CPR is attempted in approximately 30 000 people who experience an out-of-hospital cardiac arrest (OHCA) in England each year (Resuscitation Council UK (RCUK), 2017). OHCA is associated with poor outcomes. In the year 2015–2016, return of spontaneous circulation (ROSC) before arrival at hospital was achieved in 28% of patients in the UK but survival-to-hospital discharge was achieved in only 8% (National Audit Office (NAO) and NHS England (NHSE), 2017).
Early, high-quality CPR is associated with better outcomes (Meaney et al, 2013; NAO and NHSE, 2017). Perfusion monitoring during resuscitation gives an indication of the quality of CPR being performed (Meaney et al, 2013). End-tidal CO2 is used as an indicator of perfusion but reflects pulmonary rather than cerebral flow (Ahn et al, 2013).
Non-invasive cerebral oximetry has been used successfully to provide real-time information of cerebral perfusion during cardiac arrest in hospital (Kämäräinen et al, 2012; Parnia et al, 2012). Several recent studies have also investigated cerebral oximetry in patients who have experienced an OHCA but monitoring has typically begun after arrival in the emergency department, which means that important information of cerebral haemodynamics before arrival at hospital is missed (Box 1).
This review aims to identify studies where cerebral oximetry (rSO2) has been performed in the prehospital setting during OHCA and aims to answer the following questions:
Background
Cerebral oximetry uses near infrared spectroscopy (NIRS) to measure transcutaneous cerebral tissue oxygen saturation. This reflects perfusion in the cerebral microcirculation (Genbrugge et al, 2016). Most devices involve placing one or two monitoring sensors on the patient's forehead and they provide real-time, non-invasive readings of cerebral perfusion in the frontal lobes.
Cerebral oximetry has been used to provide continuous non-invasive monitoring of cerebral perfusion during surgery since the 1970s (Genbrugge et al, 2016). In recent years, it has been investigated for its usefulness during CPR with good reliability in the hospital setting (Kämäräinen et al, 2012; Parnia et al, 2012).
Positive correlations have been identified between rSO2 values on arrival at hospital and ROSC in patients who have experienced an OHCA (Ahn et al, 2013; Asim et al, 2014; Fukuda et al, 2014; Singer et al, 2017). Similarly, several large-scale studies have demonstrated improved neurological outcomes where rSO2 readings are higher on hospital arrival (Ito et al, 2014; Nishiyama et al, 2015a; 2015b).
To date, clinical trials have mostly been small, single-centre, non-randomised, unblinded observational trials, based on a convenience sample of patients.
A systematic review conducted in April 2015 included a meta-analysis of nine studies of patients with both in-hospital cardiac arrest (IHCA) and OHCA (Sanfilippo et al, 2015). In total, 315 patients were included across all studies, of whom 71.5% (n=225) had experienced OHCA, and 37.7% of (n=119) achieved ROSC.
The authors found that ROSC was significantly correlated with higher initial and mean rSO2 values overall (standardised mean difference (SMD) –0.79; 95% CI –1.29, –0.30; P=0.0002 and SMD –1.28; 95% CI –1.74, –0.83 [CI –1.74 to –0.83]; P<0.0001 respectively) but found conflicting data on the strength of correlation between rSO2 and CPR quality.
Similarly, no strong relationship was identified between rSO2 and neurological outcome. The authors concluded that larger-scale trials were required in order to identify meaningful associations. Any significant impact of cerebral oximetry in OHCA remains largely unknown.
Methods
A literature search was conducted to identify published clinical trials of cerebral oximetry in OHCA (Table 1). The PubMed and EMBASE databases were searched using specific search terms (Cerebral oximetry OR NIRS OR Near infrared spectroscopy) AND (Cardiac arrest OR resuscitation OR CPR) AND (pre-hospital OR out-of-hospital OR paramedic OR ambulance). The authors followed the approach set out in the PRISMA statement (Liberati et al, 2009) for conducting systematic reviews (Figure 1). The search was limited to studies in humans and included all articles published before June 2017. This returned 46 articles.
| PICOS criteria | |
|---|---|
| Population | Adult patients experiencing out-of-hospital cardiac arrest in whom resuscitation was attempted by emergency personnel in an out-of-hospital environment |
| Intervention | Intra-arrest cerebral oximetry monitoring |
| Comparator | No comparison with another monitoring tool |
| Outcomes | Return of spontaneous circulation, neurological survival, adverse event, feasibility |
| Study design | Prospective and retrospective clinical studies, systematic reviews |
Articles were then manually searched by abstract to identify those meeting the inclusion criteria. This produced 13 papers for review.
