AWAreness during REsuscitation (AWARE)-II:

Supplementary Methods and Results Section

Methods

Study Rationale and Overview

It was necessary to establish the setup of this study in stages. In particular, this reflected the

practical challenges associated with attempts to apply and incorporate real-time brain

monitoring systems together with tests of visual and auditory awareness, without interrupting

active and ongoing cardiopulmonary resuscitation (CPR) efforts.

Staff needed to be trained and given time to learn how to carry and unpack a bag with dedicated

study equipment at the scene of cardiac arrest, all without interfering with the time-sensitive

efforts of emergency medical and nursing staff. Research staff were required to take five pieces

of equipment: 1) tablet computer, 2) wireless headphones, 3) brain oximetry 4) EEG and 5) End

tidal carbon dioxide (ETCO2) monitors. As staff needed to wait for an opportunity to access the

head of the bed without disturbing on-going resuscitation efforts, the time taken to set up study

equipment varied.

The first four items were used for this study (as reported in the main manuscript), while the

ETCO2 device was used to monitor the background quality of resuscitation, but not to address

study questions related to this manuscript. The AWARE-II study platform and the use of devices

was designed to support different research questions. Thus, the data from each of these

devices and the overall cohort addressed different research questions for different manuscripts.

The use of ETCO2 falls into this category.

research teams and has many practical limitations (such as packed rooms with very limited

space for maneuvering), staff at each site were asked to take time (typically the first 4-6 weeks,

after completion of their training) by focusing only on 1-2 pieces of equipment

(tablet/headphones and then rSO2). This enabled them to get used to the environment and to

maneuver themselves to the head of the bed, while also allowing for the clinical teams to get

used to their presence.

At first, the feasibility of the proposed methods was tested at a single site between 04/2013 to

05/2015. The study was then extended to a multi-site pilot study across five sites between

05/2015 and 05/2017, using the tablet, headphones and rSO 2. The EEG portion of the study

was introduced later across all sites in 05/2017 and the number of study sites was expanded to

a total of 25 across the United States, the United Kingdom, Bulgaria and Cyprus. Of these 25

hospitals, 21 provided study data (Table 1 in the main manuscript). Data collection started in

04/2013 and continued until 11/2021. However, between 03/2020 and 11/2021 activity was

severely curtailed or stopped across all study sites due to the repurposing of clinical research

staff to COVID-19 studies, as well as restrictions placed on research staff with respect to

attending cardiac arrests, in order to protect them from potentially becoming infected. Thus,

most of the data was collected from cardiac arrest events prior to 03/2020.

In order to explore any possible relationship that may exist between reports of consciousness

and cognitive activity with the quality of brain resuscitation, including cerebral oxygenation and

cortical electrical activity, the portable EEG monitor and portable brain oxygen monitor were

taken to the scene of cardiac arrest. Considering the time taken to arrive at the emergency

scene and unpack the study bag, and then set up the equipment in the prespecified order -

together with the fact that the time to collect data only lasted until the end of CPR - it was
accepted at the outset that in many cases there would not be enough time for all subjects to

have all instruments attached. The very short recruitment window of ~10-20 minutes (the period

of ongoing CPR until ROSC or until CPR had been terminated), meant that sites were allowed

to participate, knowing that many times dedicated research staff may not always be available to

attend a sudden cardiac arrest event within such a very short recruitment window (e.g. when

recruiting other patients into other studies).

Each site was also only provided one set of equipment, so when equipment was being used on

one patient, it could not be used to attend to other events. Furthermore, since EEG was only

introduced after 2017 and was the last of the four pieces of equipment to be set up, far fewer

subjects had EEG and they represented patients with the most prolonged cardiac arrests lasting

up to 106 minutes (median 31 [IQR 23-45] minutes).

Establishing Tests of Explicit Recall and Implicit Learning During Active

Cardiopulmonary Resuscitation

In order to examine for potential episodes of unconscious learning during CPR, specific auditory

stimuli were developed in accordance with studies that have demonstrated implicit

(unconscious) learning during anesthesia (discussed in the main manuscript). Staff pre-

recorded the names of three specific fruits (apple, pear banana) on an android application

installed on a tablet computer. Ten specific visual stimuli (pictures of people, animals,

monuments, and newspaper clippings) were also pre-recorded onto the application. The tablet

was connected to a headphone using Bluetooth and switched on when staff arrived at the scene

of cardiac arrest. The image remained until the tablet had been removed and switched off at the

end of CPR. Only one image was used per cardiac arrest event. This was to avoid making
survivors have to recall multiple images, which we considered to be too challenging in view of

likely debilitation after cardiac arrest.

To assess explicit recall of the images, survivors were asked to state whether they recalled

seeing an image while they had been unconscious. To test for implicit learning, if they didn’t

recall the image, they were informed that an image had been displayed while they had been

unconscious, and they were then shown all 10 images and asked to guess which image they

thought had been displayed. Similarly, to test for explicit recall of auditory stimuli, survivors were

asked to recall any specific audio cues they may have heard while they had been unconscious.

