Is This Patient Dead, Vegetative, or Severely ...

Online article and related content Is This Patient Dead, Vegetative, or Severely
current as of August 27, 2009. Neurologically Impaired?: Assessing Outcome for
Comatose Survivors of Cardiac Arrest
Correction
Citations Christopher M. Booth; Robert H. Boone; George Tomlinson; et al.
Topic collections
JAMA. 2004;291(7):870-879 (doi:10.1001/jama.291.7.870)
Related Letters
http://jama.ama-assn.org/cgi/content/full/291/7/870

Contact me if this article is corrected.

This article has been cited 27 times.
Contact me when this article is cited.

Neurology; Coma/ Vegetative State; Patient-Physician Relationship/ Care;
End-of-life Care/ Palliative Medicine; Prognosis/ Outcomes; The Rational Clinical
Examination
Contact me when new articles are published in these topic areas.

Usefulness of the Glasgow Coma Score in Survivors of Cardiac Arrest
Zsolt T. Stockinger. JAMA. 2004;291(19):2313.

In Reply:
Christopher M. Booth et al. JAMA. 2004;291(19):2313.

Subscribe Email Alerts
http://jama.com/subscribe http://jamaarchives.com/alerts

Permissions Reprints/E-prints
[email protected] [email protected]
http://pubs.ama-assn.org/misc/permissions.dtl

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009

THE RATIONAL CLINICIAN’S CORNER
CLINICAL EXAMINATION

Is This Patient Dead, Vegetative,
or Severely Neurologically Impaired?

Assessing Outcome for Comatose Survivors of Cardiac Arrest

Christopher M. Booth, MD Context Most survivors of cardiac arrest are comatose after resuscitation, and mean-
ingful neurological recovery occurs in a small proportion of cases. Treatment can be lengthy,
Robert H. Boone, MD, MSc expensive, and often difficult for families and caregivers. Physical examination is poten-
tially useful in this clinical scenario, and the information obtained may help physicians
George Tomlinson, PhD and families make accurate decisions about treatment and/or withdrawal of care.

Allan S. Detsky, MD, PhD, FRCPC Objective To determine the precision and accuracy of the clinical examination in pre-
dicting poor outcome in post–cardiac arrest coma.
CLINICAL SCENARIO
Case 1 Data Sources and Study Selection We searched MEDLINE for English-
A 65-year-old man experienced a wit- language articles (1966-2003) using the terms coma, cardiac arrest, prognosis, physi-
nessed ventricular fibrillation cardiac cal examination, sensitivity and specificity, and observer variation. Other sources came
arrest at home 24 hours ago. A neigh- from bibliographies of retrieved articles and physical examination textbooks. Studies
bor had performed cardiopulmonary were included if they assessed the precision and accuracy of the clinical examination
resuscitation for 5 minutes until the in prognosis of post–cardiac arrest coma in adults. Eleven studies, involving 1914 pa-
paramedics arrived and performed suc- tients, met our inclusion criteria.
cessful defibrillation. His electrocar-
diogram revealed a large anterior Data Extraction Two authors independently reviewed each study to determine eli-
myocardial infarction for which he un- gibility, abstract data, and classify methodological quality using predetermined crite-
derwent urgent coronary angioplasty. ria. Disagreement was resolved by consensus.
Although still unresponsive, he with-
draws from a painful stimulus and his Data Synthesis Summary likelihood ratios (LRs) were calculated from random effects
pupillary and corneal reflexes are pre- models. Five clinical signs were found to strongly predict death or poor neurological out-
sent. The family asks you about his come: absent corneal reflexes at 24 hours (LR, 12.9; 95% confidence interval [CI], 2.0-68.7),
chance of meaningful recovery. absent pupillary response at 24 hours (LR, 10.2; 95% CI, 1.8-48.6), absent withdrawal
response to pain at 24 hours (LR, 4.7; 95% CI, 2.2-9.8), no motor response at 24 hours
Case 2 (LR, 4.9; 95% CI, 1.6-13.0), and no motor response at 72 hours (LR, 9.2; 95% CI, 2.1-
49.4). The proportion of individuals’ dying or having a poor neurological outcome was
A 26-year-old woman presented to the calculated by pooling the outcome data from the 11 studies (n=1914) and used as an
emergency department with severe estimate of the pretest probability of poor outcome. The random effects estimate of poor
pleuritic chest pain and dyspnea. While outcome was 77% (95% CI, 72%-80%). The highest LR increases the pretest probabil-
waiting for a computed tomographic ity of 77% to a posttest probability of 97% (95% CI, 87%-100%). No clinical findings
scan in the radiology department, she were found to have LRs that strongly predicted good neurological outcome.
had an asystolic cardiac arrest. The re-
suscitation lasted 20 minutes, after Conclusions Simple physical examination maneuvers strongly predict death or poor
which she was found to have reactive outcome in comatose survivors of cardiac arrest. The most useful signs occur at 24
pupils. Three days later the family is hours after cardiac arrest, and earlier prognosis should not be made by clinical exami-
considering withdrawing care be- nation alone. These data provide prognostic information, rather than treatment rec-
cause she is still comatose. On exami- ommendations, which must be made on an individual basis incorporating many other
variables.
CME available online at
www.jama.com JAMA. 2004;291:870-879 www.jama.com

Author Affiliations are listed at the end of this article. The Rational Clinical Examination Section Editors:
Corresponding Author: Allan S. Detsky, MD, PhD, David L. Simel, MD, MHS, Durham Veterans Affairs
FRCPC, Mount Sinai Hospital, Suite 427, 600 Univer- Medical Center and Duke University Medical Center,
sity Ave, Toronto, Ontario, Canada M5G 1X5 Durham, NC; Drummond Rennie, MD, Deputy Edi-
([email protected]). tor, JAMA.

