Advertisement

Benign EEG for prognostication of favorable outcome after cardiac arrest: A reappraisal

Open AccessPublished:November 14, 2022DOI:https://doi.org/10.1016/j.resuscitation.2022.11.003

      Abstract

      Aim

      The current EEG role for prognostication after cardiac arrest (CA) essentially aims at reliably identifying patients with poor prognosis (”highly malignant” patterns, defined by Westhall et al. in 2014). Conversely, “benign EEGs”, defined by the absence of elements of “highly malignant” and “malignant” categories, has limited sensitivity in detecting good prognosis. We postulate that a less stringent “benign EEG” definition would improve sensitivity to detect patients with favorable outcomes.

      Methods

      Retrospectively assessing our registry of unconscious adults after CA (1.2018–8.2021), we scored EEGs within 72 h after CA using a modified “benign EEG” classification (allowing discontinuity, low-voltage, or reversed anterio-posterior amplitude development), versus Westhall’s “benign EEG” classification (not allowing the former items). We compared predictive performances towards good outcome (Cerebral Performance Category 1–2 at 3 months), using 2x2 tables (and binomial 95% confidence intervals) and proportions comparisons.

      Results

      Among 381 patients (mean age 61.9 ± 15.4 years, 104 (27.2%) females, 240 (62.9%) having cardiac origin), the modified “benign EEG” definition identified a higher number of patients with potential good outcome (252, 66%, vs 163, 43%). Sensitivity of the modified EEG definition was 0.97 (95% CI: 0.92–0.97) vs 0.71 (95% CI: 0.62–0.78) (p < 0.001). Positive predictive values (PPV) were 0.53 (95% CI: 0.46–0.59) versus 0.59 (95% CI: 0.51–0.67; p = 0.17). Similar statistics were observed at definite recording times, and for survivors.

      Discussion

      The modified “benign EEG” classification demonstrated a markedly higher sensitivity towards favorable outcome, with minor impact on PPV. Adaptation of “benign EEG” criteria may improve efficient identification of patients who may reach a good outcome.

      Keywords

      Abbreviations:

      ACNS (American Clinical Neurophysiology Society), CA (Cardiac Arrest), CHUV (Centre Hospitalier Universitaire Vaudois), CPC (Cerebral Performance Category), EEG (Electroencephalogram), cEEG (continuous Electroencephalogram), FOUR (Full Outline of UnResponsiveness), ICU (Intensive Care Unit), NSE (Neuron-Specific Enolase), PPV (Positive Predictive Value), ROC (Receiver Operating Characteristic), ROSC (Return of Spontaneous Circulations), SIRPID (Stimulus Induced Rhythmic, Periodic or Ictal Discharges), SSEP (Somatosensory Evoked Potentials)

      Introduction

      The role of electroencephalogram (EEG) to prognosticate unconscious patients after cardiac arrest (CA) was described more than 50 years ago
      • Binnie C.D.
      • Lloyd D.S.
      • Margerison J.H.
      • Maynard D.
      • Prior P.F.
      • Scott D.F.
      EEG prediction of outcome after resuscitation from cardiac or respiratory arrest.
      . Added to neurological examination, cortical somatosensory evoked potentials (SSEP), serum biomarkers (such as neuron-specific enolase, NSE) and neuroimaging, EEG represents one of the most widely used prognostic tools
      • Friberg H.
      • Cronberg T.
      • Dunser M.W.
      • Duranteau J.
      • Horn J.
      • Oddo M.
      Survey on current practices for neurological prognostication after cardiac arrest.
      . EEG interpretation in an intensive care unit (ICU) environment has been increasingly guided by recommendations of the American Clinical Neurophysiology Society (ACNS)
      • Hirsch L.J.
      • Fong M.W.K.
      • Leitinger M.
      • LaRoche S.M.
      • Beniczky S.
      • Abend N.S.
      • et al.
      American Clinical Neurophysiology Society's Standardized Critical Care EEG Terminology: 2021 Version.
      • Hirsch L.J.
      • LaRoche S.M.
      • Gaspard N.
      • Gerard E.
      • Svoronos A.
      • Herman S.T.
      • et al.
      American Clinical Neurophysiology Society's Standardized Critical Care EEG Terminology: 2012 version.
      , allowing a high generalizability of the findings. It has been shown that EEG interpreted according to these guidelines may orient on poor and good outcome
      • Rossetti A.O.
      • Tovar Quiroga D.F.
      • Juan E.
      • Novy J.
      • White R.D.
      • Ben-Hamouda N.
      • et al.
      Electroencephalography predicts poor and good outcomes after cardiac arrest: A Two-center study.
      • Ruijter B.J.
      • Tjepkema-Cloostermans M.C.
      • Tromp S.C.
      • van den Bergh W.M.
      • Foudraine N.A.
      • Kornips F.H.M.
      • et al.
      Early electroencephalography for outcome prediction of postanoxic coma: A prospective cohort study.
      . In particular, the definition of “highly malignant” patterns by Westhall and colleagues in 2014
      • Westhall E.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Horn J.
      • Ullen S.
      • et al.
      Standardized EEG interpretation accurately predicts prognosis after cardiac arrest.
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Friberg H.
      • et al.
      Interrater variability of EEG interpretation in comatose cardiac arrest patients.
      shows a high specificity towards poor prognosis. However, the definition of a “benign” EEG by exclusion of any “malignant” and “highly malignant” feature seems less robust, as its sensitivity is relatively low
      • Westhall E.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Horn J.
      • Ullen S.
      • et al.
      Standardized EEG interpretation accurately predicts prognosis after cardiac arrest.
      .
      We postulated that this definition may be too stringent, especially in terms of background continuity, amplitude, and anterio-posterior amplitude reversal (i.e., when, as opposed to physiological activity, the EEG background shows higher amplitude in the front than in the back), which can be influenced by sedation, and that timing may be critical to detect patients with favorable outcome. Our hypothesis was that the removal of the three aforementioned items would strengthen the sensitivity of benign EEG for identifying good outcomes. The aim of the present study is therefore to propose and test a modified “benign” EEG definition with features associated with good outcome, in order to increase sensitivity without major compromise on false positivity (positive predictive value, PPV). This would potentially impact prognostic evaluation of post-CA patients, allowing earlier and improved sensitivity of detection of those with likely favorable outcome.

