Early Prognosis in Coma After Cardiac Arrest: a Prospective Clinical, Electrophysiological, and Biochemical Study of 60 Patients

61060ournal ofNeurology, Neurosurgery, and Psychiatry 1996;61:610-615 J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.61.6.610 on 1 December 1996. Downloaded from Early prognosis in coma after cardiac arrest: a prospective clinical, electrophysiological, and biochemical study of 60 patients

C Bassetti, Fulvio Bomio, Johannes Mathis, Christian W Hess

Abstract estimate, neuronal damage must be assessed Background-The univariate study of indirectly by exploring cerebral functions and clinical, electrophysiological, or bio- evaluating their pattern of recovery over time. chemical variables has been shown to In general, the longer the delay in recovery of predict the outcome in postanoxic coma function the worse the prognosis. Due to the in about 50% of patients for each type of greater vulnerability of the cerebral cortex to variable. Previous studies did not, how- hypoxia than the brainstem, recovery usually ever, consider the prognostic accuracy of occurs in a caudorostral sequence,5 and the a multivariate approach. maximal recovery time still compatible with a Methods-Sixty patients in coma for good outcome depends on the neurological more than six hours after cardiac arrest function being considered. Several studies were prospectively examined by means of have shown that the duration and depth of repeated clinical examinations (including coma are good predictors of outcome.3'5 6 7 The Glasgow coma score (GCS)), EEG, and absence of brainstem reflexes for more than six medianus nerve somatosensory evoked to 24 hours after cardiopulmonary resuscita- potentials (SEPs). In 16 patients, the tion,3589 absent motor responses to pain for early concentrations of serum neuron more than one or two days,371011 the persis- specific enolase and ionised calcium were tence of a Glasgow coma score (GCS) < 5 for also measured. more than two or three days,67 and of a GCS Results-Within the first year after car- < 8 for more than one week36 all herald a poor diac arrest, 20% of patients made a good outcome. However, clinical prognostication neurological recovery; 80% remained in a within the first few days after cardiopulmonary vegetative state or died. Clinical exami- resuscitation is afflicted with incontestable nation correctly predicted outcome in limitations. Firstly, there is a risk of false pes- 58% of patients, SEP in 59%, and EEG in simistic predictions,'2 which may be as high as 41%. The combination of clinical exami- 5% to 1 0%.13 14 Secondly, brainstem functions nation, SEP, and EEG raised the percent- normalise in most patients within a few hours age of correct predictions to 82%, without and accurate prediction of outcome in these false pessimistic predictions. Concen- patients is not possible within the first two trations of serum neuron specific enolase days after cardiopulmonary resuscitation, as

and ionised calcium were of no additional even a GCS of 3 is compatible with a good http://jnnp.bmj.com/ prognostic help. Multivariate regression outcome.3 7 15 16 Thirdly, the accuracy of clini- analysis identified the association of GCS cal parameters is limited in intubated patients < 8 at 48 hours with abnormal or absent who often need muscle relaxation and seda- early cortical SEPs as highly predictive of tion. For these reasons, an intense search for a bad outcome (risk = 97%, 95% confi- electrophysiological and biochemical criteria dence interval = 86-99%). for prognosis in postanoxic coma has been Conclusion-The combination of GCS at undertaken. The absence of continuous EEG 48 hours, SEP, and if these are non- activity for more than eight hours after car- on September 24, 2021 by guest. Protected copyright. conclusive, EEG, permits a more reliable diopulmonary resuscitation,5 the presence of a prediction of outcome in postanoxic coma periodic or "burst suppression" EEG,'7-19 and than clinical examination alone. the absence of early cortical SEPs'6 20-22 in the first two to three days have been shown to be Department of (J Neurol Neurosurg Psychiatry 1996;61:610-615) reliable predictors of a bad outcome. More Neurology recently, biochemical parameters such as C Bassetti J Mathis serum glucose at admission,2 serum ionised CW Hess Keywords: cardiac arrest; postanoxic coma; Glasgow calcium,23 CSF creatine kinase,22 24 CSF lac- coma scale; electroencephalography; somatosensory tate,25 and serum neuron specific Department of evoked potentials enolase,26 Medicine, University have been suggested as potential "markers" of Hospital (Inselspital) diffuse cerebral ischaemia. Their prognostic Bern, Switzerland F Bomio Up to 44% of patients with cardiac arrest can accuracy, however, remains to be confirmed. Correspondence