870 JAMA, February 18, 2004—Vol 291, No. 7 (Reprinted) ©2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009

ASSESSING OUTCOME FOR COMATOSE SURVIVORS OF CARDIAC ARREST

nation, her pupils are now unreactive lation can only be determined when the Table 1. Glasgow Coma Scale*
and she has no motor response or brain- true outcome of each patient is known,
stem reflexes. The nurse reports that the rather than at the time of presenta- Best Motor Response 6
patient had myoclonus 12 hours ago. tion. Recent interest has developed in 5
the potential role of neurophysiologic Obeying commands 4
Why Is the Clinical testing.6-8 A recent systematic review Localizing to pain 3
found somatosensory-evoked poten- Withdrawing to pain 2
Examination Important? tials very useful in predicting “waken- Abnormal flexion (decorticate) 1
ing” of comatose patients.7 Other re- Extensor response (decerebrate)
With the development of closed-chest search suggests that elevated serum None
cardiac massage in 1960 and the cre- levels of neuron-specific enolase may
ation of intensive care units shortly predict poor outcome in comatose sur- Best Verbal Response† 5
thereafter, it became possible to sur- vivors of cardiac arrest.8 Although these 4
vive cardiac arrest. Half a century later, results are promising, it will take some Oriented 3
cardiovascular disease is the leading time before the precise operating char- Confused conversation 2
cause of death in North America and Eu- acteristics of these tests are fully un- Inappropriate words 1
rope—accounting for approximately half derstood and before the technology is Incomprehensible sounds
of all deaths in the United States. At least widely available in clinical practice. None
225000 people die annually in the
United States from cardiovascular dis- The physical examination has the po- Eye Opening
ease before they reach a hospital. Twice tential to be extremely useful in this
as many will have cardiac arrest and at- common clinical scenario because of its Spontaneous 4
tempted resuscitation during hospital- universal availability and ease of per- To speech 3
ization. Survival rates for prehospital car- formance. From a compassionate stand- To pain 2
diac arrest range from 2% to 33%, and point, the clinical evaluation yields the None 1
reported inpatient survival rates range first information that is relayed to fam-
between 0% and 29%.1,2 Most survi- ily members desperate for informa- *The score for the scale is summed across the 3 com-
vors of cardiac arrest (Ϸ80%) are co- tion. Thus, it is crucial for physicians ponents and ranges from 3 to 15. A lower score indi-
matose after resuscitation. After trauma to understand the precision and accu-
and drug overdose, cardiac arrest is now racy of the clinical examination in de- cates more severe neurological deficits. Original Glasgow
the third most common cause of coma.3,4 termining prognosis in hypoxic- Coma Scale in Teasdale and Jennett11
With increasing public education in ba- ischemic coma.
sic life support and with the use of au- †Intubated patients cannot be given a score for the ver-
tomated defibrillators in public places, Pathophysiology
such as in airports and shopping malls, bal component, so their total scores accordingly range
post–cardiac arrest coma has become a Unlike traumatic or focal ischemic
common and important clinical syn- causes of coma, cardiac arrest pre- from 2 to 10.
drome. sents a global ischemic insult to the
brain. The extent of cerebral damage is How to Examine a
With the increased success of resus- largely influenced by the duration of in-
citation from cardiac arrest comes a mul- terrupted cerebral blood flow. Accord- Comatose Patient
titude of medical, ethical, and eco- ingly, minimizing both the arrest (no-
nomic questions. Once spontaneous flow) time and cardiopulmonary Glasgow Coma Scale. Prior to 1974, the
circulation has been restored, recovery resuscitation (low-flow) time are criti- clinical assessment of coma relied on
is far from certain. Possible outcomes cal. With the return of spontaneous cir- qualitative, descriptive terminology, and
range from complete neurological re- culation comes a transient period of ce- the presence or absence of brainstem re-
covery to death to the persistent vegeta- rebral hyperemia, which is followed by flexes. Plum and Posner10 described the
tive state. In admitted patients who sur- vasospasm and protracted global and widely used definition of coma as “a state
vive the initial cardiac arrest, rates of multifocal hypoperfusion. Cerebral oxy- of unarouseable unresponsiveness.” In
meaningful neurological recovery range gen stores and consciousness are lost 1974, Teasdale and Jennett11 published
from 10% to 30%.5 This uncertainty fur- within 20 seconds of the onset of car- the first description of the Glasgow Coma
thers the emotional distress of a griev- diac arrest while glucose and adeno- Scale (GCS, TABLE 1), which has since
ing and anxious family. Accordingly, it sine triphosphate stores are lost by 5 been used worldwide as a means of clas-
is important for families and physi- minutes. A cascade of complex chemi- sifying coma. Although originally de-
cians to have an understanding of a pa- cal derangements ensue, which leads to scribed for traumatic coma, it is equally
tient’s chance of meaningful recovery. neuronal death and culminates in the applicable to the assessment of nontrau-
post–cardiac arrest coma.9 matic coma. This ordinal scale is calcu-
Unfortunately the result of the gold lated from the sum of 3 components: mo-
standard test for prognosis in this popu- tor response, verbal response, and eye
opening. In assessing the motor re-
sponse, it is important to apply central
pain because spinal reflexes may occur
with peripheral stimulation and do not
represent a true motor response. A pain-
ful stimulus may be applied to the su-
praorbital region (deep pinching of the
skin) or the sternum (firm twisting pres-
sure applied with the examiner’s knuck-
les). The minimum GCS score is 3 and
maximum is 15.

Physical Examination Maneuvers. In
addition to the GCS, various brain-

©2004 American Medical Association. All rights reserved. (Reprinted) JAMA, February 18, 2004—Vol 291, No. 7 871

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009

ASSESSING OUTCOME FOR COMATOSE SURVIVORS OF CARDIAC ARREST

Box. Glasgow-Pittsburgh Cerebral Performance Categories* lated sudden muscular contractions and
may be either focal or generalized con-
1. Good Cerebral Performance tractions of axial and limb muscula-
Conscious: Alert, able to work and lead a normal life. May have minor psycho- ture. In patients with seizures, the
logical or neurological deficits (mild dysphasia, nonincapacitating hemiparesis, or physical examination should be re-
minor cranial nerve abnormalities). peated after the postictal period.

2. Moderate Cerebral Disability Finally, mechanically ventilated pa-
Conscious. Sufficient cerebral function for part-time work in sheltered environ- tients are frequently sedated and/or para-
ment or independent activities of daily life (dressing, traveling by public trans- lyzed. Accordingly, when performing a
portation, and preparing food). May have hemiplegia, seizures, ataxia, dsysar- detailed neurological examination it is
thria, dysphasia, or permanent memory or mental changes. crucial that these medications be at least
temporarily discontinued.
3. Severe Cerebral Disability
Conscious. Dependent on others for daily support because of impaired brain func- Outcomes of Interest
tion (in an institution or at home with exceptional family effort). At least limited
cognition. Includes a wide range of cerebral abnormalities from ambulatory with The neurological outcome of coma-
severe memory disturbance or dementia precluding independent existence to para- tose patients is most often described us-
lytic and able to communicate only with eyes, as in the locked-in syndrome. ing the Cerebral Performance Catego-
ries (CPC) 1 to 5 as shown in the BOX.13
4. Coma, Vegetative State
Not conscious. Unaware of surroundings, no cognition. No verbal or psychologi- METHODS
cal interactions with environment.
Search Strategy
5. Death and Quality Review
Certified brain dead or dead by traditional criteria.
We conducted a computerized biblio-
*Adapted with permission from Cummings et al.13 graphic search of MEDLINE and
EMBASE for the years 1966 to 2003
stem reflexes are used in the physical brain and vestibular reflexes, the eyes to determine the precision and accu-
examination of comatose patients.10,12 will move in a direction opposite to that racy of components of the clinical ex-
The pupillary reflex involves cranial in which the head is moved. If this re- amination in prognosis of hypoxic-
nerves II and III. Shining a penlight into flex is lost, the globes will remain fixed ischemic coma. Search terms included
one eye and then the other tests the pa- within the head and the eyes will con- coma, cardiac arrest, prognosis, physi-
tient’s pupillary light response; the ex- tinue to stare in whatever direction the cal examination, sensitivity and specific-
aminer observes the direct and consen- head is pointed. This reflex should not ity, and observer variation. The search
sual response (constriction of the be tested in cases of suspected cervical was conducted using a previously pub-
opposite eye). The corneal reflex in- trauma. Cold water caloric testing (ocu- lished search strategy for The Rational
volves cranial nerves V and VII. Touch- lovestibular reflex) also tests the ves- Clinical Examination series.14 We
ing the cornea with a piece of cotton tibular and oculomotor systems. To checked the reference lists of all re-
or tissue should cause both eyes to perform the test, first examine the tym- view articles and primary studies for ad-
blink. The gag and cough reflexes test panic membrane to ensure there is ditional articles that were not identi-
cranial nerves IX and X. To elicit a gag, no perforation or impacted cerumen. fied on the computerized search.
apply a tongue depressor to the poste- With the head 30° higher than the Standard physical examination text-
rior pharynx. The soft palate should rise horizontal, irrigate up to 120 mL of ice books and personal communications
symmetrically. In patients who are in- cold water into the auditory canal. In with the authors of primary studies pro-
tubated, assess the cough (or carinal) the unconscious patient with intact vided additional citations. Finally, we
reflex by applying deep suction through brainstem function, there will be slow manually reviewed published ab-
the endotracheal tube to the carina. The tonic deviation of eyes towards the ir- stracts from the annual scientific meet-
suction will produce a gasp followed by rigated ear. ings of the American Neurological As-
several rapid coughs. sociation, the American Academy of
It is also important to note the pres- Neurology, the Society of Critical Care
Vestibular signs are also commonly ence of seizures or myoclonus when ex- Medicine, and the European Society for
examined in the comatose patient. The amining the comatose patient, for some Intensive Care Medicine for the years
oculocephalic (or “Doll’s eye”) reflex clinicians believe they may be useful in 1997-2003.
involves observing the patient’s eyes prognosis of comatose survivors of car-
during passive rotation of the skull. In diac arrest. Seizures may be general- One of the authors (C.M.B.) ini-
a comatose patient with intact mid- ized or focal. Myoclonus refers to iso- tially screened the titles and abstracts
of the search results and classified them
as primary studies, review articles, or

