Table 1.
Summary of findings regarding prognostic value of electroencephalographic reactivity in critically ill and postacute patients presenting with disorders of consciousness
| Stimuli used for EEG reactivity testing | Study | Causes | Number of patients | Main reported prognostic value of EEG reactivity | Outcome times | ||||
|---|---|---|---|---|---|---|---|---|---|
| Main statements | Se % | Sp% | PPV% | NPP% | |||||
| Only nociceptive and/or tactile | Tsetsou et al. (2018) [81] | CA/H (TH) | 61 | EEG-R predicted good outcome | 95 (75–99) | 66 (49–80) | 60 (42–76) | 96 (79–99) | CPC at 3 months |
| Rossetti et al. (2017) [73] | CA/H (TH) | 357 | Reactive EEG predicted good outcome with accuracy = 86.6% (82.6–90.0) | 80.4 (75.9–84.4) | CPC at 3 months | ||||
| Topjian et al. (2016) [80] | CA/H (children) | 128 | Absence of reactivity was associated with worse EEG background category (p < 0.001), which is associated with death aOR = 3.63 (2.18–6.0) and unfavorable neurological outcome aOR = 4.38 (2.51–7.17). | PCPC at hospital discharge | |||||
| Li et al. (2015) [66] | Mixed | 22 | EEG-R to thermal stimulation(warm water 42 ± 2 °C) was elicited in 11 patients, and 9 of them showed improved outcomes. Among the 10 patients with no EEG-R, 9 patients did not improve. | mGOS at 1 year | |||||
| Lan et al. (2015) [64] | Mixed (children) | 103 | The poor-prognosis group had the lower proportion of events in reactive EEG patterns. Compared with patients with good prognosis, patients with poor prognosis had less frequent reactive EEG patterns as well as sleep architecture (p < 0.004). | Pediatric CPC | |||||
| Kang et al. (2014) [61] | Mixed | 56 | Performance of the variable reactive EEG for recovery of awareness: OR = 21.648 (2.212 to 211.870). | 66.7 (44.7–83.6) | 75.0 (56.2–87.9) | 66.7 (44.7–83.6) | 75.0 (56.2–87.9) | GOS at 1 year follow-up | |
| Visual only | Bagnato et al. (2017) [33] | Mixed | 28 | 5 of the 16 patients with consciousness improvement showed EEG-R on baseline EEG (at admission), which was absent in all patients without improvement. | CRS-R at 6 months | ||||
| Only patients with consciousness improvement showed the reappearance of EEG-R after 6 months. Nine of the 16 patients with consciousness improvement, corresponding to 81.9% of patients who did not show EEG-R at admission, had reappearance of EEG-R at the 6-month follow-up. On the contrary, none of the patients without consciousness improvement showed reappearance of EEG-R. | |||||||||
| Nita et al. (2016) [38] | Mixed (children) | 5 | Intermittent photic stimulation induced reactivity of the burst-suppression pattern and standardized burst ratio reactivity appeared to reflect coma severity. | GCS | |||||
| Bagnato et al. (2015) [50] | Mixed | 106 | Mean CRS-R scores were lower for patients without EEG-R than for patients with EEG-R, at admission (5.4 ± 3.1 versus 10.7 ± 4.3) and after 3 months (10.6 ± 7 versus 21.2 ± 3.5). | CRS-R at 3 months | |||||
| Moreover, patients without EEG-R had less CRS-R score improvement after 3 months than patients with EEG-R (ANOVA, F1,99 = 21.5; p < 0.001). | |||||||||
| Auditory + nociceptive and/or tactile | Steinberg et al. (2018) [76] | Mixed | 585 | Reactive background EEG predicted survival aOR = 2.89 (1.49–5.59) and functionally favorable survival aOR = 1.51 (0.66–3.45). | CPC at hospital discharge | ||||
| Duez et al. (2018) [55] | Mixed | 30 | Nonreactive EEG predicted poor outcome | 40 (23–68) | 100 (69–100). | CPC at 3 months | |||
| Johnsen et al. (2017) [37] | Neurosurgical | 39 | Nonreactive EEG predicted poor outcome | 61 (42–77) | 33 (06–76) | 83 (62–95) | 13 (2–42) | GOS at 3 months | |
| Azabou et al. (2016) [8] | CA/H | 61 | Nonreactive EEG predicted an unfavorable outcome with AUC 0.82. | 84 | 80 | 98 | 31 | GOS at 1 year | |
| Kang et al. (2015) [62] | Mixed | 106 | EEG-R predicted 1-month awakening from coma with AUC = 0.79 (0.71–0.88). | 85.4 (71.6–93.5) | 74.1 (60.7–84.4) | 73.2 (59.5–83.8) | 86.0 (72.6–93.7 | CRS-R and CPC at 1 month | |
| Sivaraju et al., (2015) [75] | CA/H (TH) | 100 | Nonreactive EEG was associated with poor outcome | 79 (66–88) | 86 (66–95) | 92 (81–98) | 65 (47–79) | GOS at discharge | |
| Gilmore et al. (2015) [28] | Septic | 98 | Nonreactive EEG was associated with mortality | Mortality and mRS at 1 year | |||||