The full text of all 13 papers was reviewed. Those where cerebral oximetry monitoring was begun only on hospital arrival were excluded. This left five papers for inclusion.
Abstract and full text filtering was performed by both researchers independently in a blinded manner and results compared to ensure consensus of inclusion. All papers included were reviewed using the Critical Appraisal Skills Programme (CASP) (2018) framework.
Findings
Five studies that met the specified inclusion criteria were identified (Table 2). The first feasibility trial of cerebral oximetry monitoring during CPR in OHCA was published by Newman et al (2004). This was a small feasibility trial that enrolled 16 patients experiencing non-traumatic OHCA. Monitoring was performed by a trained physician responding to the scene using an INVOS 3000 COx device. Consistent rSO2 readings were not reliably obtained for any participant during CPR. The authors concluded that CPR in OHCA may not generate sufficient cerebral blood flow in the frontal lobes to be detected. They did, however, note that the monitoring device used could not record rSO2 readings below 15% or where the depth of cortical tissue exceeded 2 cm, which may have limited the findings in this small cohort of patients.
| Author | Year | Type of study | Sample size | Level of evidence* | Feasibility in OHCA | Useful monitor of CPR | Predictor of ROSC | Predictor of neurological survival |
|---|---|---|---|---|---|---|---|---|
| Genbrugge et al | 2015 | Prospective single-centre observational | 49 | 3 | Yes | – | Yes | Yes |
| Meex et al | 2013 | Prospective single-centre observational | 14 | 3 | Yes | – | – | – |
| Newman et al | 2004 | Prospective single-centre observational | 16 | 3 | No | – | – | – |
| Schewe et al | 2014 | Prospective single-centre observational | 10 | 3 | Yes | Yes | Yes | – |
| Storm et al | 2016 | Prospective observational pre-clinical trial | 29 | 3 | Yes | – | No | No |
Following this, Meex et al (2013) published a small prospective feasibility trial of cerebral tissue oxygen saturation (SctO2) monitoring during resuscitation in cardiac arrest. SctO2 monitoring was observed in 16 patients, nine of whom had experienced an OHCA. In these nine cases, monitoring was performed on scene by an emergency physician from arrival to termination of resuscitation or transport to hospital. Five patients who experienced an OHCA achieved ROSC. Reliable SctO2 values were achieved in 14 of the 16 patients (88%). It is not stated whether the two cases in which monitoring failed were IHCA or OHCA. No significant difference was found between mean initial SctO2 values for survivors and non-survivors, although the mean duration of CPR before monitoring for OHCA was started was 29±9 minutes compared to 16±6 minutes for IHCA. The authors report the initial rSO2 values, the highest values during CPR, the highest values during ROSC and the final values before transport or termination of resuscitation. The lowest values are not reported and neurological survival was not identified. These data cannot therefore inform prognostication but do demonstrate that it is possible to obtain reliable rSO2 readings in the prehospital environment.
Schewe et al (2014) conducted a similar feasibility trial where cerebral oximetry monitoring was carried out by a physician during the resuscitation of 10 patients who had experienced an OHCA. Three patients achieved ROSC. Though the numbers are too small to draw conclusions, ROSC was positively associated with a rise in rSO2 in all three patients, and a fall in rSO2 allowed early detection of re-arrest in two patients. Within the study protocol, physicians were given half an hour of training in the use of the monitoring device. Feasibility was defined by achieving >80% of total intended rSO2 recording time; 89.8% of the intended monitoring time was achieved, implying feasibility of use in the out-of-hospital environment. In addition, the authors identified that the monitor design could be small (305 x 108 x 130 mm), relatively lightweight (approximately 1 kg) and have sufficient battery longevity (up to 4 hours from fully charged) to make it practical for use in out-of-hospital settings.