872 JAMA, February 18, 2004—Vol 291, No. 7 (Reprinted) ©2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009

ASSESSING OUTCOME FOR COMATOSE SURVIVORS OF CARDIAC ARREST

not relevant. Because we were inter- outcome data in this fashion and be- cal variables at various times after car-
ested in both the precision and accu- cause we could not combine studies that diac arrest. Likelihood ratios were mod-
racy of the clinical examination in post– had differing definitions of good vs poor eled using a method described by Warn
cardiac arrest coma, we included outcomes. Furthermore, we thought it et al22 for relative risks, also using the
primary studies of each type. Because was reasonable to assume that most cli- prior distributions used therein. Post-
a preliminary review of the literature re- nicians, patients, and families would not test probabilities were computed from
vealed very few precision studies, the consider severe neurological disability the estimated pretest probability and
inclusion criteria for this type of study (defined as CPC 3) a desirable out- LRs.23 All analyses were done using the
were broadened. Precision studies were come. WinBUGS software package.24
included if they assessed the interob-
server agreement in the neurological ex- The methodological quality of each Likelihood Ratios
amination of comatose adult patients. primary study was assessed in dupli-
We included both traumatic and non- cate using modified criteria previ- Likelihood ratios are a method of con-
traumatic forms of coma. ously developed for the Rational Clini- verting pretest information (ie, prob-
cal Examination series.21 Because this ability, or more precisely, odds) into
Primary studies of accuracy were in- study was assessing prognosis and not posttest information.25 The pretest in-
dependently reviewed by 2 of us diagnosis, investigators were consid- formation is the probability of a poor
(C.M.B. and R.H.B.) and included if ered blinded if the study was prospec- outcome among all comatose survivors
they assessed the accuracy of the clini- tive and clinical variables were as- of cardiac arrest. The results of the clini-
cal examination in prognosis of hy- sessed before patient outcome was cal examination, reflected in the LRs for
poxic-ischemic coma in patients older known. Level 1 studies were prospec- the findings, are combined with the pre-
than 10 years. Other criteria for study tive studies with 100 or more consecu- test information to estimate the post-
selection were the presentation of out- tive unselected patients. Level 2 stud- test probability of a poor outcome. For
come data for individual clinical vari- ies were similar but involved fewer than clinicians, the easiest way to interpret
ables measured at discrete time inter- 100 patients. Level 3 studies were ret- LRs is to keep in mind that when an ab-
vals. Selected studies also presented rospective chart reviews, and level 4 normal clinical finding is present in a co-
neurological outcome data as defined studies included selected (ie, noncon- matose survivor (eg, absent pupillary
by the Cerebral Performance Catego- secutive) patients. response), the likelihood of a poor out-
ries (CPCs) or in such a manner that come increases and the LR will be greater
an equivalent CPC score could be de- Statistical Methods than 1. Similarly, if the finding does not
termined (Box). Studies were ex- indicate a poor prognosis (eg, present
cluded if they involved patients with Two authors (C.M.B. and R.H.B.) in- pupillary response), an LR of less than
coma from other medical conditions or dependently extracted data for analy- 1 will occur.
trauma. sis; we resolved disagreement by con-
sensus. When data were missing or RESULTS
Based on our findings in a prelimi- unclear, we contacted the primary in-
nary literature search, we realized there vestigators requesting further informa- Search Results and Quality
were 2 types of accuracy studies in the tion. Published raw data were used to of the Evidence
literature. The majority of studies di- calculate positive and negative likeli-
chotomized patient outcome as good or hood ratios (LRs) for specific clinical Our search yielded 5 studies of preci-
poor. Unfortunately there is not a uni- variables. To create 2 ϫ 2 evidence sion that met our inclusion criteria
form definition of what constitutes a tables we dichotomized CPC 1 and 2 (TABLE 2).26-30 Two other studies of pre-
good vs a poor outcome. Most studies as good outcome and CPC 3 through 5 cision were excluded because neither
combined outcome data for severe neu- as poor outcome. Sensitivity was de- rates of agreement (␬) nor raw data
rological disability, vegetative state, and fined as the proportion of patients with were presented.31,32 Fourteen accu-
death (ie, CPC 3-5) as a poor outcome and a poor neurological outcome who had racy articles describing 11 different
normal or moderate disability (ie, CPC a particular physical finding; specific- studies met our inclusion criteria
1-2) as a good outcome. However, there ity was the proportion of patients who (TABLE 3).33-46 We had 100% agree-
were 6 studies that included severe neu- had a good neurological outcome who ment on the inclusion of studies for the
rological disability (ie, CPC 3) as a good did not have the particular finding. systematic review. Reasons for exclud-
outcome; 4 of which included fewer than ing relevant studies included studies
65 patients.15-20 We included only stud- Where 3 or more studies examined that did not present neurological out-
ies from which combined outcome data the same clinical variable at the same comes as CPC 1 and 2 as good and CPC
for severe neurological disability, veg- time after cardiac arrest, we calculated 3 to 5 as poor outcomes,15-20 studies in
etative state, and death (ie, CPC 3-5) summary LRs and 95% confidence in- which patients were not comatose,47-52
could be extracted. We did this be- tervals (CIs) using Bayesian random- studies that included only patients in
cause most primary studies presented effects meta-analyses. We also present persistent vegetative state,53,54 studies
the strongest LRs for individual clini- that included other forms of medical