      Methods

      Study design

      We analyzed our institutional (CHUV, Lausanne University Hospital, Switzerland), monocentric, prospectively acquired registry of consecutive adult patients treated after CA in our ICU, not dying within 24 h, and receiving at least one EEG recording within 72 h of CA, between January 2018 and August 2021 (44 months). The registry is approved by the local ethics committee (CER-VD), with consent waiver (procedures and treatments are part of standard care).

      ICU management and EEG analysis

      The protocol (as previously described
      • Rossetti A.O.
      • Tovar Quiroga D.F.
      • Juan E.
      • Novy J.
      • White R.D.
      • Ben-Hamouda N.
      • et al.
      Electroencephalography predicts poor and good outcomes after cardiac arrest: A Two-center study.
      • Oddo M.
      • Rossetti A.O.
      Early multimodal outcome prediction after cardiac arrest in patients treated with hypothermia.
      included temperature management during the first 24 h targeting 35–36 °C, with sedation by propofol (2–3 mg/kg/h) or midazolam (0.1–0.15 mg/kg/h), and analgesia by fentanyl (1.5mcg/kg/h), with a case-dependent adaptation. Neuromuscular blocking agents and noradrenaline (mean arterial pressure target: ≥65 mmHg) were administered if needed. Clinical and nonconvulsive (electrical) seizures were treated with valproate or levetiracetam as primary agents; in selected cases, propofol was added
      • Aicua Rapun I.
      • Novy J.
      • Solari D.
      • Oddo M.
      • Rossetti A.O.
      Early Lance-Adams syndrome after cardiac arrest: Prevalence, time to return to awareness, and outcome in a large cohort.
      .
      The in-house protocol foresaw repeated routine (20 min) video-EEGs. The first routine recording occurred at 12–36 h, repeated at 36–72 h (NicOne, Viasys Neurocare; after December 2020 XLtek, Natus Medical Inc., Middleton WI) using 21 electrodes (10–20 international system), with systematic reactivity testing by physicians or certified EEG technologists
      • Tsetsou S.
      • Novy J.
      • Oddo M.
      • Rossetti A.O.
      EEG reactivity to pain in comatose patients: Importance of the stimulus type.
      . In some cases, continuous EEG (cEEG) war used (according to changing internal guidelines during definite time periods). Median nerve SSEP were recorded after 24 h following CA; serum NSE was collected at 24 and 48 h. Clinical examination (Glasgow Coma Scale, FOUR -Full Outline of UnResponsiveness- score) was performed routinely at 72 hours, off sedation, by certified neurologists.
      During patients’ recruitment, life-sustaining-therapy withdrawal was considered, per house-internal protocol based on previous studies, in the presence of at least two reliable unfavorable prognostic factors (assessed off sedation and at 72 h following CA), such as absence of pupillary or corneal reflexes, treatment-resistant myoclonus / status epilepticus, unreactive EEG background in normothermia and off-sedation, bilateral absence of cortical somatosensory-evoked potentials
      • Nolan J.P.
      • Sandroni C.
      • Bottiger B.W.
      • Cariou A.
      • Cronberg T.
      • Friberg H.
      • et al.
      European Resuscitation Council and European Society of Intensive Care Medicine Guidelines 2021: Post-resuscitation care.
      ; brain MRI and serum NSE (threshold: 75 µg/l) represented additional supporting criteria
      • Rossetti A.O.
      • Tovar Quiroga D.F.
      • Juan E.
      • Novy J.
      • White R.D.
      • Ben-Hamouda N.
      • et al.
      Electroencephalography predicts poor and good outcomes after cardiac arrest: A Two-center study.
      • Oddo M.
      • Rossetti A.O.
      Early multimodal outcome prediction after cardiac arrest in patients treated with hypothermia.
      • Rossetti A.O.
      • Oddo M.
      • Logroscino G.
      • Kaplan P.W.
      Prognostication after cardiac arrest and hypothermia: a prospective study.
      . Decisions of withdrawal of intensive treatment were multi-disciplinarily, including discussion with relatives
      • Rossetti A.O.
      • Tovar Quiroga D.F.
      • Juan E.
      • Novy J.
      • White R.D.
      • Ben-Hamouda N.
      • et al.
      Electroencephalography predicts poor and good outcomes after cardiac arrest: A Two-center study.
      • Oddo M.
      • Rossetti A.O.
      Early multimodal outcome prediction after cardiac arrest in patients treated with hypothermia.
      .
      EEGs were prospectively interpreted by two authors (JN, AOR) and entered in the registry according to ACNS recommendations, before knowing the patient’s outcome. For those patients, the registry was prospectively scored using the time for reactivity testing corresponding to the routine recordings (12–36 hours, then 36–72 hours). Each recording had prospectively been assigned to the three prognostic Westhall’s categories
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Kjaer T.W.
      • Horn J.
      • et al.
      Electroencephalography (EEG) for neurological prognostication after cardiac arrest and targeted temperature management; rationale and study design.
      ; continuity, reactivity, amplitude (categorized as normal, low or suppressed), anterio-posterior gradient, and occurrence of repetitive epileptiform discharges were also entered separately. For this study, HF and AOR re-evaluated all tracings with pre-specified modified criteria of good prognosis (Fig. 1), blinded for outcome and cardiac arrest details. We focused on following elements: background continuity, anterio-posterior gradient, reactivity
      • Tsetsou S.
      • Novy J.
      • Oddo M.
      • Rossetti A.O.
      EEG reactivity to pain in comatose patients: Importance of the stimulus type.
      (excluding stimulus-induces rhythmic, periodic or ictal discharges (SIRPIDs)
      • Admiraal M.M.
      • van Rootselaar A.F.
      • Horn J.
      Electroencephalographic reactivity testing in unconscious patients: a systematic review of methods and definitions.
      , and occurrence of repetitive epileptiform elements (spikes, spike-waves, sharp waves, including triphasic appearance). The registry, with full EEG reports, as well as original and EEG tracings were accessed for clarification.
      Figure thumbnail gr1
      Fig. 1Benign EEG «Westhall criteria» and « Modified criteria » for good outcome. The difference lies in the second items group of Westhall criteria (in italic).
      Information on temperature and sedation during EEG recordings (and reactivity testing) was retrieved from the registry, as were results of SSEP, peak serum NSE, and FOUR scores. No tests were specifically performed for the means of this study.