Genbrugge et al (2015) conducted a single-centre prospective observational study of cerebral oximetry in adult non-traumatic OHCA. Monitoring was started on scene by a trained physician-nurse team and continued until termination of resuscitation or arrival at hospital. Highest, lowest and initial rSO2 values were recorded as well as the proportion of time during resuscitation where rSO2 was less than 30%. Data were collected between December 2011 and November 2013, and included 49 patients. Nineteen patients (39%) achieved ROSC and three (6%) survived with good neurological outcome at hospital discharge. No significant difference was found in initial or minimal rSO2 values recorded between those who achieved ROSC and those who did not. There was, however, a significant difference in the median rise in rSO2 from the initial value recorded, to the final value recorded 2 minutes before ROSC or before termination of resuscitation (16% vs 10%, P = 0.02) and mean rSO2 in the ROSC vs the non-ROSC group (39%±7% vs 31%±4%, P = 0.05). The proportion of time spent with rSO2 values <30% was also found to be important as those achieving ROSC spent a significantly lower proportion of the resuscitation period with low rSO2 values than those without ROSC (median=2% vs 25%, P < 0.03). Though numbers were small, it is worth noting that in the three patients who survived with good neurological outcome at hospital discharge, mean initial rSO2 was recorded at 44% (±26%), the mean rSO2 2 minutes before to ROSC was 54% (±4%) and the mean rise in rSO2 values during the resuscitation was 13% (±6%).
Only one study where cerebral oximetry monitoring was performed by paramedics was identified (Storm et al, 2016). This was a small preclinical trial including 29 patients who had experienced OHCA. In 23 patients, rSO2 monitoring was begun during CPR and, in the remaining six patients, it was applied within 2 minutes of achieving ROSC. All patients received advanced life support (ALS) and post-arrest care without deviation from normal protocols and an additional paramedic attended the scene to perform rSO2 monitoring. Initial rSO2 values were obtained and compared with neurological outcome at discharge from ICU and at 6 months after hospital discharge. Eighteen patients (62%) did not achieve ROSC. Of the 11 patients who did achieve ROSC, three were considered to have good neurological outcomes at ICU discharge and at 6 months' follow-up. The authors found that initial rSO2 values obtained immediately after arrival of emergency services personnel on scene were typically low and were not useful in predicting either ROSC or neurological outcome. Two patients with initial rSO2 values of only 22% survived with good neurological outcome. A rise in rSO2 values was observed in patients who achieved ROSC.
Discussion
Feasibility
Cerebral oximetry monitoring in OHCA was first trialled by Newman et al (2004) without success. Subsequent feasibility trials have, however, demonstrated improved capture, which may be related to improvements in the design of portable monitoring devices (Genbrugge et al, 2016). Meex et al (2013) found that reliable readings could be obtained in at least 88% of cases of OHCA within the first minute of applying monitoring. Similarly, Schewe et al (2014) found that reliable rSO2 could be obtained in 89.8% of the total intended monitoring time when used in the prehospital setting during OHCA, conferring feasibility. This suggests cerebral oximetry is possible in the prehospital environment.
Monitoring devices are available in small, lightweight designs with sufficient battery life, making them practical for prehospital use (Schewe et al, 2014). Monitoring has been shown to provide reliable values during transport by land or helicopter (Weatherall et al, 2012). The accuracy of cerebral oximetry may, however, be compromised by ambient light (Zaouter and Arbeid, 2010), which could be an important limiting factor in the prehospital environment, but it is not known if this applies to all monitoring devices. Authors in one study describe overcoming this by applying tape over the sensors to minimise interference (Meex et al, 2013).
Most trials involved rSO2 monitoring performed by physicians. Only one small preclinical trial where cerebral oximetry monitoring was carried out by paramedics was identified in this review (Storm et al, 2016). Larger scale prehospital trials are therefore required to confirm its feasibility for OHCA management by paramedics.
Monitoring CPR quality
Only one study identified in this review addressed the usefulness of rSO2 monitoring as a measure of CPR effectiveness during OHCA (Schewe et al, 2014). These authors suggest that rSO2 monitoring provided useful real-time assessment of the quality of resuscitation, which is in line with findings of previous authors where monitoring was applied on hospital arrival (Ahn et al, 2013). Further prehospital studies are needed to explore this area.