©2004 American Medical Association. All rights reserved. (Reprinted) JAMA, February 18, 2004—Vol 291, No. 7 873

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009

ASSESSING OUTCOME FOR COMATOSE SURVIVORS OF CARDIAC ARREST

coma,55,56 and studies that presented the We reached 100% agreement on the 2 as level 4. The studies and method-
same data set.3,57 One study was a sys- methodological quality scores. Of the ological quality scores are summa-
11 accuracy studies, 5 were classified rized in Table 3.
tematic review of clinical and neuro- as level 1, 3 as level 2, 1 as level 3, and
physiological variables.6

Table 2. Interobserver Agreement of Clinical Examination for Coma Precision of the Clinical

No. of Observer’s Level Variable Agreement, Examination of Coma
Assessed ␬ Statistic
Source Observers of Experience 0.72 Five studies have reported the preci-
sion of the examination of comatose pa-
Braakman et al,26 1977 12 Neurosurgeons and GCS motor 0.75 tients (Table 2). Heterogeneity in study
residents DR = 14.3% methodology, patient population, and
DR = 5.4% variables assessed, precluded a quan-
20 Neurosurgical nurses GCS motor DR = 10.9% titative synthesis of results; thus, these
DR = 4.3% studies were reviewed qualitatively. As
Teasdale et al,27 1978 7 Neurosurgeons GCS eye 0.49 presented in Table 2, interobserver
agreement was moderate to substan-
GCS verbal 0.46 tial in each of the studies. Three stud-
ies found no difference in interob-
GCS motor 0.65 server agreement between experienced
0.62 nurses, residents, and physicians.26-28
Pupil response 0.59 One study did find precision to be di-
0.68 minished in groups of less experi-
van den Berge et al,28 6 Neurosurgeons Oculocephalic 0.79 enced examiners.30 No study exam-
1979 response 0.74 ined only patients with nontraumatic
causes of coma. In summary, there was
Spontaneous eye 0.65 reasonable consistency among stud-
movement 0.69 ies, and the precision of the clinical ex-
0.70 amination of coma (including compo-
Pupil response 0.36 nents of the GCS and brainstem
0.42 reflexes) has been found to be moder-
Minderhoud et al,29 1982 4 Physicians GCS eye ate to substantial.

GCS verbal

GCS motor

Pupil response

Oculocephalic
response

Born et al,30 1987 6 Neurosurgeons GCS motor

Brainstem score

Pupil response

6 Other physicians GCS motor

Brainstem reflexes

Abbreviations: DR, reported disagreement rate; GCS, Glasgow Coma Scale.

Table 3. Studies of the Accuracy of the Clinical Examination in Prognosis of Hypoxic-Ischemic Coma*

Source Level of Study Population Site of Mean No. of Neurological Outcome
Evidence Arrest Age, y† Patients Outcomes§ Assessment

Good Poor

Berek et al,33 1997 2 Post–cardiac arrest coma PH 68 42 13 29 At discharge

Chen et al,34 1996 4 Patients in hypoxic-ischemic coma PH or IH 58 34 7 27 3 mo࿣

at 24 h

Earnest et al,35 1979 1 Post–cardiac arrest coma PH 62 100 30 70 At discharge

Edgren et al,36 1987 4 Post–cardiac arrest coma PH or IH 71 32 11 21 6 mo࿣

Edgren et al,37 1994‡ 1 Post–cardiac arrest coma PH or IH 58 262 89 173 12 mo࿣

Krumholz et al,38 1988 1 Patients in post–cardiac arrest coma PH or IH 67 114 21 93 At discharge

at 24 h

Levy et al,39 1985 1 Hypoxic-ischemic coma PH or IH 61 210 26 184 12 mo࿣

Madl et al,40 2000 1 Patients in post–cardiac arrest coma PH or IH 57 209 49 160 6 mo࿣

at 24 h

Madl et al,41 1993 2 Post–cardiac arrest coma PH or IH 58 66 17 49 At discharge࿣

Sasser,42 1999‡ 1 Patients in post–cardiac arrest coma PH or IH 63 937 230 707 6 mo࿣

at 12 h

Snyder et al,43-45 1980/1981 2 Post–cardiac arrest coma PH 64 63 25 38 6 mo࿣

Widjiks et al,46 1994 3 Post–cardiac arrest coma PH 63 107 15 92 6 mo

Abbreviations: IH, in-hospital cardiac arrest; PH, prehospital cardiac arrest.

*The 14 sources represent 11 studies
†When the mean age was not provided, the median age of the study population is listed.

‡This article includes patients from the first Brain Resuscitation Clinical Trial (BRCT), also included in Sasser’s dissertation, which involves all 3 BRCTs.

§Good neurological outcome refers to cerebral performance categories (CPCs) 1 and 2. Poor outcome includes CPCs 3 through 5. See the Box for a definition of CPCs.
࿣Outcome refers to best ever CPC in specified time period.

874 JAMA, February 18, 2004—Vol 291, No. 7 (Reprinted) ©2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009

ASSESSING OUTCOME FOR COMATOSE SURVIVORS OF CARDIAC ARREST

Accuracy of the Clinical of cardiac arrest were assessed for the 2915 patients were randomly as-
Examination of Coma use of a barbiturate (thiopental). In signed to receive standard- or high-
BRCT II59 (1984-1989), 516 comatose dose epinephrine during cardiac ar-
Fourteen articles involving 11 differ- patients were randomly assigned to re- rest. All 3 of the BRCT studies reported
ent studies of the accuracy of the clini- ceive placebo or a calcium channel negative results; there was no differ-
cal examination were included (Table blocker (lidoflazine) following car- ence found in survival or neurological
3). These studies provided a sample size diac arrest. In BRCT III42 (1989-1992), outcome among treatment groups. Two
of 1914 comatose survivors of cardiac
arrest. The proportion of individuals dy- Table 4. Useful Clinical Findings in the Prognosis of Post–Cardiac Arrest Coma Organized by
ing or having a poor neurological out- Time After Onset of Coma (Not Pooled)*
come was calculated by pooling the out-
come data from the 11 studies and was LR of Poor Neurological Outcome
used as an estimate of the pretest prob- (95% Confidence Interval)
ability of poor outcome (Table 3). The
random effects estimate of poor out- Clinical Finding Study Positive Negative
come was 77% (95% CI, 72%-80%).
This value represents an estimate of the At Time of Coma Onset 0.5 (0.4-0.6)
pretest probability of death or a poor 0.6 (0.4-0.7)
outcome for the entire population of co- Absent pupillary reflex Earnest et al35 7.2 (1.9-28.0) 0.7 (0.6-0.8)
matose survivors of cardiac arrest, and 0.4 (0.3-0.6)
it is combined with the LRs for vari- Absent motor response Levy et al39 3.5 (1.4-8.6) 0.4 (0.3-0.6)
ous clinical findings to revise the esti- 0.2 (0.1-0.6)
mated probability of a poor clinical Absent corneal reflex Levy et al39 3.2 (1.1-9.5) 0.3 (0.2-0.5)
outcome. 0.1 (0.0-0.7)
Absent oculocephalic reflex Earnest et al35 2.5 (1.3-4.8)
Motor Response and
Brainstem Reflexes Absent spontaneous eye movement Levy et al39 2.2 (1.3-4.0)