      Outcomes

      The outcome was prospectively assessed, for clinical follow-up, using Cerebral Performance Categories (CPC
      • Booth C.M.
      • Boone R.H.
      • Tomlinson G.
      • Detsky A.S.
      Is this patient dead, vegetative, or severely neurologically impaired? Assessing outcome for comatose survivors of cardiac arrest.
      at 3 months by semi-structured interviews with patients, relatives, or primary care physicians, performed by a blinded investigator; good outcome was defined as CPC 1–2
      • Grossestreuer A.V.
      • Abella B.S.
      • Sheak K.R.
      • Cinousis M.J.
      • Perman S.M.
      • Leary M.
      • et al.
      Inter-rater reliability of post-arrest cerebral performance category (CPC) scores.
      ). Patients for whom outcome could not be obtained were excluded.

      Statistics

      Clinical variables were compared with Fisher, chi2 or 2-sided t-tests, as needed. Predictive performances of “Westhall” and the modified criteria for “benign EEG” for good outcomes (CPC 1–2) and awakening with contact (CPC 1–3) at 3 months were calculated, using 95% binomial confidence intervals. Performances of the two approaches were assessed by comparisons of proportions, and comparisons of areas under the Receiver Operating Characteristic (ROC) curves. Results with a p-value < 0.05 were considered statistically significant. Calculations were performed using Stata version 16 (College Station, TX).

      Results

      During the study period of 44 months, 381 patients were included in our registry (suppl. flowchart); their characteristics are summarized in Table 1. Of them, 363 (95%) had an early EEG within the first 24–36 h and 313 (82%) at 36–72 h (Table 2); the majority (292 patients, 77%) had both recordings. The lower number of recordings performed at 72 h was due to patient deceasing or regaining consciousness during this period of time. A continuous EEG (cEEG) was recorded in 32 patients (8%). We conducted the data analysis for both early and late EEG, as well as for “any EEG”, meaning a benign EEG occurring at any time-point.
      Table 1Patients characteristicsROSC = return of spontaneous circulation, SSEP = somatosensory evoked potentials, CPC = cerebral performance category scale, SD = standard deviation.
      All patients

      381
      Benign EEG after Westhall (15) 163Modified benign EEG 252p (test)
      Age, years61.9 (SD:15.4)61.6 (SD: 16.0)62.4 (15.2)0.596 (t)
      Female sex104 (27.2.%)45 (27.6%)63 (25.0%)0.554 (Chi2)
      Shockable cardiac rhythm173 (45.4%)65 (39.8%)116 (46.0%)0.217 (Chi2)
      Cardiac origin240 (62.9%)121 (75%)181 (72%)0.590 (Chi2)
      Time to ROSC (minutes, median)17 (IQR:8–27)15 (IQR: 8–28)17 (IQR: 8–27)0.891 (U)
      Bilaterally absent SSEP (tested n = 340)87/340 (25.5%)1/143 (0.7%)5/220 (2.2%)0.501 (Fisher)
      Bilaterally absent pupillary reflex (tested n = 381)90 (23.6%)9 (5.5%)18 (7.1%)0.549 (Fisher)
      CPC 1101 (24.3%)45 (27.6%)56 (22.2%)0.317 (Fisher)
      CPC 2129 (31.0%)52 (31.9%)77 (30.5%)
      CPC 391 (21.9%)37 (22.7%)54 (21.4%)
      CPC 41 (0.2%)01 (0.4%)
      CPC 593 (22.4%)29 (17.7%)64 (25.4%)
      Table 2Details regarding EEG: early (12–36 h after CA, EEG1) and late (36–72 h after CA, EEG2). SD: standard deviation; other sedation: ketamine, thiopental.
      Total patients n = 381EEG 1 (n = 363) 95%EEG 2 (n = 313) 82%
      Mean timing of recording (h)19.0 (SD: 5.75)51.1 (SD: 9.5)
      Mean temperature (°C)35.6 (SD: 0.8)36.7 (SD: 0.6)
      Propofol (%)221 (61%)70 (22%)
      Midazolam (%)152 (42%)53 (17%)
      Other sedation (%)3 (0.8%)5 (1,5%)
      The modified definition of “benign EEG” allowed identifying a greater number of patients with a benign recording as compared to the original one (252 (66%) versus 163 (43%) of 381); the distribution of clinical variables was however comparable in both groups of “benign EEGs” (Table 1). Sensitivities and positive predictive values for good functional outcome (CPC 1–2) and awakening reaching contact to the environment (CPC 1–3) are illustrated in Table 3, Table 4. Comparisons of these performances showed a marked improvement of sensitivities at any time-point and for each outcome, paralleling a slight, statistically non-significant of PPV reduction (Table 5). Comparing ROC curves, the modified definition showed higher areas towards survival (0.796 vs 0.757, p = 0.066) and good outcome (0.746 vs 0.723, p = 0.295), in both cases however not reaching statistical significance.
      Table 3Original definition (Westhall): Sensitivity and PPV (positive predictive value) (95% confidence interval) of good outcome and awakening with contact. CPC = cerebral performance category scale.
      EEGTotal scoredTrue positiveFalse positiveTrue negativeFalse negativeSensitivityPPV
      CPC 1–2 (good outcome)Early3626627205640.51 (0.42–0.59)0.71 (0.61–0.80)
      Late3137455155290.72 (0.62–0.80)0.57 (0.48–0.66)
      Any3819766179390.71 (0.62–0.78)0.59 (0.51–0.67)
      CPC 1–3 (survival)Early36278151561130.41 (0.34–0.49)0.84 (0.75–0.91)
      Late31310722135490.68 (0.61–0.76)0.83 (0.75–0.89)
      Any38113429153650.67 (0.60–0.73)0.82 (0.75–0.67)
      Table 4Modified definition; sensitivity and PPV (positive predictive value) (95% confidence interval) of good outcome and awakening with contact. CPC = cerebral performance category scale.
      EEGTotal scoredTrue positiveFalse positiveTrue negativeFalse negativeSensitivityPPV
      CPC 1–2 (good outcome)Early36212079153100.92 (0.86–0.96)0.60 (0.53–0.67)
      Late313969211870.93 (0.86–0.97)0.51 (0.43–0.58)
      Any38113311912630.97 (0.93–0.99)0.53 (0.46–0.59)
      CPC 1–3 (survival)Early36216039132310.84 (0.78–0.89)0.80 (0.74–0.86)
      Late31314345112130.91 (0.86–0.95)0.76 (0.69–0.82)
      Any38118864118110.94 (0.92–0.97)0.74 (0.68–0.78)
      Table 5Comparisons (P-values) of proportions of performances between the two definitions of benign EEG. CPC = cerebral performance category scale.
      CPC 1–2 (good outcome)CPC 1–3 (awake)
      SensitivityPPVSensitivityPPV
      Early<0.0010.077<0.0010.477
      Late<0.0010.269<0.0010.140
      Any<0.0010.178<0.0010.070