to: be resuscitated successfully.' Most remain Despite considerable information about the Dr C Bassetti, Department comatose initially and only 10%-20% make a prognostic accuracy of single parameters of Neurology, University '3 Hospital (Inselspital) 3010 good recovery. Prediction of individual out- alone, the value of a multivariate approach has Bern, Switzerland. come is a difficult task with important med- not yet been determined. Clearly, the com- Received 29 January 1996 ical, ethical, and socioeconomic implications.4 bined analysis of, for example, clinical exami- and in revised form 13 August 1996 As the duration and severity of cerebral nation, EEG, and SEP has the potential Accepted 19 August 1996 ischaemia are usually unknown or difficult to advantage of giving a more complete func- Early prognosis in coma after cardiac arrest: a prospective clinical, electrophysiological, and biochemical study of 60 patients 611 J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.61.6.610 on 1 December 1996. Downloaded from tional profile of a patient. To assess whether a Table 1 Grading ofEEG and medianus somatosensory multivariate approach provides a better pre- evoked potentials (SEPs) diction of outcome within the first three days EEG Grade I Dominant normally distributed a activity, reactive after cardiopulmonary resuscitation, we pro- Grade II Dominant 0-6 activity, reactive spectively studied a consecutive series of Grade III 6-0 activity without a activity Reactive or non-reactive patients in postanoxic coma with a standard Grade IV* "a or 0 coma", non-reactive examination protocol which included repeated Grade IV Burst suppression activty Low voltage 6 activity, non-reactive clinical examinations, EEG, and SEP, serum Periodic general phenomena with isoelectric ionised calcium, and serum neuron specific intervals Grade V Very low voltage EEG enolase. Isoelelectric EEG SEP Grade I Normal N20/P25 response bilaterally (normal amplitude + normal central conduction Patients and methods time) We studied a series of 60 consecutive patients Grade II Abnormal N20/P25 response unilaterally or bilaterally (amplitude reduction > 50% or after cardiac arrest and cardiopulmonary CCT > 7-2 ms, or both) resuscitation who remained comatose, as Grade III Absent N20/P25 bilaterally defined elsewhere,3 for more than six hours. The average age was 62 years, ranging from 19 to 86 years, and 49 were men. Patients with craniocerebral trauma, pre-existing intra- Biochemical studies were carried out in 16 cranial lesions, or drug intoxication were patients. Serum neuron specific enolase con- excluded. All patients were initially intubated, centrations were determined in heparinised most received slight hyperventilation, intra- blood 12 and 24 hours after cardiopulmonary venous nutrition, and slight sedation with resuscitation. Concentrations in serum were midazolam or morphine. Intensive care treat- measured by a commercially available ment was guaranteed in all patients for at least radioimmunoassay (Pharmacia-PDF-Sweden). the first three days after cardiopulmonary The concentration of serum neuron specific resuscitation. enolase measured in healthy subjects in our Repeated clinical examination was per- laboratory is < 20 m mol/l. Ionised calcium, formed in each patient at regular intervals total calcium concentrations, and pH were between six to 12 hours, at 24 hours, at 48 determined in an arterial sample of hours, and at one week after cardiopulmonary heparinised blood four and 12 hours after car- resuscitation. All clinical assessments were diopulmonary resuscitation. Measurements performed by two of us (CB, FB). Glasgow were performed with a calcium selective elec- coma score (best 15, worst 3), corneal, pupil- trode and corrected for a pH of 7-4. The nor- lar, and oculocephalic brainstem reflexes, the mal range of ionised calcium concentrations in presence of spontaneous flexor ("decortica- healthy subjects in our laboratory is tion") or extensor ("decerebration") postur- 1 20-1 30 mmol/l, and for total calcium it is ing, and myoclonic or epileptic phenomena 2-10-2-50 mmol/l. were noted. Outcome was defined as the best cerebral Electrophysiological studies were carried performance achieved at any follow up time out within the first three days after cardiopul- and classified by analogy with the GOS into