Six studies examined the association be- ICS Ͻ4 Berek et al33 2.2 (1.1-4.5)
tween motor and brainstem function
and the recovery of comatose survi- GCS Ͻ5 Madl et al40 1.4 (1.1-1.6)
vors of cardiac arrest. Data for specific
clinical findings were pooled if they Absent verbal effort Levy et al39 1.2 (0.9-1.6)
were assessed in at least 3 studies.
TABLE 4 shows potentially useful clini- Absent cough reflex At 12 Hours 13.4 (4.4-40.3) 0.3 (0.2-0.4)
cal findings from individual studies. Absent corneal reflex Sasser42 9.1 (3.9-21.1) 0.3 (0.2-0.4)
Summary measures for pooled vari- Absent gag reflex Sasser42 8.7 (4.0-18.9) 0.4 (0.4-0.5)
ables are shown in TABLE 5. Absent pupillary reflex Sasser42 4.0 (2.5-6.6) 0.5 (0.5-0.6)
GCS Ͻ5 Sasser42 3.5 (2.4-5.2) 0.4 (0.3-0.4)
In 1987, Edgren et al36 reported mo- Absent motor response Sasser42 3.2 (2.2-4.6) 0.4 (0.3-0.5)
tor and brainstem function in 32 co- Absent withdrawal to pain Sasser42 2.4 (1.9-3.1) 0.2 (0.1-0.2)
matose patients at 24 and 48 hours af- Absent verbal effort Sasser42 1.6 (1.4-1.9) 0.1 (0.0-0.1)
ter cardiac arrest. It is important to note Sasser42
that patients were weaned from inten-
sive care at 72 hours if they did not re- Absent cough reflex At 24 Hours 84.6 (5.3-1342.0) 0.4 (0.3-0.5)
spond to pain and had no evidence of Absent gag reflex Sasser42 24.9 (6.3-98.3) 0.5 (0.4-0.5)
brainstem reflexes. Chen and col- GCS Ͻ5 Sasser,42 0.4 (0.3-0.4)
leagues34 examined similar clinical vari- Absent eye opening to pain Sasser42 8.8 (5.1-15.1) 0.3 (0.3-0.4)
ables in a study of 34 comatose pa- Absent spontaneous eye movement Sasser42 5.9 (3.9-9.0) 0.5 (0.4-0.7)
tients. As in Edgren’s study patients Absent eye opening to pain Levy et al39 3.5 (1.4-8.8) 0.4 (0.3-0.5)
with absent brainstem reflexes at 24 Absent oculocephalic reflex Levy et al39 3.0 (1.5-6.2) 0.5 (0.5-0.6)
hours were excluded from this study. Absent spontaneous eye movement Sasser42 2.9 (1.8-4.6) 0.3 (0.2-0.3)
Absent verbal effort Sasser42 2.7 (2.1-3.4) 0.1 (0.0-0.1)
The Brain Resuscitation Clinical Sasser42 2.4 (2.0-2.9)
Trials (BRCTs) were a series of 3 large
prospective, randomized, multicenter GCS Ͻ6 At 48 Hours 2.8 (1.3-5.9) 0.3 (0.1-0.5)
studies of pharmacologic interven- GCS Ͻ10 Madl et al41 1.3 (1.0-1.7) 0.0 (0.0-0.7)
tions in cardiac arrest. In BRCT I58 Madl et al41
(1979-1984), 262 comatose survivors
Absent withdrawal to pain At 72 Hours 36.5 (2.3-569.9) 0.3 (0.2-0.4)
Absent spontaneous eye movement Levy et al39 11.5 (1.7-79.0) 0.6 (0.5-0.7)
Absent verbal effort Levy et al39 0.3 (0.2-0.5)
Absent eye opening to pain Levy et al39 7.4 (2.0-28.0) 0.5 (0.4-0.6)
Levy et al39 6.9 (1.8-27.0)

At 7 Days

Absent withdrawal to pain Levy et al39 29.7 (1.9-466.0) 0.4 (0.3-0.6)

Absent verbal effort Levy et al39 14.1 (2.0-97.7) 0.4 (0.2-0.6)

Abbreviations: GCS, Glasgow Coma Scale; ICS, Innsbruck Coma Scale33; LR, likelihood ratio.

*Clinical findings that have a positive LR greater than 2 and lower confidence interval boundary greater than 1 are
presented with the corresponding negative LR.

©2004 American Medical Association. All rights reserved. (Reprinted) JAMA, February 18, 2004—Vol 291, No. 7 875