| Ribeiro et al. (2015) [71] | CA/H | 36 | Reactivity of the first EEG might predict better survival in post-cardiac arrest patients with hypoxic encephalopathy and generalized or bilateral lateralized periodic epileptiform discharges on first EEG (p = 0.0794). | Survival at hospital | |||||
| Su et al. (2013) [141] | CA/H (Stroke) | 162 | Dominant alpha wave without reactivity and dominant slow-wave rhythmic activity without reactivity were found to be correlated with poor outcome with ORs = 1.19 (0.27–5.14), and 1.82 (0.61–5.42), respectively. | mRS at 3 months | |||||
| Howard et al. (2012) [59] | CA/H | 39 | EEG-R to external stimuli (p = 0.039) and the presence of spontaneous fluctuations in the EEG (p = 0.003) were significantly associated with a favorable outcome. | mGOS at hospital discharge | |||||
| Zhang et al. (2011) [83] | CA/H (stroke) | 161 | Unfavorable EEG patterns, lack of EEG reactivity, pathologic N20 of SSEP, and pathological wave V of BAEP were associated with unfavorable outcome. | (92.4–97.0) | (82.5–99.5) | GOS at 6 months | |||
| Logi et al. (2011) [14] | Mixed | 50 | EEG-R is a good prognostic factor of recovery of consciousness in the postacute phase of brain injury; nevertheless, its absence is not invariably associated with a poor prognosis. | 68.7 | 88.9 | LCFS at 5 months | |||
| EEG reactivity predicted recovery of consciousness after 5 months from EEG recording with OR = 0.08 (0.01–0.44), p = 0.004 and 0.05 (0.01–0.53), p = 0.013, respectively, in univariable and multivariable logistic regression models. | |||||||||
| Rossetti et al. (2010) [72] | CA/H | 111 | Unreactive EEG background was found in 3 of 45 (8%) survivors versus 53 of 65 (81%) nonsurvivors p = 0.001 (Fisher’s exact test). Unreactive EEG background was incompatible with good long-term neurological recovery (CPC 1–2) and was strongly associated with in-hospital mortality: aOR for death = 15.4 (3.3–71.9). | CPC at 3 and 6 months | |||||
| Gütling et al. (1995) [58] | severe TBI | 50 | All but one patient with preserved EEG reactivity (96%) had a good global outcome, but 93% of the patients in whom EEG reactivity was absent had a bad outcome. Using discriminant analysis, EEG-R correctly classified 92% of the patients into good or bad global outcome groups. EEG-R is an excellent long-term global outcome predictor, superior to the central conduction time of the somatosensory evoked potentials and GCS. | GOS at 1,5 years | |||||
| Auditory + nociceptive and/or tactile + visual | Li et al. (2018) [65] | CA/H | 73 | EEG-R predicted survival with OR = 8.75 (1.48–51.95), p = 0.017. | 82.1 | 84.1 | 86.8 | 78.7 | GOS at 6 months |
| Fernández-Torre et al. (2018) [57] | CA/H | 26 | In patients with a diagnosis of postanoxic alpha coma, theta coma, or alpha-theta coma, there was increased association of EEG-R with survival (p = 0.07). | CPC at 5 months | |||||
| Fantaneanu et al. (2016) [56] | CA/H (TH) | 60 | EEG-R varies depending on the stimulus modality as well as the temperature. EEG to nipple pressure is the most sensitive EEG-R test for outcome during hypothermia, with a good specificity, and is associated with good outcomes during either hypothermic or normothermic periods. | 75 | 79.5 | CPC at hospital discharge | |||
| Braksick et al. (2016) [52] | Mixed | 416 | Absence of EEG-R was independently associated with in-hospital mortality: | In-hospital mortality | |||||
| OR = 8.14 (4.20–15.79) | |||||||||
| Mohammad et al. (2016) [68] | Septic (children) | 119 | A nonreactive background was noted in 48% (57 of 119) of patients on their first EEG and predicted abnormal outcome in children with encephalitis (OR = 3.8, p < 0.001). | LOS at last follow-up | |||||
| Juan et al. (2015) [60] | CA/H | 197 | Seventy-two patients (37%) had a nonreactive EEG background during TH, with 13 (18%) evolving toward reactivity in NT. Compared with those remaining nonreactive (n = 59), they showed significantly better recovery of brainstem reflexes (p < 0.001), better motor responses (p < 0.001), transitory consciousness improvement (p = 0.008), and a tendency toward lower NSE (p = 0.067). | CPC at 3 months | |||||
| Oddo and Rossetti (2014) [69] | CA/H (TH) | 134 | AUC for nonreactive hypothermic EEG for predicting mortality and poor outcome were 0.86 (0.81–0.92) and 0.81 (0.75–0.87), respectively | CPC at 3 months | |||||