monary resuscitation in all but four patients; two groups: (1) good outcome: awakening http://jnnp.bmj.com/ EEG and SEP were always recorded 12 hours with reappearance of signs of cognition or more after cardiopulmonary resuscitation, (Glasgow outcome scale (GOS) 3-5); (2) bad as even an isoelectric EEG before this interval (poor) outcome: persistent vegetative state or does not exclude full recovery.27 In 48 patients death (GOS 1-2). Outcome was evaluated at both EEG and SEPs were recorded within 24 one, two, and seven days, as well as one month hours of each other. An eight channel EEG and one year after cardiopulmonary resuscita- was recorded in 59 patients for at least 30 min- tion. We decided to include patients with a utes, using the international 10-20 system and GOS = 3 in the "good outcome" group on September 24, 2021 by guest. Protected copyright. including testing of the reactivity to acoustic, because some recovery may occur in this painful, and photic stimulation. Recordings patient group beyond our observation time. were classified by a visual analysis in a five grade scale of increasing severity (table 118). STATISTICAL ANALYSIS SEPs were obtained from 56 patients by elec- Predictive values were determined for favour- trical stimulation of the median nerve at the able and unfavourable factors. Favourable pre- wrist.28 The central conduction time and the dictive values (percentage of patients with the P25/N20 amplitude ratio20 of SEPs were test result whose good outcome was correctly defined as abnormal when exceeding control predicted), and unfavourable predictive values data from healthy subjects in our laboratory by (percentage of patients with the test result more than 2-5 SD. Results of the SEP were whose bad outcome was correctly predicted) classified in three grades of increasing severity and their approximate 95% confidence inter- (table 1). Clinical, EEG, and SEPs findings vals (95% CIs) were calculated for clinical, were classified in three prognostic groups EEG, and SEP parameters. The number (and (favourable; uncertain; unfavourable) accord- percentage) of correct and wrong predictions ing to currently suggested prognostic criteria of both good and bad outcome were calcu- (see above and table 2). Only very selective cri- lated for clinical examination, EEG, and SEP teria for unfavourable outcome were chosen to separately, according to the prognostic criteria avoid false pessimistic predictions. shown in table 2. We decided to use these cri- 612 Bassetti, Bomio, Mathis, Hess J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.61.6.610 on 1 December 1996. Downloaded from Table 2 Published prognostic criteria ofclinical examination, EEG and SEPs in the awakened and reached their best level of func- literature tion within the first week, and nine patients GCS at 48-72 hours eventually became independent in daily activi- BR at 6-12 hours EEG grade SEPs ties. Thirteen of 60 patients (22%) died within Favourable > 8 I-II Normal the first three days, six of whom were certified and normal BR Uncertain 5-7 III, IV* Abnormal brain dead. Within the next month a further and normal BR N20/P25 31 patients died, mostly (29/3 1) without Unfavourable < 5 IV-V Absent or > 1 BR absent N20/P25 bilaterally awakening from coma. Four patients remained in a vegetative state for more than one month *"a or 6 coma". GCS = Glasgow coma score; BR = brainstem reflexes (corneal, pupillary, oculocephalic); SEP and died 70, 72, 89, and 280 days after car- - medianus somatosensory evoked potentials. diopulmonary resuscitation. Table 3 sum- marises the predictive values of clinical teria because their accuracy has already been findings, EEG, and SEP. proved in several large series of patients (see introduction). The prognostic value of the CLINICAL EXAMINATION combination of the three modalities was Only eight of 28 patients with normal brain- assessed as follows. A good outcome was pre- stem reflexes at 6-12 hours had a good out- dicted if there were (1) no unfavourable find- come (favourable predictive value = 29%, ings, and (2) at least one favourable finding on 95% CI 13-49%). On the other hand, all 10 clinical examination, EEG, or SEP. Con- patients in whom two or more brainstem versely, a bad outcome was predicted if there reflexes were absent at 6-12 hours died in were one or more unfavourable findings on coma (unfavourable predictive value = 100%, clinical examination, EEG, or SEP. The num- 69-100%). The GCS in the first 24 hours ber (and percentage) of correct and wrong ranged between 3 and 14 in patients with good predictions of both good and bad outcome outcome, and between 3 and 9 in patients with were calculated for this multivariate approach. bad outcome. However, seven of nine patients A multivariate logistic regression analysis was with a GCS > 8 at 48 hours had a good out- used to identify from a series of different para- come (favourable predictive values = 78%, meters (brainstem reflexes at 6-12 hours, 40-97%), whereas all 20 patients with a GCS < GCS at 24 hours, GCS at 48 hours, motor 5 at 48 hours had a bad outcome responses at 24 and at 48 hours, EEG, and (unfavourable predictive value = 100%, SEP) the combination which best predicts the 83-100%). Spontaneous eye opening was