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009

ASSESSING OUTCOME FOR COMATOSE SURVIVORS OF CARDIAC ARREST

articles described the association be- tion. In 1994, Edgren et al37 reported I who had survived to 72 hours. In an
tween clinical neurological signs and the neurological examination and out- analysis of all 3 BRCT studies, Sasser42
outcome in the BRCT study popula- comes of the 109 individuals in BRCT assessed the prognostic utility of mo-
tor response and brainstem reflexes at
Table 5. Pooled Clinical Signs in the Prognosis of Post–Cardiac Arrest Coma 12 and 24 hours after cardiac arrest. As
in all studies of cardiac arrest, there was
LR of Poor Neurological Outcome (95% a high degree of early mortality. Ac-
Confidence Interval) cordingly, only 1450 patients of the
original 3693 studied in all 3 BRCTs
Source Positive Negative survived to 12 hours. Of this group, 506
At Time of Coma Onset* patients were sedated or anesthetized
Absent Withdrawal to Pain 0.4 (0.1-1.1) at the time of neurological examina-
Summary LR 1.7 (0.7-4.2) 0.4 (0.3-0.6) tion and therefore were not included in
Earnest et al35 3.7 (1.6-8.2) 0.4 (0.2-0.7) Sasser’s review. Of the remaining 944
Levy et al39 1.4 (1.0-1.9) 0.5 (0.2-1.2) patients, outcome data was available for
Snyder et al43 1.4 (0.9-2.1) 937. This is the largest population of
At 24 Hours 0.2 (0.1-0.6) comatose survivors of cardiac arrest re-
0.4 (0.2-0.8) ported to date.
Absent Withdrawal to Pain 4.7 (2.2-9.8) 0.2 (0.2-0.3)
Summary LR 3.9 (1.1-14) 0.2 (0.1-0.2) Summary measures for clinical vari-
Edgren et al36 6.8 (2.3-19.8) 0.3 (0.1-0.7) ables that were assessed in at least 3
Levy et al39 5.1 (3.6-7.3) studies are presented in Table 5. Five
Sasser42 6.5 (1.0-42.0) 0.8 (0.4-1.4) pooled variables were found to have a
Snyder et al43 1.0 (0.8-1.2) 95% CI lying entirely above 1. The clini-
10.2 (1.8-48.6) 0.8 (0.6-1.1) cal signs at 24 hours with the highest
Absent Pupil Response 0.9 (0.0-19.1) 0.8 (0.7-0.9) LRs were absent corneal reflexes (LR,
Summary LR 5.6 (0.3-95.0) 0.6 (0.6-0.7) 12.9; 95% CI, 2.0-68.7), absent pupil-
Chen et al34 lary reflexes (LR, 10.2; 95% CI,
Edgren et al36 10.7 (0.7-170.0) 0.6 (0.3-1.3) 1.8-48.6), absent motor response (LR,
Levy et al39 39.2 (5.6-276.6) 0.8 (0.6-1.1) 4.9; 95% CI, 1.6-13.0), and absent with-
Sasser42 0.6 (0.5-0.8) drawal to pain (LR, 4.7; 95% CI,
4.9 (1.6-13.0) 0.4 (0.3-0.4) 2.2-9.8). At 72 hours after cardiac ar-
Absent Motor Response 3.7 (0.2-59.1) 0.7 (0.5-1.1) rest, absent motor response was found
Summary LR 5.5 (1.4-21.0) to accurately predict death or poor neu-
Chen et al34 7.6 (4.6-12.6) 0.6 (0.2-1.9) rological outcome (LR, 9.2; 95% CI,
Levy et al39 3.5 (0.5-24.3) 0.9 (0.7-1.2) 2.1-49.4). No clinical findings were
Sasser42 0.7 (0.7-0.8) found to accurately predict good neu-
Snyder et al43 12.9 (2.0-68.7) 0.4 (0.4-0.5) rological outcome (ie, no useful nega-
1.8 (0.2-15.4) tive LRs).
Absent Corneal Reflex 0.9 (0.4-2.1)
Summary LR 14.8 (0.9-233.0) 1.0 (0.8-1.2) Coma Scales
Edgren et al36 90.9 (5.7-1442.9) 0.8 (0.7-1.0)
Levy et al39 0.9 (0.8-1.0) Four studies assessed composite coma
Sasser42 scores as prognostic indicators in post–
0.7 (0.3-1.3) cardiac arrest coma. Madl and col-
At 72 Hours 0.9 (0.7-1.2) leagues41 reported 2 studies that as-
0.6 (0.5-0.7) sessed the role of the GCS in predicting
Absent Pupil Response 3.4 (0.5-23.6) 0.7 (0.6-0.8) neurological recovery. In 1993 this
Summary LR 0.6 (0.3-1.1) group reported on a series of 66 coma-
tose patients who survived cardiac ar-
Chen et al34 0.9 (0.0-19.1) 0.8 (0.3-2.1) rest.41 The GCS at 48 hours was com-
0.7 (0.5-1.0) pared with survival and functional
Edgren et al37 5.3 (0.3-84.0) 1.0 (0.8-1.2) recovery. A second study of 209 pa-
0.8 (0.6-1.1) tients measured GCS on admission to
Levy et al39 5.8 (0.4-94.0) the intensive care unit following car-
diac arrest.40 In the BRCT reports, GCS
Absent Motor Response 9.2 (2.1-49.4) scores at 12, 24, and 72 hours were
Summary LR

Chen et al34 2.0 (0.1-34.8)

Edgren et al37 12.6 (0.8-193.0)

Levy et al39 16.5 (1.1-261.0)

Snyder et al43 3.0 (0.2-38.8)

Seizure or Myoclonus† 1.4 (0.5-3.9)
Summary LR

Krumholz et al38 1.7 (0.8-3.4)

Levy et al39 1.1 (0.5-2.3)

Snyder et al44 1.7 (0.7-4.2)

Abbreviation: LR, likelihood ratio.

*Times reflect number of hours since cardiac arrest.
†These figures refer to the presence of seizures or myoclonus at any time after cardiac arrest.

876 JAMA, February 18, 2004—Vol 291, No. 7 (Reprinted) ©2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009

ASSESSING OUTCOME FOR COMATOSE SURVIVORS OF CARDIAC ARREST

compared with neurological recov- rological outcome. In discussing this signs accurately predict the patient’s
ery.37,42 In 1997, Berek et al33 exam- with the family, it is important to ex- outcome. Finally, no clinical findings
ined the utility of Innsbruck Coma Scale plain that although there are no signs were found to have LRs that strongly
in 42 comatose patients who survived suggestive of poor outcome, the physi- predicted good neurological outcome.
prehospital arrest.33 The Innsbruck cal examination is much less useful in
Coma Scale (ICS) includes an assess- predicting good outcome. Conse- The results of our meta-analysis
ment of the GCS components in addi- quently, his probability of poor neuro- should be interpreted in the context of
tion to various brainstem reflexes. A logical outcome remains unchanged (ie, study limitations. To calculate LRs from
score from 0 to 23 is assigned. A lower 77%). 2ϫ2 tables, there must be a delinea-
score indicates more severe neurologi- tion between what constitues a good vs
cal deficits. In the second case, the young woman a poor neurological outcome. We chose
has no brainstem reflexes or response to define poor outcome as death, veg-
Although the composite coma scores to painful stimuli at 3 days. Unfortu- etative state, or severe neurological im-
did predict poor neurological out- nately these findings all suggest an ex- pairment (precluding independent
come, they were not as predictive as the tremely poor chance of meaningful neu- living). We made this decision be-
individual motor and brainstem reflex rological recovery. The most powerful cause that is where most primary stud-
components. This is demonstrated in of these indicators elevates her post- ies dichotomize outcome. Further-
Table 4. test probability of poor neurological more, we believe most patients, families,
outcome to 97%. Although the exist- and physicians would not consider se-
Seizures ing literature does not examine the vere neurological impairment to be a
combined effects of different physical desirable outcome. However, in apply-
Four studies have examined whether examination predictors, because she has ing the results of this study to indi-
seizures in the postarrest period accu- multiple poor prognostic findings her vidual patients, physicians must real-
rately predict outcome. In 1988 Krum- prognosis may be even worse. You note ize that some families and patients may
holz et al38 described 114 comatose sur- that the observation of reactive pupils have different perceptions of what con-
vivors of cardiac arrest. Nearly half of immediately following cardiac arrest situtes an acceptable neurological out-
the patients (44%) had some seizure ac- and the presence of myoclonus are not come. It was not the purpose of this
tivity. In Snyder’s study44 of 63 pa- useful in determining her neurologi- study to provide an ethical framework
tients, 19 (30%) of 63 had seizures or cal prognosis. for treatment decisions in the manage-
myoclonus. In 1994 Widjiks and col- ment of comatose survivors of cardiac
leagues46 described the prevalence of THE BOTTOM LINE arrest; rather, we attempted to summa-
myoclonus status in a group of 107 pa- rize the exisiting literature to provide
tients. Forty (37%) of the 107 patients In this systematic review we found that guidance to clinicians and families
had myoclonus status within 24 hours. the precision of the neurological ex- about prognostic probabilities.
In Levy’s study39 of 210 patients, 53 amination in comatose patients is mod-
(25%) had seizure or myoclonic activ- erate to substantial. Based on our re- Any study of prognosis in the criti-
ity. Most clinicians infer that seizures sults, we suggest that in patients who cally ill is potentially influenced by the
portend a poor prognosis in comatose lack pupillary and corneal reflexes at tendency for poor prognoses to be self-
survivors of cardiac arrest. However, 24 hours and have no motor response fulfilling. It is difficult to determine if
none of the individual studies or the at 72 hours, the chance of meaningful poor neurological outcomes are caused
summary measures established that neurological recovery is extremely by decisions to withdraw or withhold
seizures accurately predict outcome small. This meta-analysis includes al- therapy based on a perceived poor neu-
(Table 5). most 2000 patients and is the largest rological prognosis. This has the po-
such review to date. In addition to pro- tential to artificially elevate positive LRs.
SCENARIO RESOLUTION viding other information, it corrobo- Although there is no empirical evi-
rates the findings of the oft-quoted Levy dence that this occurred in our study
In both cases an estimate of the pre- study39 in which none of the 210 pa- population, this clinical reality does re-
test probability (derived from our over- tients who had any of these 3 clinical main a limitation of the existing litera-
all study population) of poor neuro- findings ever regained an indepen- ture.
logical outcome is 77%. (It is important dent lifestyle.
to recognize that this figure will vary It would be potentially useful to as-
based on comorbid disease, duration of In our study population the ran- sess whether combinations of neuro-
cardiopulmonary resuscitation, and dom effects estimate of poor outcome logical findings could improve the ac-
other clinical variables.) The 65-year- was 77% (95% CI, 72%-80%). The curacy of prognosis in comatose
old man who withdraws to pain and has highest LR increases the pretest prob- survivors of cardiac arrest. Unfortu-
intact brainstem reflexes 24 hours af- ability of 77% to a posttest probability nately we were unable to perform this
ter cardiac arrest has none of the clini- of 97% (95% CI, 87%-100%). Imme- analysis because the available litera-
cal findings associated with poor neu- diately after cardiac arrest, no clinical ture does not provide this data. In 3