One recognised limitation is that the relationship between brain perfusion and the cardiac output generated by CPR may be affected by certain pathologies. For example, a massive pulmonary embolism may obstruct blood flow to the brain or mean only poorly oxygenated blood is provided. Since rSO2 monitors levels of oxygenated blood, this may mean low values are recorded even in cases of high-quality CPR. Other pathologies, such as hyperbilirubinaemia (Ahn et al, 2013), may result in false readings. Improving the quality of chest compressions in these cases would not result in improved rSO2 values. It is also not fully established how rSO2 values may be affected by elements of care such as inotrope administration. All values therefore need to be interpreted within the context of the clinical presentation and concurrent treatments provided.
Prognostication
Only two studies (Genbrugge et al, 2016; Storm et al, 2016) in this review addressed whether rSO2 values could be useful in prognostication (Table 3). The results of these demonstrated a positive correlation between rSO2 values and both rates of ROSC and good neurological outcome. Though these are low-level with small sample sizes, this does suggest rSO2 could be a useful measure in prognostication during OHCA. Exact figures to guide decisions are yet to be established.
| Author | rSO2 value required for ROSC | rSO2 value required for good neurological outcome |
|---|---|---|
| Genbrugge et al (2015) | - | ≥44% |
| Storm et al (2016) | ≥15% | ≥22% |
Though rare, ROSC has been recorded in patients who have experienced an OHCA with initial rSO2 values on hospital arrival of 15% (Asim et al, 2014; Storm et al, 2016) or less (Nishiyama et al, 2015a). Initial rSO2 values may therefore be a poor predictor of outcome. Genbrugge et al (2015) identified no significant difference between initial or lowest rSO2 values obtained between those achieving ROSC and those without ROSC as standalone figures. However, they found the overall trend (i.e. the rise in rSO2 values throughout the resuscitation) and the proportion of the resuscitation spent with low (<30%) rSO2 values were important in predicting ROSC. These were, however, all unblinded trials so it is not possible to establish whether monitoring information obtained during the resuscitation influenced decisions to terminate care, creating some bias.
Even when ROSC is achieved, the majority of patients experiencing OHCA do not survive to hospital discharge. This is largely because of irreversible ischaemic brain injury (Lemiale et al, 2013). Higher rSO2 values intra-arrest or at the point of ROSC seem to correlate with better neurological outcomes at 90 days, although the minimum rSO2 values required to obtain ROSC may be much lower than the values required for good neurological outcome (Ito et al, 2014; Nishiyama et al, 2015a). Good neurological outcome has been observed with low initial rSO2 values (<15%) but this is uncommon (Nishiyama et al, 2015b). It is likely that this needs to be interpreted within the overall patient presentation taking into account pathology and the length of time where cerebral blood flow may have been inadequate before CPR was started or during interruptions. Studies where rSO2 values are first recorded on hospital arrival need to be interpreted with caution as the quality of CPR may be temporarily compromised while the patient is being transferred to a hospital bed. These initial readings may therefore not reflect the overall quality of cerebral blood flow before arrival.
ROSC relies on spontaneous cardiac output. Only brainstem perfusion is required to support essential cardiac output. Cerebral oximetry monitors frontal lobe perfusion. It is therefore possible that ROSC may occur with brainstem perfusion but insufficient frontal lobe circulation and be sustained in the absence of good regional cerebral oximetry values. Therefore, cerebral oximetry may be a better predictor of neurological outcome than of ROSC. This may explain why ROSC is found with initial rSO2 values as low as 15% but good neurological outcome has been more reliably indicated by values of ≥40–44% (Ito et al, 2014; Nishiyama et al, 2015a; Genbrugge et al, 2016).
Using cerebral oximetry for prognostication and termination of resuscitation encourages decision-making based on the patient's anticipated quality of life. Ethically, it is difficult for a health professional to judge what the individual patient may consider an acceptable quality of life, which means this is still a complex decision that needs to be made on a case-by-case basis.
Conclusions
Cerebral oximetry may provide useful information in the context of OHCA. It has been used effectively in the prehospital environment but needs to be tested further for use by paramedics. Monitoring of cerebral blood flow allows real-time feedback of CPR quality but requires interpretation within the context of the clinical presentation. Cerebral oximetry may be used to assist in prognostication but cannot be used as a standalone measure of futility.
There is limited evidence of the usefulness of cerebral oximetry in monitoring CPR quality or guiding decisions around prognostication in the prehospital phase of cardiac arrest care. Furthermore, larger-scale studies are required to inform the use of cerebral oximetry in decision-making for this group of patients.