outcome. The comparison of patients with often seen within the first 48 hours in patients good and bad outcome was performed for with both good and bad outcome. Hence, the parametric variables with the unpaired t test, favourable predictive values of spontaneous and for non-parametric variables with the eye opening was only 23% (5%-54%) at 24 Mann-Whitney U test, X2 test, and compari- hours, and 53% (27-79%) at 48 hours. Poor son of proportions test. motor responses to pain (absent or stereotyped responses) heralded a poor outcome, particu- larly at 48 hours (favourable predictive value Results = 92%, 78-98%). Myoclonic or other epileptic Of 60 patients in coma for more than six hours phenomena were seen within the first three

after cardiopulmonary resuscitation, 12 (20%) days after cardiopulmonary resuscitation in 25 http://jnnp.bmj.com/ had a good outcome and 48 (80%) had a bad patients, and were limited in eight of them to outcome. Five of 12 patients with good out- the periocular region. Twenty four patients come remained in coma (GCS < 8) for two or had a bad outcome (unfavourable predictive more days after cardiopulmonary resuscita- value = 96%, 80-100%). Spontaneous tion. However, all but one of the 12 patients "decerebration" and "decortication" spasms were found in 11 patients including four patients with a good outcome. One patient Table 3 Prognostic value ofclinicalfindings, EEG, and SEPs in 60 patients with with good outcome had these up to 34 hours on September 24, 2021 by guest. Protected copyright. postanoxic coma after cardiopulmonary resuscitation. Time after cardiac arrest (h) 12-72 ELECTROPHYSIOLOGY 6-12 24 48 Four of 10 patients with favourable EEG Predictors ofgood outcome had a (percentage of patients with the test result whose good outcome was correctly prer dicted) (grade I-II) good recovery (favourable 3 BRpresent 29 (28) 33 (21) 33 (27) predictive value = 40%, 12-74%). An GCS > 8 50 (4) 50 (8) 78 (9) unfavourable EEG (grade IV-V) was found in Spontaneous eye opening 17 (6) 23 (13) 53 (15) Favourable EEG (grade I-II) 40 (10) 20 (34%) of 59 patients. All 20 patients had a Favourable SEP (N20/P25 normal 50 (20) bad outcome (unfavourable predictive value bilaterally) = Fourteen of 59 patients Predictors of bad outcome 100%, 83-100%). (percentage of patients with the test result whose bad outcome was correctly prediicted) (24%) had an "a or 0 coma" EEG (grade IV). > 1 BR absent 100 (10) 100 (5) 100 (3) In 11 of them cortical SEPs were absent bilat- GCS < 5 81 (27) 87 (31) 100 (20) Poor motor response* 79 (38) 85 (48) 92 (37) erally and none recovered. In two patients "a Seizures or myoclonus Unfavourable EEG (Grade IV-V) 190 (20) coma was "incomplete"'8 as the EEG activity Unfavourable SEP (N20/P25 was either diffusely distributed or intermit- absent bilaterally) 100 (23) tently reactive. Both patients had normal

Values are % with total number of patients in parentheses (95% CIs are given in the text). SEPs, and one of them had a good recovery. BR = brainstem reflexes (comeal, pupillary, oculocephalic); GCS = Glasgovw coma score; SEPs were obtained in 56 patients. Normal SEP = medianus nerve somatosensory evoked potential. *No or only stereotyped (extensor, flexor) motor response to pain. SEPs were recorded in 20 patients, and 10 of Early prognosis in coma after cardiac arrest: a prospective clinical, electrophysiological, and biochemical study of 60 patients 613 J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.61.6.610 on 1 December 1996. Downloaded from Table 4 Comparison ofpatients with good and bad outcome afterpostanoxic coma for the clinical examination, 41% for the EEG, Good outcome Bad outcome and 59% for the SEPs. The combination of (n = 12) (n = 48) P value three modalities raised the percentage of cor- Age (y (SD)) 64.8 (12-4) 60-8 (16-5) NS rect predictions of outcome to 82%. There Brainstem reflexes were no false negative predictions. A correct (at 6-12 hours, n (%)): All three present 8 (67) 20 (48) NS prediction of bad outcome was possible in 36 Two or more absent 0 10 (24) < 0-05 of 48 (75%) patients combining clinical exam- Poor motor responses (no or only stereotyped responses, n ination and SEP alone, a total which was At 6-12 hours 8 (80) 30 (97) NS raised by one to 37 (77%) by also considering At 24 hours 7 (70) 39 (87) NS At 48 hours 3 (25) 34 (92) < 0-05 the EEG. Prediction of a bad outcome was Spontaneous eye opening (n (%)): easier in patients who died within the first At6-12hours 1 (11) 5 (17) NS At24hours 3 (33) 10 (22) NS three days (92%) than in those who survived At48 hours 8 (67) 7 (18) < 0 05 longer (71%). Electrophysiological tests were Glasgow coma scale (score (SD)): At 6-12hours 5 (3) 4 (2) NS of significant prognostic help in nine (36%) of At 24 hours 6 (3) 5 (2) NS 25 patients