©2004 American Medical Association. All rights reserved. (Reprinted) JAMA, February 18, 2004—Vol 291, No. 7 877

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009

ASSESSING OUTCOME FOR COMATOSE SURVIVORS OF CARDIAC ARREST

studies, combinations of findings were Analysis and interpretation of data: Booth, Boone, 17. Longstreth WT Jr, Diehr P, Inui TS. Prediction of
assessed. In the analysis of 262 pa- Tomlinson, Detsky. awakening after out-of-hospital cardiac arrest. N Engl
tients by Edgren et al,37 no combina- Drafting of the manuscript: Booth, Tomlinson, Detsky. J Med. 1983;308:1378-1382.
tion of findings was found to be more Critical revision of the manuscript for important in- 18. Mullie A, Verstringe P, Buylaert W, et al, for the
predictive than the individual vari- tellectual content: Booth, Boone, Tomlinson, Detsky. Cerebral Resuscitation Study Group of the Belgian So-
ables. Sasser,42 who performed a very Statistical expertise: Booth, Tomlinson, Detsky. ciety for Intensive Care. Predictive value of Glasgow
detailed analysis of combined neuro- Obtained funding: Detsky coma score for awakening after out-of-hospital car-
logical findings and demographic, co- Administrative, technical, or material support: Booth, diac arrest. Lancet. 1988;1:137-140.
morbidity, and cardiopulmonary re- Boone, Detsky. 19. Pifferi S, Codazzi D, Savioli M, Langer M. Early
suscitation variables, did not find any Supervision: Detsky. prediction of neurologic prognosis after post-anoxic
additional predictive value of the algo- Funding/Support: This project was partially funded coma. Intensive Care Med. 1998;24:535-536.
rithm (sensitivity 59%, specificity 93%). through the financial support of Griffen Rotman. 20. Sacco RL, VanGool R, Mohr JP, Hauser WA. Non-
Only Levy et al39 found practical and Acknowledgment: We thank David Simel, MD, for traumatic coma: Glasgow Coma Score and coma eti-
useful algorithms that combined vari- his valuable guidance during the course of this study. ology as predictors of 2-week outcome. Arch Neu-
ous neurological findings. These are We also thank Tom Stewart, MD, Marika Hohol, MD, rol. 1990;47:1181-1184.
clearly presented in their article. Karen Welty-Wolfe, MD, James Tulsky, MD, and How- 21. Holleman DR, Simel DL. Does the clinical exami-
ell Sasser, PhD, for their expert advice and comments nation predict airflow limitation? JAMA. 1995;273:
Finally, it is important to note that on this article. 313-319.
the 11 studies included in this meta- 22. Warn DE, Thompson SG, Spiegelhalter DJ.
analysis represent a diverse and hetero- REFERENCES Bayesian random effects meta-analysis of trials with
geneous population with various co- binary outcomes: methods for the absolute risk dif-
morbidities. It is also unclear what effect 1. Eisenberg MS, Mengert TJ. Cardiac resuscitation. ference and relative risk scales. Stat Med. 2002;21:
individual medications and/or hypo- N Engl J Med. 2001;344:1304-1313. 1601-1623.
thermic cooling may have on the bed- 2. Becker LR, Ostrander MP, Barrett J, Kondos GT. 23. Sackett DL, Straus SE, Richardson WS, Rosen-
side clinical examination. Conse- Outcome of CPR in a large metropolitan area. Ann berg W, Haynes RB. Evidence-Based Medicine: How
quently, the applicability of our results Emerg Med. 1991;20:355-361. to Practise and Teach EBM. 2nd ed. Edinburgh, Scot-
to individual patients must be made 3. Bates D, Caronna JJ, Cartlidge NE, et al. A pro- land: Churchill Livingstone; 2000.
with caution and as part of the larger spective study of nontraumatic coma: methods and 24. WinBUGS [computer program]. Version 1.4. Cam-
clinical picture. We do not suggest a di- results in 310 patients. Ann Neurol. 1977;2:211- bridge, England: MRC Biostatistices Unit; 2003.
rect extension of our results to the de- 220. 25. Sackett DL. A primer on the precision and accu-
cision to proceed with or withdraw 4. Shewmon DA, De Giorgio CM. Early prognosis in racy of the clinical examination. JAMA. 1992;267:
from medical care. Rather, we present anoxic coma. Neurol Clin. 1989;7:823-843. 2638-2644.
information that we hope will allow the 5. Elheff SM, Hanly DF. Postresuscitation prognosti- 26. Braakman R, Avezaat CJ, Maas AI, Roel M,
decision to be made on a more ratio- cation and declaration of brain death. In: Paradis NA, Schouten HJ. Inter observer agreement in the
nal basis. Halperin HR, Novak RA, eds. Cardiac Arrest: The Sci- assessment of the motor response of the Glasgow
ence and Practice of Resuscitation Medicine. Balti- Coma Scale. Clin Neurol Neurosurg. 1977;80:100-
In summary, simple physical exami- more, Md: Williams & Wilkins; 1996:910-922. 106.
nation maneuvers strongly predict 6. Zandbergen EG, de Haan RJ, Stoutenbeck CP, 27. Teasdale G, Knill-Jones R, Van der Sande J. Ob-
death or poor neurological outcome in Loelman JH, Kijdra A. Systematic review of early pre- server variability in assessing impaired consciousness
comatose survivors of cardiac arrest. Al- diction of poor outcome in anoxic-ischaemic coma. and coma. J Neurol Neurosurg Psychiatry. 1978;41:
though decisions to proceed with care Lancet. 1998;352:1808-1812. 603-610.
or withdraw care may take place at later 7. Robinson LR, Micklesen PJ, Tirschwell DL, Lew HL. 28. van den Berge JH, Schouten HJ, Boomstra S, van