      Discussion

      This analysis shows that a comprehensive definition of “benign EEG” not including the criteria of amplitude, anterio-posterior amplitude gradient, and allowing discontinuity of the background, appears to be more sensitive to identify post CA patients with favorable prognosis than the original definition
      • Westhall E.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Horn J.
      • Ullen S.
      • et al.
      Standardized EEG interpretation accurately predicts prognosis after cardiac arrest.
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Friberg H.
      • et al.
      Interrater variability of EEG interpretation in comatose cardiac arrest patients.
      .
      Identification of patients with favorable outcome after CA has received increasing attention in the last years
      • Sandroni C.
      • Nolan J.P.
      • Andersen L.W.
      • Bottiger B.W.
      • Cariou A.
      • Cronberg T.
      • et al.
      ERC-ESICM guidelines on temperature control after cardiac arrest in adults.
      • Cronberg T.
      • Greer D.M.
      • Lilja G.
      • Moulaert V.
      • Swindell P.
      • Rossetti A.O.
      Brain injury after cardiac arrest: from prognostication of comatose patients to rehabilitation.
      , but is still scarcely mentioned in current prognostic recommendations
      • Nolan J.P.
      • Sandroni C.
      • Bottiger B.W.
      • Cariou A.
      • Cronberg T.
      • Friberg H.
      • et al.
      European Resuscitation Council and European Society of Intensive Care Medicine Guidelines 2021: Post-resuscitation care.
      . Recent studies
      • Duez C.H.V.
      • Johnsen B.
      • Ebbesen M.Q.
      • Kvaloy M.B.
      • Grejs A.M.
      • Jeppesen A.N.
      • et al.
      Post resuscitation prognostication by EEG in 24 vs 48h of targeted temperature management.
      • Admiraal M.M.
      • van Rootselaar A.F.
      • Hofmeijer J.
      • Hoedemaekers C.W.E.
      • van Kaam C.R.
      • Keijzer H.M.
      • et al.
      Electroencephalographic reactivity as predictor of neurological outcome in postanoxic coma: A multicenter prospective cohort study.
      already highlighted the importance of background reactivity. However, our goal in the present work was to provide an improved definition of benign EEG using several criteria, not just reactivity, with increased sensitivity (48% in the original study, however on a very limited number of patients with good outcome)
      • Westhall E.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Horn J.
      • Ullen S.
      • et al.
      Standardized EEG interpretation accurately predicts prognosis after cardiac arrest.
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Friberg H.
      • et al.
      Interrater variability of EEG interpretation in comatose cardiac arrest patients.
      , and similar PPV; higher sensitivities lead to non-significantly higher ROC areas.
      Our cohort appears relatively similar to those described in other studies, including the original description
      • Westhall E.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Horn J.
      • Ullen S.
      • et al.
      Standardized EEG interpretation accurately predicts prognosis after cardiac arrest.
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Friberg H.
      • et al.
      Interrater variability of EEG interpretation in comatose cardiac arrest patients.
      : mean age 67 years (62 in ours), 70% male (73%), 72% shockable rhythm (reflecting the targeted recruitment towards this rhythm in the TTM trial; 45% in ours), median time to ROSC 30 minutes (17 minutes). The higher time to ROSC in the original cohort could reflect that some TTM trial centers possibly performed EEG more frequently in unclear prognostic situations, thus somewhat concentrating poor prognosis patients, which would be in line with the worse outcome: (24% CPC 1–2, clearly lower than most CA cohorts, versus 55% in ours). Timing of EEG recording, reactivity testing as well as sedation did not differ between the cohorts.
      It could be argued that sensitivity is not more important than specificity to identify subjects with potential good outcome, as active care would not be discontinued in the absence of an isolated benign EEG. Identification of good outcome, however, practically relies on maximizing sensitivity (minimizing false negatives, or in other words not missing patients with good outcome), rather than specificity
      • Sandroni C.
      • Nolan J.P.
      • Andersen L.W.
      • Bottiger B.W.
      • Cariou A.
      • Cronberg T.
      • et al.
      ERC-ESICM guidelines on temperature control after cardiac arrest in adults.
      • Rossetti A.O.
      • Rabinstein A.A.
      • Oddo M.
      Neurological prognostication of outcome in patients in coma after cardiac arrest.
      . The latter does represent a key element of poor outcome prognostication, where minimizing false positives is paramount
      • Rossetti A.O.
      • Rabinstein A.A.
      • Oddo M.
      Neurological prognostication of outcome in patients in coma after cardiac arrest.
      ; however, identifying patients with favorable prognosis may not be efficient using high specificity (e.g., serum NSE, where 65% of good outcome patients were missed applying highly specific thresholds
      • Moseby-Knappe M.
      • Mattsson-Carlgren N.
      • Stammet P.
      • Backman S.
      • Blennow K.
      • Dankiewicz J.
      • et al.
      Serum markers of brain injury can predict good neurological outcome after out-of-hospital cardiac arrest.
      ).
      We thus opted to optimize sensitivity (which is independent from the prevalence of the considered outcome in the studied population, and includes false negatives in the denominator) rather than PPV; this, including false positives in the denominator, was only slightly and non-significantly decreased with the modified definition, and remains relatively low for both “benign EEG” approaches, especially for identification of CPC 1–2. In case of false positives for good outcome, patient’s evolution or additional prognostic tests
      • Sandroni C.
      • D'Arrigo S.
      • Cacciola S.
      • Hoedemaekers C.W.E.
      • Westhall E.
      • Kamps M.J.A.
      • et al.
      Prediction of good neurological outcome in comatose survivors of cardiac arrest: a systematic review.
      would likely contribute to correct the overestimation.