in whom prediction based on clini- At48 hours 10 (4) 5 (2) < 0 05 Seizure or myoclonic activity cal signs was considered to be uncertain. In (within first 3 days, n (%)) 1 (8) 24 (52) < 0-05 addition, unfavourable EEG or SEP findings EEG findings (59 patients, n (%)): Favourable or uncertain 11 (100) 28 (58) were found in 23 (66%) of 35 patients with Unfavourable 0 20 (42) < 0 05 unfavourable clinical signs. Conversely, nor- SEP findings (56 patients, n (%)): Favourable or uncertain 11 (100) 22 (49) mal or abnormal biochemical tests were of no Unfavourable 0 23 (51) < 0 05 additional prognostic help. A correct predic- SEP = medianus nerve somatosensory evoked potentials. tion of good outcome was possible applying the multivariate approach in 11 of 12 patients. them had a good outcome (favourable predic- There were 39% false optimistic predictions. tive value = 50%, 30-74%). Delayed or low A multivariate analysis identified the combina- amplitude cortical (N20/P25) evoked poten- tion of GCS after 48 hours and SEP as the tials were found in 12 patients, and only one of best predictive combination. The risk of a bad them had a good outcome. Cortical SEPs were outcome was calculated to be 97% (86%- absent in 23 patients. All of them had a bad 99%) for patients with a GCS < 8 and abnor- outcome (unfavourable predictive value = mal (or absent) SEP, but only 12% (2-46%) 100%, 90-100%). Table 4 shows the compar- for patients with a GCS > 8 and normal SEPs ison of patients with good and bad outcome. (table 5). The absence of two or more brainstem reflexes at 6-12 hours, poor motor responses at 48 hours, absent spontaneous eye opening at 48 Discussion hours, seizure or myoclonic activity, and Coma following resuscitation after cardiac unfavourable EEG and SEP findings were arrest is a serious condition, and a good out- associated with a bad outcome. There was also come can be expected in less than 20% of a significant difference between patients with patients, as shown in this and previous larger good and bad outcome in the GCS at 48 series."3Although awakening from coma usu- hours, but not in the GCS at 6-12 hours and ally occurs within the first three days after car- 24 hours. diopulmonary resuscitation'° 25 we confirm that

recovery after this interval can occur.36 Because http://jnnp.bmj.com/ BIOCHEMICAL PARAMETERS decisions about limitation or maintenance of Serum neuron specific enolase concentrations intensive care often need to be made within the were above normal in seven patients after 12 first few days after cardiopulmonary resuscita- hours, and in eight patients after 24 hours. tion, prognostic criteria in postanoxic coma are The unfavourable predictive values of high needed. In this study we tried to avoid some of serum neuron specific enolase were 86% at 12 the methodological problems of previous stud- hours and 100% at 24 hours. Ionised calcium ies of prognosis in coma."3 Maximal treatment concentrations were below normal in nine was guaranteed for at least the first three days on September 24, 2021 by guest. Protected copyright. patients after four hours, and in 12 patients after cardiopulmonary resuscitation, with the after 12 hours despite normal concentrations exception ofpatients in whom the clinical crite- of total calcium. The unfavourable predictive ria of brain death were satisfied; data were col- values of low ionised calcium were 89% at lected prospectively by two observers only; four hours and 83% at 12 hours. electrophysiological studies were performed within a narrow interval of time from the onset UNVARIATE AND MULTIVARIATE APPROACH of coma; and 95% CIs for the suggested predic- The percentage of total correct predictions of tive model were provided. There are three outcome obtained applying a univariate major findings from our data. Firstly, normal or approach (criteria shown in table 2) was 58% only mildly abnormal EEG and SEP findings are unreliable predictors of a good outcome. Secondly, currently suggested univariate clini- Table S Risk ofbad outcome in coma after cardiac arrest (multivariate logistic regression cal, EEG, and SEP criteria for prognosis of bad analysis of 60 patients) outcome (table 2) are reliable, but identify only GCS > 8 at 48 hours GCS < 8 at 48 hours GCS < 8 at 48 hours about 50% of the patients who die or remain in and or and SEP normal SEP abnormal or absent SEP abnormal or absent a vegetative state. Thirdly, the combination of GCS at 48 hours, SEP, and EEG raises the per- 12% (95% CI, 2%-46%) 68% (95% CI, 47%-84%) 97% (95% CI, 86%-99%) centage of accurate prediction of bad outcome GCS = Glasgow coma score; SEP = medianus nerve somatosensory evoked potential. to 77% ofthe patients. 