0.913; P <⁠0.001). In the logistic regression analysis that considered factors such as age, sex, initial heart rhythm (shockable and nonshockable), AACCI, rSO₂, and ETCO₂ in the first 2 minutes of CPR to predict ROSC, it was found that patient age (P = 0.03) and early values of rSO₂ in the first 2 minutes of resuscitation (P <⁠0.001) were significant in predicting ROSC (R² = 0.905), while sex, heart rhythm, AACCI, and ETCO₂ did not show any significant relationship (Table 2).

Table 2. Logistic regression, end point of return of spontaneous circulation

Variable	B	OR	95% CI	P value
Nagelkerke R2 =0.905 a Sex: 1, female; 2, male b Rythms: 1, nonshockable; 2, shockable Abbreviations: B, coefficient of regression; OR, odds ratio; others, see Figure 1
Age	0.218	1.244	1.021–1.517	0.03
Sexa	–1.145	0.318	0.018–4.369	0.37
Rhythmsb	–1.835	0.16	0.005–4.467	0.27
ACCI	–1.149	0.317	0.08–1.306	0.06
rSO2 first 2 minutes of CPR	0.46	1.584	1.221–2.045	<⁠0.001
ETCO2 first 2 minutes of CPR	0.12	1.127	0.907–1.4	0.28

The mean values of rSO₂ were evaluated throughout the CPR period. In the group of patients who achieved ROSC but did not survive 30 days (death <⁠30 days), the mean (SD) value of rSO₂ was 60.3% (5.9%). The group of patients who survived more than 30 days with CPC 3–4 (71.2% [2.7]) and the group of patients with 30‑day survival and CPC 1–2 (72.5%)³ showed a notable difference in comparison with the group of patients without 30‑day survival (P <⁠0.001; Figure 4; Supplementary material, Table S2). ROC curve analysis showed that lower rSO₂ values are associated with a higher risk of death (AUC, 0.992; 95% CI, 0.98–1). The end point of the analysis was survival for at least 30 days regardless of the CPC value. The cutoff point was determined to be at least 66% (sensitivity 92% and specificity 100%). If the mean rSO₂ value during CPR was greater than 66%, all patients in the study group survived 30 days.

24‑hour monitoring

The measurements were obtained from 51 patients, of whom 13 were pronounced dead before 24 hours. In 3 patients, 24‑hour rSO₂ measurement was impossible. The mean (SD) value of rSO₂ in the group of patients who died before 30 days (n = 35) was 63.7% (6.1%) (95% CI, 61.7–65.8), while in those who survived at least 30 days with CPC 3–4 (n = 7) it reached 73.6% (6%) (95% CI, 65.9–81.1). The highest mean (SD) value of rSO₂ of 77.2% (0.8%) (95% CI, 76.6–77.8) was recorded in the group of patients who survived at least 30 days with CPC 1–2. Statistical analysis showed that the difference between the CPC 3–4 group and the CPC 1–2 group was insignificant (P = 0.46), but there was a difference between patients who died within 30 days and those who survived with CPC 3–4 (P = 0.04), and those who survived with CPC 1–2 (P <⁠0.001; Figure 5).

Discussion

Our study confirms that both higher ETCO₂ and rSO₂ values are associated with ROSC in patients with IHCA. However, rSO₂ demonstrated greater predictability than ETCO₂. Monitoring rSO₂ during postresuscitation care may also be useful for long‑term neuroprognostication. Measurement of rSO₂ using NIRS can serve as a tool for monitoring during CPR and prognostication. However, available studies on this topic vary significantly in many aspects, such as clinical conditions, monitoring devices, or time and type (mean, median, ΔrSO₂, minimum or maximum) of readings considered. It should also be noted that the rSO₂ value can be affected by various confounders, such as skin pigmentation, improper electrode placement, arterial occlusion, hemoglobin concentration, bleeding, and craniocerebral injuries.^12,14,15 Considering the technical aspects, it should be noted that monitoring rSO₂ using NIRS technology may be difficult to access in most European hospitals due to the cost of purchasing the device and disposable electrodes.

A study by Schewe et al¹⁵ showed that mean (SD) rSO₂ values during resuscitation were lower in patients who did not achieve ROSC (31.6% [7.4%]) vs those who did (37.2% [7.7%]). In the ROSC patients (n = 3), an increase in these values was observed from 37.2% (7.7%) before ROSC to 52.3% (14.9%) at the time of ROSC, representing a relative increase of 40.6%. However, it should be noted that this was a study conducted with 10 patients and ROSC was achieved in 2 of them.¹⁵ In a meta‑analysis by Schnaubelt et al,¹⁶ mean (SD) values of rSO₂ were shown to be higher in the patients who achieved ROSC than in the non‑ROSC group (41% [12%] vs 30% [12%]; P = 0.009). ROSC was not observed when the mean rSO₂ remained below 26%.¹⁶ In a systematic review by Sanfilippo et al,¹⁷ the rSO₂ values were on average by 1.03 SD higher in the ROSC group than in the non‑ROSC group (95% CI, 1.39–0.67; P <⁠0.001). Our study confirmed these findings, but in a large group of patients and with resuscitation performed by a specialized team, which makes the outcomes more credible for the quality of the CPR performed.