| Crepeau et al. (2013) [54] | CA/H | 54 | Nonreactive EEG was associated with poor outcome with OR = 17.05 (3.22–90.28). | CPC at hospital discharge | |||||
| Sutter et al. (2013) [78] | Mixed | 105 | Nonreactive EEG background was independently associated with death in encephalopathic patients with triphasic waves: OR = 3.73 (1.08–12.80, p = 0.037). | Mortality and CPC at discharge | |||||
| Bisschops et al. (2011) [51] | CA/H (TH) | 103 | EEG was unreactive in 15 of 23 patients (65.2%) with an unfavorable outcome and in none of the 4 patients with a good outcome (p = 0.015). | 100 (75–100) | GOS at hospital discharge | ||||
| Rossetti et al. (2010) [74] | CA/H (TH) | 34 | Nonreactive cEEG background during therapeutic hypothermia had false-positive rate of 0 (0–18%) for mortality. All survivors had cEEG background reactivity, and the majority of them (14 [74%] of 19) had a favorable outcome. | 100% (74 to 100%) | CPC at 2 months | ||||
| Ramachandrannair et al. 2005 [70] | Mixed (children) | 33 | Among the 19 children with nonreactive EEG, 13 (65%) had unfavorable outcomes, including 10 deaths. Outcome was better in children with EEG-R (p = 0.023). EEG-R was associated with a lower PCOPCS score at follow-up (p = 0.002). | PCOPCS at 1 year | |||||
| Amantini et al. (2005) [49] | Severe TBI | 60 | Awakening prediction with EEG-R: LR+ = 1.6 (0.8–3.2). | 66.7 | 60.0 | 83.3 | 37.5 | GOS at 1 year | |
| Good outcome prediction with EEG-R: LR+ = 1.8 (1.2–2.9). | 79.3 | 58.1 | 63.9 | 75.0 | |||||
| Young et al. (1999) [82] | Mixed | 214 | Nonreactive EEG was one of the individual factors strongly related to mortality: OR > 2.0. | > 0.80 | |||||
| EEG-R was among factors that favored survival rather than death. | |||||||||
| Kaplan et al. (1999) [44] | Mixed | 36 | Presence of EEG reactivity in alpha coma correlated with survival (χ2 = 5.231; p = 0.022). If the EEG showed no reactivity after cardiac arrest, patients were likely to die (χ2 = 3.927; p = 0.0475). | GOS after hospital discharge | |||||
| Not described | Søholm et al. (2014) [30] | CA/H | 219 | A favorable EEG pattern (including reactivity) was independently associated with reduced mortality with HR 0.43 (0.24–0.76), p = 0.004 (false-positive rate, 31%) and a nonfavorable EEG pattern (including no reactivity) was associated with higher mortality (HR = 1.62, 1.09–2.41, p = 0.02) after adjustment for known prognostic factors (false-positive rate, 9%). | 30-day mortality and CPC at hospital discharge | ||||
| Kessler et al. (2011) [63] | CA/H (TH) Children | 35 | During hypothermia, patients with EEGs in categories 2 (continuous but unreactive EEG) or 3 (discontinuity, burst suppression, or lack of cerebral activity) were far more likely to have poor outcome than those in category 1 (continuous and reactive EEG) (OR = 10.7, p = 0.023, and OR = 35, p = 0.004, respectively). Similarly, for EEG obtained during normothermia, patients with EEGs in categories 2 or 3 were far more likely to have poor outcomes than those in category 1 (OR = 27, p = 0.006, and OR = 18, p = 0.02, respectively). | PCPC at hospital discharge | |||||
| Thenayan et al. (2010) [79] | CA/H | 29 | Of the 18 patients with nonreactive EEG, only 1 recovered awareness; of the 11 patients with EEG-R, 10 recovered awareness. | 90 (57–100) | 94 (70–100) | Awakening during hospitalization | |||
| Claassen et al. 7(2006) [53] | SAH | 116 | Outcome was poor in all patients with absent EEG reactivity | 3-month mRS | |||||
Abbreviations: ANOVA Analysis of variance, BAEP Brainstem auditory evoked potential, Se Sensitivity, Sp Specificity, PPV Positive predictive value, NPV Negative predictive value, aOR Adjusted OR, CA/H Cerebral anoxia/hypoxia, TH Target therapeutic hypothermia, NT Normothermia, Mixed = Heterogeneous population of critically ill or postacute patients with disorders of consciousness from various causes (toxic, septic, metabolic, or vascular). CPC Cerebral Performance Categories scale, PCPC Pediatric Cerebral Performance Category scale, PCOPCS Pediatric Cerebral and Overall Performance Category scale, GCS Glasgow Coma Scale, GOS Glasgow Outcome Scale, mGOS Modified Glasgow Outcome Scale, mRS Modified Rankin Scale, LCFS Level of Cognitive Functioning Scale, LOS Liverpool Outcome Score, CRS-R Coma Recovery Scale–Revised, cEEG Continuous electroencephalography, SAH Subarachnoid hemorrhage, NSE Neuron-specific enolase, LR+ Positive likelihood ratio