614 Bassetti, Bomio, Mathis, Hess J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.61.6.610 on 1 December 1996. Downloaded from PROGNOSIS OF GOOD OUTCOME arrest; SEPs are faster to perform and easier to Prognosis of good outcome in postanoxic interpret, and also they are not influenced by coma is difficult. We confirm that early recov- sedating medications. Also, we found that ery of brainstem functions and spontaneous EEG only rarely added prognostic information eye opening do not necessarily herald a good when clinical examination and SEP results outcome, in that the cortex may have been were considered together. Although we con- damaged to a degree that prevents cognitive firmed previous reports that over 80% of recovery. Also, the percentage of false opti- patients with raised serum neuron specific mistic predictions is high because of a signifi- enolase and decreased ionised calcium have a cant extracerebral mortality.7 29 Nevertheless, bad outcome,2326 the limited number of 78% of our patients with a GCS > 8 after 48 patients evaluated in this and previous studies hours had a good outcome, a percentage simi- and the non-negligible percentage of false pes- lar to the 82% reported by Mullie et al in a simistic predictions call for caution. The series of 360 patients.6 Conversely, favourable determination of CSF enzymes, including cre- EEG (grade I-II) and normal SEPs were unre- atine kinase, may yield a higher sensitivity and liable predictors of good recovery. Our results specificity in predicting outcome after cardiac agree with those of previous series in that arrest. None of these biochemical tests can, 50%-60% of patients with favourable EEG however, be suggested for routine use at the and SEP findings still had a poor out- present time.38 come. 821222530 Hence, as long as the patient remains clinically in coma, the absence of rele- MULTIVARIATE PROGNOSIS OF BAD OUTCOME vant EEG and SEP abnormalities should be In this, as well as previous reports the univariate considered an indispensible condition rather application of selective criteria allowed the than a "guarantee" for a good outcome. A identification of only about 50% of the possible exception is the preservation of long patients with bad outcome.367 Therefore, the latency (N70) cortical SEPs.'6 most relevant result of this study is the increase in the percentage of correct predic- UNIVARIATE PROGNOSIS OF BAD OUTCOME tions of bad outcome from 58% with clinical We confirm that the absence of more than one examination alone to 77% with the combina- brainstem reflex for more than 6-12 hours,339 a tion of clinical examination and electrophysi- GCS < 5 at 48 hours or later, 67 an EEG grade ology. In other words, the recording of SEP IV-V,'7 19 and the bilateral absence of cortical and EEG allowed the prediction of outcome in SEPs1620-22 at 12-72 hours after cardiopul- one third of those patients, in whom clinical monary resuscitation, all herald a bad out- signs alone were not (yet) conclusive. In addi- come without false pessimistic predictions. We tion, unfavourable electrophysiological para- also corroborate previous studies that found meters supported the clinical assessment in the following criteria to be unreliable predic- predicting a bad outcome in two thirds of tors of poor outcome: prolonged absence of a patients with unfavourable clinical signs. It is single brainstem reflex,7 a low GCS, and poor of clinical relevance to point out that the utility motor responses within the first 24 hours after of electrophysiological recordings applied not cardiopulmonary resuscitation,7 16 spontaneous only to patients who died within the first three flexor and extensor posturing,3' an "a coma" days but also to those who survived longer.

EEG,'82 532 and abnormal but still detectable Conversely, in the 16 patients tested, bio- http://jnnp.bmj.com/ cortical SEPs. Poor motor responses to pain at chemical tests were of no additional prognos- 48-72 hours3 7 "and epileptic and myoclonic tic help. A multivariate logistic regression phenomena in the first three days33 after car- analysis of our data identified the association diopulmonary resuscitation were confirmed to of a GCS < 8 at 48 hours with an abnormal or be the single most sensitive clinical predictors absent SEP as highly predictive of a bad out- of bad outcome. However, a few patients with come (risk 97%, 95% CI 86%-99%). This these findings enjoyed a good outcome, both means that the combination of clinical signs in our series and previous ones. 123436 For this and SEPs may allow the application of less on September 24, 2021 by guest. Protected copyright. reason we think that the use of a combination selective prognostic criteria for prognostica- of clinical signs (for example, GCS), should be tion of bad outcome than those suggested preferred to using a single clinical parameter when these variables are used in isolation for