If they did not recall any audio cues and in order to test for implicit learning, they were informed

that three fruits had been mentioned to them while they had ben unconscious. They were then

asked to think of and guess the names of three fruits they thought may have been mentioned to

them.

Study Definitions and Outcome Measures

Using the Glasgow Outcome Scale (GOS), unfavorable outcomes were defined as GOS1-3.

Survival with a favorable neurological outcome was defined as a GOS score 4-5. The five GOS

categories are; 1: death, 2: vegetative state, 3: severe cerebral disability (neurological damage

and dependence on others but preserved consciousness) 4: moderate cerebral disability (but

with independent activities of daily life), 5: good cerebral performance (normal life with possible

minor psychological and/or neurological deficits).

EEG definitions were adapted from the American Clinical Neurophysiology Society (ACNS)

Standardized Critical Care EEG terminology (2021). Standard EEG terms referring to activity

observed across multiple cranial regions were adapted to reflect the limited scope of the 4-lead
EEG device used, which measured frontotemporal discharges only. For example, “generalized

rhythmic delta activity (GRDA)” was modified to “rhythmic delta activity (RDA)” and “generalized

periodic discharges (GPD)” to “localized periodic discharges (LPD).” All rhythms with

superimposed sharp waves were categorized as “epileptiform.” Background suppression (<

10uV), commonly referred to clinically as “isoelectric”, was labeled “suppression,” and implied a

lack of measurable cortical function. EEG was interpreted by at least one trained researcher

(E.R. and E.L.H) and one board-certified neurophysiologist (R.S. and D.A.), blinded to outcome.

EEG and rSO2 values were matched by exact timestamp.

Pilot testing Brain Monitoring Using Real-Time Cerebral Oximetry and

Electroencephalography During Active Cardiopulmonary Resuscitation.

Cerebral oximetry is a noninvasive measure of tissue oxygenation validated against invasive

arteriovenous oxygen tension that does not depend on pulsatile flow. One adhesive rSO 2 sensor

was applied to the lateral forehead prior to EEG. Sensors contained two light sources and

detectors, one penetrating ~3 cm into the frontal region of the brain reflecting venous

hemoglobin saturation and one collecting absorption from superficial tissue subtracted from

overall measurements64. Using an internal algorithm, devices calculated % rSO 2 using the

unique absorption spectra of oxy- and deoxyhemoglobin (detection range 0-100%, normal 60-

80%).

Prior to study use, EEG equipment was tested on healthy controls and alpha (seen awake with

eyes closed) and beta (seen awake with eyes open) waves visualized. The EEG sensors, which

comprised two temporal, two frontal, and two midline grounding electrodes attached to a single

adhesive strip were then placed across the forehead. Prior to application, the forehead was
wiped dry and sensors secured with a clean bandage if necessary. Each captured EEG image

on the screen displayed 4 sec of continuous EEG data. These were recorded on the EEG

device and later downloaded to the electronic data capture system. The EEG images were then

defined as “Interpretable”, (with discernable brain activity in at least one lead), “not interpretable”

contained exclusively artifactual findings, usually due to motion during CPR. The interpretable

EEG images were further classified as “near-normal” (alpha, beta, theta, delta activity),

“pathologic” (rhythmic delta activity, localized periodic discharges, burst suppression &

attenuation, epileptiform activity) or “absent” (suppression).

Research staff observed the monitor continuously during CPR for the purpose of identifying any

potential sensor or measurement defects. Although some rhythms classified as “near-normal”

should qualify as “normal” (e.g. frontal beta waves, theta/delta waves during sleep) we chose

this more conservative term given the limitations to capturing continuous data and/or across

multiple cranial regions associated with the frontotemporal device. However, the EEG and rSO2

values were not used to manage patients by clinical staff who did not have prior knowledge of

the potential utility of this technology during CPR.

In order to minimize data collection errors, research staff were trained and certified in study

procedures for collecting cerebral oximetry and EEG data, the completion of study case report

forms and data entry into REDCap, a web-based data entry system prior to study

commencement. Study protocols were reinforced during monthly teleconference meetings

conducted for the length of data collection. All rSO2 and EEG data were recorded and

automatically stored on the equipment without the need for further input from research staff,

thus minimizing operator bias errors. Brain oximetry artifact values were recognized by values

that were three standard deviations from the mean. Missing, or incomplete data, were defined

as any missing or incomplete values during each 4-second sampling period.

Cardiac arrest survivors were identified either through correspondence sent to the principal

investigator by cardiac arrest survivors, or through the near-death experience research

foundation (NDERF) website. For this, we are indebted to Dr. Jeffrey Long and Jody Long for

making the narratives provided by survivors available to the pubic and academics in a

deidentified manner.