      Comparison of clinical characteristics of patients selected according to the original or the modified «benign EEG» definition did not show any relevant difference, suggesting that these two populations are comparable, and supporting that the modified “benign EEG” definition allows a more efficient identification of subjects with favorable outcome.
      We conducted our analyses on two different time-points (12–36 hours, 36–72 hours), as well as together (“any EEG”), for our initial hypothesis was that early and late recordings might show different performances, with better accuracy at the early time-point
      • Rossetti A.O.
      • Tovar Quiroga D.F.
      • Juan E.
      • Novy J.
      • White R.D.
      • Ben-Hamouda N.
      • et al.
      Electroencephalography predicts poor and good outcomes after cardiac arrest: A Two-center study.
      • Ruijter B.J.
      • Tjepkema-Cloostermans M.C.
      • Tromp S.C.
      • van den Bergh W.M.
      • Foudraine N.A.
      • Kornips F.H.M.
      • et al.
      Early electroencephalography for outcome prediction of postanoxic coma: A prospective cohort study.
      . Indeed, sensitivities of the original definition between early or late EEG showed a striking difference (Table 3), whereas they remained relatively stable over the two different time-points for the modified version (Table 4). This may be the consequence of EEG amplitude reduction and background discontinuity during sedation. It can be conversely inferred that sedation (mostly present early after resuscitation) did not bias EEG prediction using the modified criteria, in line with recent data at normally prescribed doses
      • Ruijter B.J.
      • Tjepkema-Cloostermans M.C.
      • Tromp S.C.
      • van den Bergh W.M.
      • Foudraine N.A.
      • Kornips F.H.M.
      • et al.
      Early electroencephalography for outcome prediction of postanoxic coma: A prospective cohort study.
      . This suggests that not only the modified definition is more sensitive, but also that it is more robust over time, including at an early stage of prognostication. Indeed, the modified «benign EEG» definition was based on the assumption that the items of background continuity, amplitude, and reversal of anterior-posterior amplitude gradient were too easily impacted by sedation and without robust evidence upon their description
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Friberg H.
      • et al.
      Interrater variability of EEG interpretation in comatose cardiac arrest patients.
      .
      In the 5 patients with conflicting prognostic information having a “benign EEG” (new definition) but bilaterally absent SSEPs, 1 reached CPC 3, while the other 4 died, underscoring that discordant prognostication warrants watchful waiting.
      The inclusion of background discontinuity (attenuation < 50%) as a malignant item in the original classification was described in the original analysis
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Friberg H.
      • et al.
      Interrater variability of EEG interpretation in comatose cardiac arrest patients.
      to be responsible for several false positive predictions, as opposed to burst-suppression (attenuation > 50%), which has very high specificity to poor prognosis. Other studies
      • Ruijter B.J.
      • Tjepkema-Cloostermans M.C.
      • Tromp S.C.
      • van den Bergh W.M.
      • Foudraine N.A.
      • Kornips F.H.M.
      • et al.
      Early electroencephalography for outcome prediction of postanoxic coma: A prospective cohort study.
      • Sivaraju A.
      • Gilmore E.J.
      • Wira C.R.
      • Stevens A.
      • Rampal N.
      • Moeller J.J.
      • et al.
      Prognostication of post-cardiac arrest coma: early clinical and electroencephalographic predictors of outcome.
      • Willems L.M.
      • Trienekens F.
      • Knake S.
      • Beuchat I.
      • Rosenow F.
      • Schieffer B.
      • et al.
      EEG patterns and their correlations with short- and long-term mortality in patients with hypoxic encephalopathy.
      also found that discontinuous background can be associated with good outcome. Normal background voltage has been associated with good prognosis
      • Ruijter B.J.
      • Tjepkema-Cloostermans M.C.
      • Tromp S.C.
      • van den Bergh W.M.
      • Foudraine N.A.
      • Kornips F.H.M.
      • et al.
      Early electroencephalography for outcome prediction of postanoxic coma: A prospective cohort study.
      • Sivaraju A.
      • Gilmore E.J.
      • Wira C.R.
      • Stevens A.
      • Rampal N.
      • Moeller J.J.
      • et al.
      Prognostication of post-cardiac arrest coma: early clinical and electroencephalographic predictors of outcome.
      , but the relationship of early low-voltage with poor outcome is inconstant
      • Sivaraju A.
      • Gilmore E.J.
      • Wira C.R.
      • Stevens A.
      • Rampal N.
      • Moeller J.J.
      • et al.
      Prognostication of post-cardiac arrest coma: early clinical and electroencephalographic predictors of outcome.
      , especially in the first 24 hours post CA. The limited predictive value of low voltage was also highlighted in the original analysis as responsible for false prediction of negative outcomes
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Friberg H.
      • et al.
      Interrater variability of EEG interpretation in comatose cardiac arrest patients.
      . Recent work indeed suggests that, especially if at least nearly continuous, low voltage background may be related to favorable outcome
      • Scarpino M.
      • Carrai R.
      • Lolli F.
      • Lanzo G.
      • Spalletti M.
      • Valzania F.
      • et al.
      Neurophysiology for predicting good and poor neurological outcome at 12 and 72 h after cardiac arrest: The ProNeCA multicentre prospective study.
      . Finally, to our knowledge, there is no validation of the predictive value of the anterior-posterior amplitude development; its inclusion in the original criteria drew a parallel with the alpha coma pattern, which also includes lack of reactivity
      • Westhall E.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Horn J.
      • Ullen S.
      • et al.
      Standardized EEG interpretation accurately predicts prognosis after cardiac arrest.
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Kjaer T.W.
      • Horn J.
      • et al.
      Electroencephalography (EEG) for neurological prognostication after cardiac arrest and targeted temperature management; rationale and study design.