prognostication of postanoxic coma. (table 2). There have been only a few studies Prognostically unfavourable SEPs and EEG assessing the prognostic value of a multivariate were found both in 40%-50% of patients with approach and none including clinical examina- bad outcome. The lower than expected EEG tion, EEG, SEP, and biochemical parameters. sensitivity for severe brain damage may be in Edgren et a125 correctly predicted the outcome part related to the fact that we included a in about 75% of 32 patients combining a mod- coma patterns in the uncertain rather than ified GCS at 48 hours with the EEG at 48 unfavourable EEG signs. This decision was hours, and the CSF lactate at 24 hours. In that prompted by a few reports (confirmed in this study, however, the percentage of false pes- study) of patients with a coma and good simistic predictions was as high as 29%. In a outcome.'x 25 32 37 Although SEP and EEG study of 40 patients, Rothstein et al found, as have a similar sensitivity in detecting severe we did, that SEPs were more reliable than postanoxic brain damage, and the EEG may EEG in predicting outcome. These authors help in the recognition of subclinical seizure did not consider, however, any clinical para- activity, we think that SEP should be preferred meter, and the exact timing of electrophysio- for prognostication in coma after cardiac logical recordings was not specified. Becker et Early prognosis in coma after cardiac arrest: a prospective clinical, electrophysiological, and biochemical study of 60 patients 615 J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.61.6.610 on 1 December 1996. Downloaded from al performed repeated clinical examination, 7 Edgren E, Hedstrand U, Kelsey S, Sutton-Tyrell K, Safar and P and BRCT I Study Group. Assessment of neurological EEG, SEP, proton magnetic resonance prognosis in comatose survivors of cardiac arrest. Lancet spectroscopy (MRS) in a series of 25 patients 1994;343: 1055-9. with coma.29 of 8 Earnest MP, Breckenridge JC, Yarnell PR, Oliva PB. postanoxic Demonstration Quality of survival after out-of hospital cardiac arrest: brain lactate in MRS and absent cortical SEPs predictive value of early neurologic evaluation. Neurology were found to be 1979;29:56-60. reliable predictors of bad 9 Snyder BD, Gumnit RJ, Leppik IE. Neurologic prognosis outcome. The limitations of this study were after cardiopulmonary arrest: IV. Brainstem reflexes. Neurology 1981;31:1092-7. the high cardiac mortality, which reduced the 10 Longstreth WT, Inui TS, Cobb LA, Copass MK. number of patients available for analysis of Neurologic recovery after out-of-hospital cardiac arrest. Ann Intern Med 1983;98:588-92. cerebral prognosis, and the lack of analysis of 11 Hamel MB, Goldman L, Teno J, et al. Identification of the prognostic value of clinical examination comatose patients at high risk for death or severe disabil- ity. JAAMA 1995;273:1842-8. and EEG. 12 Golby A, McGuire D, Bayne L. Unexpected recovery from The accuracy of the multivariate approach anoxic-ischemic coma. Neurology 1995;45:1629-30. used in this to 13 Shewmon DA, De Giorgio CM. Early prognosis in anoxic study needs be verified in a coma. Neurol Clinl989;7:823-43. larger group of patients before it can be 14 Bates D. Defining prognosis in medical coma. Jf Neurol in clinical Neurosurg Psychiatry 1991;54:569-71. adopted practice. Nevertheless, the 15 Sacco RL, Van Gool R, Mohr JP, et al. Non traumatic following approach to prognostication after coma. Glasgow coma score and coma etiology as predictors of 2-week outcome. Arch Neurol 1990;47: 1181-4. cardiac arrest can be suggested integrating our 16 Madl C, Grimm G, Kramer L, et al. Early prediction of data with the information available in the liter- individual outcome after cardiopulmonary resuscitation. In Lancet 1993;341:855-60. ature. the first 48 hours after cardiopul- 17 Prior. The EEG in acute cerebral anoxia. Amsterdam: monary resuscitation serial clinical Excerpta Medica, 1973. 18 Bassetti C, Karbowski K. Prognostischer Wert der examination including assessment of brain- Elektroenzephalographie bei nicht-traumatischen stem reflexes and GCS should be performed. Komata. Schweiz med Wschr 1990;120:1425-34. 