The practice of neurological prognostication in comatose CA patients
As shown in table 3, a formalized prognostic protocol is only available in 6.8% respondents. Compared with clinical examination (85.4%), neuroimaging (65.6%) and biomarkers (30.1%), electrophysiological measurement (21.9%) is less used. Among stem reflexes, pupillary reflex (91.9%) and corneal reflex (68.1%) was used by most respondents. 67.9% respondents had used brain CT for predicting outcome, but 18.4% for MRI. Among biomarkers, NSE was a little more widely used than S-100β. However, only 4.4% respondents had performed SSEP and 11.4% had performed EEG for prognostication in CA patients.
Disscussion
This survey serves as the first nation-wide analysis of neurological prognostication in survivors after CA in China and can serve as a baseline for future work. The main finding is that post CA neurological assessment has not been well understood and performed by emergency physicians who are more likely to take care of such patients in China. In practice, they were more likely to use clinical evaluation rather than other objective prognostic tools.
Clinical examination is the most commonly used method to predict outcome after CA in our study. However, the details were not well understood. Assessment of brainstem reflexes, motor responses to pain, and myoclonus represent the standard test in clinical examination.5,6 But in our survey, only 50% recognized that brainstem reflexes and motor responses to pain correlated with outcome. Corneal reflex was significantly less performed than pupillary reflex despite of its easy use. 62.7% confirmed the progGnostic value of myoclonus. In fact, myoclonus is variably associated with outcome. Myoclonic jerk is not consistently associated with poor outcome and status myoclonus starting within 48h from ROSC is highly predictive of poor outcome.7,8 This was partly explained by the lack of developed post-CA care centers in China. Moreover, the lack of special training in neurological prognostication, together with limited attendance of emergency physicians contributed to the insufficient understanding of clinical evaluation.
Guidelines suggest EEG for prognostication after CA. Certain EEG patterns such as electrocerebral silence, burst suppression, absence of reactivity and status epilepticus, was associated with a high rate of poor outcome, though not uniformly.9-11 The other patterns (including low-voltage, seizures alone and discontinuous background) were less reliable because of limited evidence.12 But epileptiform, burst suppression, isoelectric was considered to predict poor outcome by only 35.0%, 27.4%, 33.3% respondents respectively in our survey. Only 11.4% had performed EEG in routine practice in this study. However, recent surveys showed that EEG was widely used.13,14 Friberg et al reported that 74% of members from European Society of Emergency Medicine had used intermittent EEG or continuous EEG for CA patients in practice.14 The low level utility of EEG in China is likely due to several causes. One primary reason is lack of national recommendations and formalized SOP for neurological prognosis in CA patients which can improve the practice and makes the assessment less physician and/or hospital-dependent. However, national recommendation was not available until 2018 one year after our study.15 A written SOP is only available in 6.8% respondents in present study, which is far behind that of European countries. 3 In addition, there also exist barriers to access EEG services in timely manner even at developed university-affiliated hospitals in China.
Among evoked potentials, somatosensory evoked potentials (SSEP) are best studied after CA. Bilateral absence of SSEP N20 response is strongly associated with poor outcomes.5,6 Surprisingly, the present study showed that only 7.3% respondents knew the predictive value of SSEP N20 and only 4.4% had performed SSEP for CA patients in practice. There was still a huge gap with that in European countries. It was reported that more than half of respondents or ICUs had used SSEP in post CA patients.3,14 In addition to the lack of national guidelines and local protocols, the need for special skill and experience is one of the main reasons why SSEP is less used.
Despite the evidences of neuroimaging for prognostication were limited and its use only in combination with other predictors was recommended in guidelines, our survey showed that next to clinical examination, brain CT was the most widely used methods to predict outcome after CA which is in line with previous studies.14-16 This was likely related to the extensive use of brain CT to exclude brain hemorrhage or infarction after CA and to the fact that this technique is easily available. When it comes to details, however, only 30% of respondents know that a marked reduction of the grey matter/white matter ratio was associated with poor outcome.17,18 MRI has a high sensitivity for identifying ischemic brain injury and was considered to be an even more accurate tool for prognostication, however, its use can be problematic in the most clinically unstable patients.19 This partly explained the finding that the use of brain MRI for predicting outcome was reported by only 18.8% respondents in our survey.
Biomarkers were more commonly used than EEG and SSEP in present study, mainly because of their simplicity, independence from sedation and low cost. Although a wide range of proteins have been identified as indicators of brain injury, only NSE and S-100β have been extensively studied.20,21 S-100β is less well documented than NSE. However, there is a great variability in thresholds due to heterogeneity. The cut off levels varied in different studies and at different time point.20-22 Furthermore, the threshold of NSE in TTM-patients might be higher than in those without cooling as hypothermia may significantly reduce NSE levels.23 So no optimal threshold of NSE or S-100βwas established to predict poor neurological outcome with a false positive rate of 0%.
No single test can accurately predict the prognosis of patients following CA. A multimodal approach combining clinical examination, EEG, biomarkers and brain imaging was recommended to increase the certainty of prognostication. Combining any of two or more tools improved prognostic performance compared to either each individual method.24,25 Ji Hoon Kim and his college reported a stepwise model combining brain CT and MRI, NSE, EEG, SSEPs and pupil light reflex predicted poor outcomes with a 0% false positive rate.25 But in our study, the low utility of EEG, SSEP and biomarkers illustrated that multimodal approach was not well performed in practice.
The optimal time of neurological evaluation is under discussion. The current advisory suggests that multimodal modalities should not be performed before 72h from ROSC, especially in the ear of TTM.6,26 Previous studies had shown that TTM and the related pharmacological sedation may influence the neurological examination and also modify the predictive value of NSE or the accuracy of SSEPs.27 So prolonging clinical observation after rewarming, usually beyond 72h from ROSC, may be suggested when interference from residual effects of drugs is suspected.28 But our survey reported only a few respondents knew that neurological prognostication should be performed later than 72h from CA either in TTM or non-TTM patients. This misunderstanding may cause premature decisions to withdrawal of life support in practice.