      • Kaplan P.W.
      • Genoud D.
      • Ho T.W.
      • Jallon P.
      Etiology, neurologic correlations, and prognosis in alpha coma.
      .
      The use of positive items (such as reactivity) to define a “benign EEG” instead of merely the absence of items associated to bad prognosis, which probably overstates the value of voltage development and low-voltage background, as had already been observed
      • Backman S.
      • Cronberg T.
      • Friberg H.
      • Ullen S.
      • Horn J.
      • Kjaergaard J.
      • et al.
      Highly malignant routine EEG predicts poor prognosis after cardiac arrest in the Target Temperature Management trial.
      . The elements of discontinuity (burst-suppression) and background voltage remain however of unchanged importance for the “highly malignant” category
      • Westhall E.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Horn J.
      • Ullen S.
      • et al.
      Standardized EEG interpretation accurately predicts prognosis after cardiac arrest.
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Friberg H.
      • et al.
      Interrater variability of EEG interpretation in comatose cardiac arrest patients.
      , which has been convincingly validated
      • Rossetti A.O.
      • Tovar Quiroga D.F.
      • Juan E.
      • Novy J.
      • White R.D.
      • Ben-Hamouda N.
      • et al.
      Electroencephalography predicts poor and good outcomes after cardiac arrest: A Two-center study.
      • Beuchat I.
      • Solari D.
      • Novy J.
      • Oddo M.
      • Rossetti A.O.
      Standardized EEG interpretation in patients after cardiac arrest: Correlation with other prognostic predictors.
      • Bongiovanni F.
      • Romagnosi F.
      • Barbella G.
      • Di Rocco A.
      • Rossetti A.O.
      • Taccone F.S.
      • et al.
      Standardized EEG analysis to reduce the uncertainty of outcome prognostication after cardiac arrest.
      .
      Our results should be interpreted in the light of limitations. They are based on a single cohort, albeit sizeable and with characteristics similar to others
      • Rossetti A.O.
      • Tovar Quiroga D.F.
      • Juan E.
      • Novy J.
      • White R.D.
      • Ben-Hamouda N.
      • et al.
      Electroencephalography predicts poor and good outcomes after cardiac arrest: A Two-center study.
      • Ruijter B.J.
      • Tjepkema-Cloostermans M.C.
      • Tromp S.C.
      • van den Bergh W.M.
      • Foudraine N.A.
      • Kornips F.H.M.
      • et al.
      Early electroencephalography for outcome prediction of postanoxic coma: A prospective cohort study.
      • Sivaraju A.
      • Gilmore E.J.
      • Wira C.R.
      • Stevens A.
      • Rampal N.
      • Moeller J.J.
      • et al.
      Prognostication of post-cardiac arrest coma: early clinical and electroencephalographic predictors of outcome.
      • Elmer J.
      • Coppler P.J.
      • Solanki P.
      • Westover M.B.
      • Struck A.F.
      • Baldwin M.E.
      • et al.
      Sensitivity of Continuous Electroencephalography to Detect Ictal Activity After Cardiac Arrest.
      ; this suggests a good generalizability (note that ethnic information is not included in our registry, which is representative of a cohort of White, European individuals).The limited inter-rater reliability for the evaluation of reactivity on EEG in post-CA patients should be acknowledged
      • Westhall E.
      • Rosen I.
      • Rossetti A.O.
      • van Rootselaar A.F.
      • Wesenberg Kjaer T.
      • Friberg H.
      • et al.
      Interrater variability of EEG interpretation in comatose cardiac arrest patients.
      . However, reactivity has been independently related to favorable outcome in this setting
      • Admiraal M.M.
      • Horn J.
      • Hofmeijer J.
      • Hoedemaekers C.W.E.
      • van Kaam C.R.
      • Keijzer H.M.
      • et al.
      EEG reactivity testing for prediction of good outcome in patients after cardiac arrest.
      ; it was tested uniformly across patients
      • Tsetsou S.
      • Novy J.
      • Oddo M.
      • Rossetti A.O.
      EEG reactivity to pain in comatose patients: Importance of the stimulus type.
      and interpreted by two experienced authors (AOR, JN) working together for many years, supporting internal validity. It is possible that some EEG features, such as reactivity, may have influenced patients’ management and lead to a self-fulfilling prophecy. This seems however less probable than in studies analyzing poor outcome, as to our knowledge there are no widely recognized multimodal recommendations to forecast good prognosis (and there is still no protocol in this sense at our center). Of relevance, EEG scorings were conducted without knowledge of prognosis, and inspection of the raw EEG recordings was performed on selected cases to address unclear reports. We basically scored the amplitude regarding the item of anterior-posterior gradient, not the frequency. Unfortunately, our registry does not contain details on death cause. The retrospective design does not allow testing whether this modified EEG scoring may influence patient care: a prospective design would be needed. Another issue is that the post-test probability is somewhat similar compared to the pre-test probability of a good outcome (which was 55% in this cohort). The vast majority of patients did not undergo cEEG; however, repetitive routine EEG probably bears similar predictive information
      • Alvarez V.
      • Sierra-Marcos A.
      • Oddo M.
      • Rossetti A.O.
      Yield of intermittent versus continuous EEG in comatose survivors of cardiac arrest treated with hypothermia.
      and does not seem to influence outcome
      • Rossetti A.O.
      • Schindler K.
      • Sutter R.
      • Ruegg S.
      • Zubler F.
      • Novy J.
      • et al.
      Continuous vs Routine Electroencephalogram in Critically Ill Adults With Altered Consciousness and No Recent Seizure: A Multicenter Randomized Clinical Trial.
      • Fatuzzo D.
      • Beuchat I.
      • Alvarez V.
      • Novy J.
      • Oddo M.
      • Rossetti A.O.
      Does continuous EEG influence prognosis in patients after cardiac arrest?.
      . One last point is that a “burst-suppression” pattern directly classifies EEGs into the “highly malignant” category, which was not put into question for this study; a small number of these recordings (5 cases) showed however distinct reactivity to stimuli. This should be further investigated in a larger dataset.