19 Synek VM. Value of a revised EEG coma scale for progno- Electrophysiological testing need not be con- sis after cerebral anoxia and diffuse head injury. Clin sidered in the first 24 hours, unless status Electroencephalogr 1990;21:25-30. or status are 20 Walser H, Mattle H, Keller HM, Janzer R. Early cortical epilepticus myoclonicus sus- median nerve somatosensory evoked potentials. pected. If by 24-48 hours the patient does not Prognostic value in anoxic coma. Arch Neurol 1986;42: SEPs should be 32-8. awaken, recorded. If clinical 21 Brunko E, Zegers de Beyl D. Prognostic value of early cor- examination and SEPs are prognostically non- tical somatosensory evoked potentials after resuscitation from cardiac arrest. Electroencephalogr Clin Neurophysiol conclusive an EEG could be performed. We 1987;66: 15-24. think that the combination of GCS at 48 22 Rothstein TL, Thomas EM, Sumi SM. Predicting outcome in hypoxic-ischemic coma. A prospective clinical and hours, SEPs, and, when necessary, EEG can electrophysiological study. Electroencephalogr Clin assist the clinician in predicting the outcome Neurophysiol 1991;79:101-7. of coma after 23 Urban P, Scheidegger D, Buchmann B, Barth D. Cardiac cardiac arrest with reasonable arrest and blood ionized calcium levels. Ann Intern Med accuracy. As any available prognostic model in 1988;109:110-3. coma some 24 Longstreth WT, Clayson KJ, Sumi SM. Cerebrospinal postanoxic carries risk of error," fluid and serum creatine kinase BB activity after out-of- the limitation or withdrawal of medical treat- hospital cardiac arrest. Neurology 1981;31:455-8. ment should 25 Edgren E, Hedstrand U, Nordin M, et al. Prediction of out- always be made on an individual come after cardiac arrest. Crit Care Med 1987;15:820-5. basis, involving the patient's family in the deci- 26 Roine RO, Somer H, Kaste M, Viinikka L, Karonen S. sion and Neurological outcome after out of hospital cardiac arrest: considering medical as well as psy- prediction by cerebrospinal fluid enzyme analysis. Arch chosocial issues. Neurol 1989;46:753-6. 27 Jorgensen EO. EEG without detectable cortical activity and Dr G van Melle, Department of Statistics, University Hospital, cranial nerve areflexia as parameters of brain death. Lausanne, Switzerland, helped with the statistical analysis. Drs Electroencephalogr Clin Neurophysiol 1974;36:70-5. http://jnnp.bmj.com/ T Herren and T Weiss, Department of Internal Medicine, 28 Bassetti C, Mathis J, Hess CW. Multimodal electrophysio- University Hospital, Bem, Switzerland, collaborated in the logical studies including motor evoked potentials in recruitment and care of the patients. The measurement of the patients with locked-in syndrome: report of six patients. J7 serum neuron specific enolase was performed by Dr P Huber, Neurol Neurosurg Psychiatry 1994;57: 1403-6. Gynecological Laboratory Medicine, University Hospital, 29 Berek K, Lechleimer P, Luef G, et al. Early determination of Basel, Switzerland, and that of the serum ionised calcium by neurologic outcome after prehospital cardiopulmonary Dr E Peheim, Chemisches Zentrallabor, University Hospital, resuscitation. Stroke 1995;26:543-9. Bem, Switzerland. Dr F Glocker and Mrs J Schmid assisted in 30 Scollo-Lavizzari G, Bassetti C. Prognostic value of the the recording of SEPs. Drs I Drury and D Gelb, Department of EEG in post-anoxic coma after cardiac arrest. Europ Neurology, University of Michigan Hospitals, Ann Arbor, MI, Neurol 1987;26: 161-70. made comments. 31 Brendler SJ, Selverstone B. Recovery from decerebration. on September 24, 2021 by guest. Protected copyright. USA, helpful Brain 1970;93:381-92. 32 Austin E, Wilkus RJ, Longstreth WT. Etiology and prognosis of alpha coma. Neurology 1988;33:773-7. 1 Bedell SE, Delbanco TL, Cook EF, Epstein FH. Survival 33 Wijdicks EFM, Parisi JE, Sharbrough FW. Prognostic after cardiopulmonary resuscitation in the hospital. N value of myoclonus status in comatose survivors of car- EnglJMed 1983;309:569-76. diac arrest. Ann Neurol 1994;35:239-43. 2 Longstreth WT, Diehr P, Inui TS. Prediction of awakening 34 Celesia GG, Grigg MM, Ross E. Generalized status after out-of-hospital cardiac arrest. N Engl Jf Med myoclonicus in acute anoxic and toxic-metabolic 1983;308: 1378-82. encephalopathies. Arch Neurol 1988;45:781-4. 3 Levy DE, Caronna JJ, Singer BH, et al. Predicting the out- 35 Krumholz A, Stern BJ, Weiss HD. Outcome from coma come from hypoxic-ischemic coma. JAMA 1985;254: after CPR: relation to seizures and myoclonus. Neurology 1171. 1988;38:401-5. 4 Gray WA, Capone RJ, Most AS. Unsuccessful emergency 36 Arnoldus EPJ, Lammers GJ. Potsanoxic coma: good recov- medical resuscitation-Are continued efforts in the emer- ery despite myoclonic status. Ann Neurol 1995;38:697-8. gency department justified? N Engl J7 Med 1991;325: 37 Soerensen K, Thomassen A, Wernberg M. Prognostic sig- 1393-8. nificance of alpha frequency EEG rhythm in coma after 5 Jorgensen EO. Natural history of global and critical brain cardiac arrest. J Neurol Neurosurg Psychiatry 1978;41: ischemia. Resuscitation 1981;9:131-88. 840-2. 6 Mullie A, Buylaert W, Michem N, et al. Predictive value of 38 Longstreth WT. Neurologic complications of cardiac arrest. Glasgow coma score for awakening after out-of-hospital In: Aminoff A, ed. Neurology and general medicine, 2nd ed. cardiac arrest. Lancet 1988;137-40. Edinburgh: Churchill-Livingstone, 1994:159-82.