Kämäräinen et al¹⁸ noted that rSO₂ values remained low even during high‑quality CPR in a group of 9 patients. We confirmed that the rSO₂ values remained low throughout the CPR period in the non‑ROSC group but increased in the ROSC group.

Takegawa et al¹⁹ examined the ability of rSO₂ to predict ROSC during CPR. ROC curves were used to assess the predictive capacity of various rSO₂ parameters. The analysis showed that the AUC for rSO₂ was the highest at 16 minutes, with AUC values of 0.65, 0.68, 0.71, 0.72, and 0.7 for the 4, 8, 12, 16, and 20 minute periods, respectively. In our study, it was assumed that the prediction of successful resuscitation should be based on initial values, as the median (IQR) duration of CPR for ROSC patients was 15 (10–20) minutes. The mean value of the first 2 minutes (1 CPR cycle) was considered. The AUC of the ROC curve for rSO₂ was 0.978, indicating a very good predictive ability. The optimal cutoff point was established at 47.6%, with sensitivity of 94% and specificity of 92%. If we think about predicting ROSC during CPR, we want to determine its chances right from the beginning. Therefore, only the factors that can be assessed during the first CPR cycle, such as age, sex, comorbidities, and the first recorded rhythm, should be considered. In the context of our study, the mean rSO₂ value of the initial period was also such a factor. Studies show that an initial shockable rhythm is associated with a higher likelihood of ROSC.² However, in our work, this was not a significant factor, likely due to the small patient group, as shockable rhythms account for only about 10% of IHCA cases.^2,6

In the meta‑analysis by Schnaubelt et al,¹⁶ ΔrSO₂ was evaluated as a ROSC indicator. A 7% change in ΔrSO₂ predicted ROSC with 100% sensitivity and 86% specificity. In our study, the optimal cutoff point was established at ΔrSO₂ equal to or above 4 percentage points (sensitivity 80.4%, specificity 83.2%). However, as mentioned earlier, rSO₂ values are significantly higher throughout the resuscitation period in the ROSC patients. An increase of 4 percentage points is relatively small from a clinical perspective. More research is needed to determine the specific value of rSO₂ above which it can be assumed that ROSC has occurred during a given CPR cycle, similar to ETCO₂, where a sudden increase to near normal values, that is, 35–45 mm Hg, may indicate ROSC.

Engel et al²⁰ demonstrated that predictive factors for ROSC included: a trend of the last 5 minutes of rSO₂ and ETCO₂, the Δ from the first to the last reading of rSO₂ and ETCO₂, the penultimate minute values of rSO₂ and ETCO₂, and the last‑minute values of rSO₂ and ETCO₂. The rSO₂ measurement was more effective than that of ETCO₂ for predicting ROSC. Our study also demonstrated that rSO₂ was a better predictor of ROSC than ETCO₂. Storm et al²¹ found significantly higher rSO₂ values on hospital admission in patients who experienced out‑of‑hospital CA with CPC 1–2, with mean rSO₂ of 56% and 68%, as compared with rSO₂ values in patients with CPC 3–4, which were 20% and 58%, respectively. However, other works did not show any difference in rSO₂ between the patients with CPC 1–2 (62%) and CPC 3–5 (58%).¹⁶ Our study also did not find any differences between the group of patients who survived at least 30 days with CPC 3–4 and those who survived at least 30 days with CPC 1–2 (P = 0.46). Chen et al²² developed a model to predict the risk of in‑hospital death in post‑CA patients, which included 11 independent predictors: age, malignancy, bicarbonate level, blood urea nitrogen, sodium level, heart rate, respiratory rate, body temperature, SpO₂, norepinephrine use, and lactate level. The nomogram demonstrated good discriminative ability in both the training set (AUC, 0.787; 95% CI, 0.753–0.821) and the test set (AUC, 0.801; 95% CI, 0.778–0.824). Studies evaluating the predictive value of rSO₂ in postresuscitation care within such models are needed.

NIRS technology in postresuscitation monitoring may become an additional element in neuroprognostication within a multimodal approach, encompassing examinations such as electroencephalography (EEG), N20 wave of somatosensory evoked potentials, neuron‑specific enolase (NSE), computed tomography (CT), magnetic resonance imaging (MRI), and clinical assessments, including pupillary and corneal reflexes.²³ However, more research is needed to determine the role of rSO₂ monitoring in postresuscitation care in the context of neuroprognostication.