      Conclusion

      “Benign EEG” after CA may be defined in a more comprehensive way, regardless of background amplitude, anterior-posterior amplitude development and discontinuity, allowing a high sensitivity in identification of patients with favorable outcome, with reasonable PPV. Further studies are warranted to develop the present results in order to refine current knowledge on identification of patients with favorable prognosis, including a more granular analysis of the different EEG items (potentially considering cEEG), addressing interrater reliability, and including other clinical items for a multi-modal analysis.

      Conflicts of Interest

      The authors of this study have no conflicts of interest to declare.

      Acknowledgements

      The authors thank John-Paul Miroz, RN, and Yoanne Boulez, RN, for help with data acquisition.

      Appendix A. Supplementary data

      The following are the Supplementary data to this article:

      References

        • Binnie C.D.
        • Lloyd D.S.
        • Margerison J.H.
        • Maynard D.
        • Prior P.F.
        • Scott D.F.
        EEG prediction of outcome after resuscitation from cardiac or respiratory arrest.
        Electroencephalogr Clin Neurophysiol. 1970; 29: 105
        • Friberg H.
        • Cronberg T.
        • Dunser M.W.
        • Duranteau J.
        • Horn J.
        • Oddo M.
        Survey on current practices for neurological prognostication after cardiac arrest.
        Resuscitation. 2015; 90: 158-162
        • Hirsch L.J.
        • Fong M.W.K.
        • Leitinger M.
        • LaRoche S.M.
        • Beniczky S.
        • Abend N.S.
        • et al.
        American Clinical Neurophysiology Society's Standardized Critical Care EEG Terminology: 2021 Version.
        J Clin Neurophysiol. 2021; 38: 1-29
        • Hirsch L.J.
        • LaRoche S.M.
        • Gaspard N.
        • Gerard E.
        • Svoronos A.
        • Herman S.T.
        • et al.
        American Clinical Neurophysiology Society's Standardized Critical Care EEG Terminology: 2012 version.
        J Clin Neurophysiol. 2013; 30: 1-27
        • Rossetti A.O.
        • Tovar Quiroga D.F.
        • Juan E.
        • Novy J.
        • White R.D.
        • Ben-Hamouda N.
        • et al.
        Electroencephalography predicts poor and good outcomes after cardiac arrest: A Two-center study.
        Crit Care Med. 2017; 45: e674-e682
        • Ruijter B.J.
        • Tjepkema-Cloostermans M.C.
        • Tromp S.C.
        • van den Bergh W.M.
        • Foudraine N.A.
        • Kornips F.H.M.
        • et al.
        Early electroencephalography for outcome prediction of postanoxic coma: A prospective cohort study.
        Ann Neurol. 2019; 86: 203-214
        • Westhall E.
        • Rossetti A.O.
        • van Rootselaar A.F.
        • Wesenberg Kjaer T.
        • Horn J.
        • Ullen S.
        • et al.
        Standardized EEG interpretation accurately predicts prognosis after cardiac arrest.
        Neurology. 2016; 86: 1482-1490
        • Westhall E.
        • Rosen I.
        • Rossetti A.O.
        • van Rootselaar A.F.
        • Wesenberg Kjaer T.
        • Friberg H.
        • et al.
        Interrater variability of EEG interpretation in comatose cardiac arrest patients.
        Clin Neurophysiol. 2015; 126: 2397-2404
        • Oddo M.
        • Rossetti A.O.
        Early multimodal outcome prediction after cardiac arrest in patients treated with hypothermia.
        Crit Care Med. 2014; 42: 1340-1347
        • Aicua Rapun I.
        • Novy J.
        • Solari D.
        • Oddo M.
        • Rossetti A.O.
        Early Lance-Adams syndrome after cardiac arrest: Prevalence, time to return to awareness, and outcome in a large cohort.
        Resuscitation. 2017; 115: 169-172
        • Tsetsou S.
        • Novy J.
        • Oddo M.
        • Rossetti A.O.
        EEG reactivity to pain in comatose patients: Importance of the stimulus type.
        Resuscitation. 2015; 97: 34-37
        • Nolan J.P.
        • Sandroni C.
        • Bottiger B.W.
        • Cariou A.
        • Cronberg T.
        • Friberg H.
        • et al.
        European Resuscitation Council and European Society of Intensive Care Medicine Guidelines 2021: Post-resuscitation care.
        Resuscitation. 2021; 161: 220-269
        • Rossetti A.O.
        • Oddo M.
        • Logroscino G.
        • Kaplan P.W.
        Prognostication after cardiac arrest and hypothermia: a prospective study.
        Ann Neurol. 2010; 67: 301-307
        • Westhall E.
        • Rosen I.
        • Rossetti A.O.
        • van Rootselaar A.F.
        • Kjaer T.W.
        • Horn J.
        • et al.
        Electroencephalography (EEG) for neurological prognostication after cardiac arrest and targeted temperature management; rationale and study design.
        BMC Neurol. 2014; 14: 159
        • Admiraal M.M.
        • van Rootselaar A.F.
        • Horn J.
        Electroencephalographic reactivity testing in unconscious patients: a systematic review of methods and definitions.
        Eur J Neurol. 2017; 24: 245-254
        • Booth C.M.
        • Boone R.H.
        • Tomlinson G.
        • Detsky A.S.
        Is this patient dead, vegetative, or severely neurologically impaired? Assessing outcome for comatose survivors of cardiac arrest.
        JAMA. 2004; 291: 870-879
        • Grossestreuer A.V.
        • Abella B.S.
        • Sheak K.R.
        • Cinousis M.J.
        • Perman S.M.
        • Leary M.
        • et al.
        Inter-rater reliability of post-arrest cerebral performance category (CPC) scores.
        Resuscitation. 2016; 109: 21-24
        • Sandroni C.
        • Nolan J.P.
        • Andersen L.W.
        • Bottiger B.W.
        • Cariou A.
        • Cronberg T.
        • et al.
        ERC-ESICM guidelines on temperature control after cardiac arrest in adults.
        Intensive Care Med. 2022; 48: 261-269
        • Cronberg T.
        • Greer D.M.
        • Lilja G.
        • Moulaert V.
        • Swindell P.
        • Rossetti A.O.
        Brain injury after cardiac arrest: from prognostication of comatose patients to rehabilitation.
        Lancet Neurol. 2020; 19: 611-622
        • Duez C.H.V.
        • Johnsen B.
        • Ebbesen M.Q.
        • Kvaloy M.B.
        • Grejs A.M.
        • Jeppesen A.N.
        • et al.
        Post resuscitation prognostication by EEG in 24 vs 48h of targeted temperature management.
        Resuscitation. 2019; 135: 145-152
        • Admiraal M.M.
        • van Rootselaar A.F.
        • Hofmeijer J.
        • Hoedemaekers C.W.E.
        • van Kaam C.R.
        • Keijzer H.M.
        • et al.
        Electroencephalographic reactivity as predictor of neurological outcome in postanoxic coma: A multicenter prospective cohort study.
        Ann Neurol. 2019; 86: 17-27
        • Rossetti A.O.
        • Rabinstein A.A.
        • Oddo M.
        Neurological prognostication of outcome in patients in coma after cardiac arrest.
        Lancet Neurol. 2016; 15: 597-609
        • Moseby-Knappe M.
        • Mattsson-Carlgren N.
        • Stammet P.
        • Backman S.
        • Blennow K.
        • Dankiewicz J.
        • et al.
        Serum markers of brain injury can predict good neurological outcome after out-of-hospital cardiac arrest.
        Intensive Care Med. 2021; 47: 984-994
        • Sandroni C.
        • D'Arrigo S.
        • Cacciola S.
        • Hoedemaekers C.W.E.
        • Westhall E.
        • Kamps M.J.A.
        • et al.
        Prediction of good neurological outcome in comatose survivors of cardiac arrest: a systematic review.
        Intensive Care Med. 2022; 48: 389-413
        • Sivaraju A.
        • Gilmore E.J.
        • Wira C.R.
        • Stevens A.
        • Rampal N.
        • Moeller J.J.
        • et al.
        Prognostication of post-cardiac arrest coma: early clinical and electroencephalographic predictors of outcome.
        Intensive Care Med. 2015; 41: 1264-1272
        • Willems L.M.
        • Trienekens F.
        • Knake S.
        • Beuchat I.
        • Rosenow F.
        • Schieffer B.
        • et al.
        EEG patterns and their correlations with short- and long-term mortality in patients with hypoxic encephalopathy.
        Clin Neurophysiol. 2021; 132: 2851-2860
        • Scarpino M.
        • Carrai R.
        • Lolli F.
        • Lanzo G.
        • Spalletti M.
        • Valzania F.
        • et al.
        Neurophysiology for predicting good and poor neurological outcome at 12 and 72 h after cardiac arrest: The ProNeCA multicentre prospective study.
        Resuscitation. 2020; 147: 95-103
        • Kaplan P.W.
        • Genoud D.
        • Ho T.W.
        • Jallon P.
        Etiology, neurologic correlations, and prognosis in alpha coma.
        Clin Neurophysiol. 1999; 110: 205-213
        • Backman S.
        • Cronberg T.
        • Friberg H.
        • Ullen S.
        • Horn J.
        • Kjaergaard J.
        • et al.
        Highly malignant routine EEG predicts poor prognosis after cardiac arrest in the Target Temperature Management trial.
        Resuscitation. 2018; 131: 24-28
        • Beuchat I.
        • Solari D.
        • Novy J.
        • Oddo M.
        • Rossetti A.O.
        Standardized EEG interpretation in patients after cardiac arrest: Correlation with other prognostic predictors.
        Resuscitation. 2018; 126: 143-146
        • Bongiovanni F.
        • Romagnosi F.
        • Barbella G.
        • Di Rocco A.
        • Rossetti A.O.
        • Taccone F.S.
        • et al.
        Standardized EEG analysis to reduce the uncertainty of outcome prognostication after cardiac arrest.
        Intensive Care Med. 2020; 46: 963-972
        • Elmer J.
        • Coppler P.J.
        • Solanki P.
        • Westover M.B.
        • Struck A.F.
        • Baldwin M.E.
        • et al.
        Sensitivity of Continuous Electroencephalography to Detect Ictal Activity After Cardiac Arrest.
        JAMA Netw Open. 2020; 3: e203751
        • Admiraal M.M.
        • Horn J.
        • Hofmeijer J.
        • Hoedemaekers C.W.E.
        • van Kaam C.R.
        • Keijzer H.M.
        • et al.
        EEG reactivity testing for prediction of good outcome in patients after cardiac arrest.
        Neurology. 2020; 95: e653-e661
        • Alvarez V.
        • Sierra-Marcos A.
        • Oddo M.
        • Rossetti A.O.
        Yield of intermittent versus continuous EEG in comatose survivors of cardiac arrest treated with hypothermia.
        Crit Care. 2013; 17: R190
        • Rossetti A.O.
        • Schindler K.
        • Sutter R.
        • Ruegg S.
        • Zubler F.
        • Novy J.
        • et al.
        Continuous vs Routine Electroencephalogram in Critically Ill Adults With Altered Consciousness and No Recent Seizure: A Multicenter Randomized Clinical Trial.
        JAMA Neurol. 2020; 77: 1225-1232
        • Fatuzzo D.
        • Beuchat I.
        • Alvarez V.
        • Novy J.
        • Oddo M.
        • Rossetti A.O.
        Does continuous EEG influence prognosis in patients after cardiac arrest?.
        Resuscitation. 2018; 132: 29-32