times for a variety of reasons, the most Predictive value of somatosensory evoked potentials Drunen Littel S, Braakman R. Interobserver agree-
useful signs occur at least 24 hours and for awakening from coma. Crit Care Med. 2003;31: ment in assessment of ocular signs in coma. J Neurol
in the case of motor response at 72 960-967. Neurosurg Psychiatry. 1979;42:1163-1168.
hours post–cardiac arrest. The exist- 8. Zingler VC, Krumm B, Bertsch T, Fassbender K, Pohl- 29. Minderhoud JM, Huizenga J, van Woerkom TC,
ing literature does not allow for an ear- mann-Eden B. Early prediction of neurological out- Blomjous CE. The pattern of recovery after severe head
lier prognosis to be made on the basis come after cardiopulmonary resuscitation: a multi- injury. Clin Neurol Neurosurg. 1982;84:15-28.
of the clinical examination alone. modal approach combining neurobiochemical and 30. Born JD, Hans P, Albert A, Bonnal J. Interob-
electrophysiological investigations may provide high server agreement in assessment of motor response and
Author Affiliations: Departments of Medicine (Drs prognostic certainty in patients after cardiac arrest. Eur brain stem reflexes. Neurosurgery. 1987;20:513-
Booth, Boone, Tomlinson, and Detsky), and Health Neurol. 2003;49:79-84. 517.
Policy Management and Evaluation (Dr Detsky), Uni- 9. Safar P, Behringer W, Bottiger BW, Sterz F. Cere- 31. Teasdale G, Jennett B. Assessment and progno-
versity of Toronto, and Department of Medicine, bral resuscitation potentials for cardiac arrest. Crit Care sis of coma after head injury. Acta Neurochir (Wien).
Mount Sinai Hospital (Drs Tomlinson and Detsky) and Med. 2002;30(suppl):S140-S144. 1976;34:45-55.
University Health Network (Drs Tomlinson and Detsky), 10. Plum F, Posner JB. The Diagnosis of Stupor and 32. Van Gijn J, Bonke B. Interpretation of plantar re-
Toronto, Ontario. Coma. 2nd ed. Philadelphia, Pa: FA Davis; 1972. flexes: biasing effect of other signs and symptoms.
Author Contributions: Study concept and design: 11. Teasdale G, Jennett B. Assessment of coma and J Neurol Neurosurg Psychiatry. 1977;40:787-789.
Booth, Tomlinson, Detsky. impaired consciousness: a practical scale. Lancet. 1974; 33. Berek K, Schinneri A, Traweger C, Lechleitner P,
Acquisition of data: Booth, Boone. 2:81-84. Baubin M, Aichner F. The prognostic significance of
12. Haerer AF. DeJong’s: The Neurologic Examina- coma-rating, duration of anoxia and cardiopulmo-
tion. 5th ed. Philadelphia, Pa: JB Lipincott Co; 1992. nary resuscitation in out-of-hospital cardiac arrest.
13. Cummings RO, Chamberlain DA, Abramson NS, J Neurol. 1997;244:556-561.
et al. Recommended guidelines for uniform report- 34. Chen R, Bolton CF, Young GB. Prediction of out-
ing of data from out-of-hospital cardiac arrest: the Ut- come in patients with anoxic coma: a clinical and elec-
stein style. Circulation. 1991;84:960-975. trophysiologic study. Crit Care Med. 1996;24:672-
14. Lederle FA, Simel DL. Does this patient have ab- 678.
dominal aortic aneurysm? JAMA. 1999;281:77-82. 35. Earnest MP, Breckinridge JC, Yarnell PR, Olivia
15. Bassetti C, Bomio F, Mathis J, Hess CW. Early prog- PB. Quality of survival after out-of-hospital cardiac ar-
nosis in coma after cardiac arrest: a prospective clini- rest: predictive value of early neurologic evaluation.
cal, electrophysiological, and biochemical study of 60 Neurology. 1979;29:56-60.
patients. J Neurol Neurosurg Psychiatry. 1996;61: 36. Edgren E, Hedstrand U, Nordin M, Rydin E, Ron-
610-615. quist G. Prediction of outcome after cardiac arrest. Crit
16. Bertini G, Margheri M, Giglioli C, et al. Prognos- Care Med. 1987;15:820-825.
tic significance of early clinical manifestations in post- 37. Edgren E, Hedstrand U, Kelsey S, Sutton-Tyrell K,
anoxic coma: a retrospective study of 58 patients re- Safar P. Assessment of neurological prognosis in co-
suscitated after prehospital cardiac arrest. Crit Care matose survivors of cardiac arrest. Lancet. 1994;343:
Med. 1989;17:627-633. 1055-1059.

878 JAMA, February 18, 2004—Vol 291, No. 7 (Reprinted) ©2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009

ASSESSING OUTCOME FOR COMATOSE SURVIVORS OF CARDIAC ARREST

38. Krumholz A, Stern BJ, Weiss HD. Outcome from wenson RB, Ramirez-Lassepas M. Neurologic prog- cal findings after cardiac arrest to outcome. BMJ. 1974;
coma after cardiopulmonary resuscitation: relation to nosis after cardiopulmonary arrest, IV: brainstem re- 3:437-439.
seizures and myoclonus. Neurology. 1988;38:401- flexes. Neurology. 1981;31:1092-1097. 53. Andrews K. Recovery of patients after four months
405. 46. Widjiks EF, Parisi JE, Sharbrough FW. Prognostic or more in the persistent vegetative state. BMJ. 1993;
39. Levy DE, Caronna JJ, Singer BH, Lapinski RH, Fry- value of myoclonus status in comatose survivors of car- 306:1597-1600.
dman H, Plum F. Predicting outcome from hypoxic- diac arrest. Ann Neurol. 1994;35:239-243. 54. Dougherty JH Jr, Rawlinson DG, Levy DE, Plum
ischemic coma. JAMA. 1985;253:1420-1426. 47. Grubb NR, Elton RA, Fox KA. In-hospital mortal- F. Hypoxic-ischemic brain injury and the vegetative
40. Madl C, Kramer L, Domanovits H, et al. Im- ity after out-of-hospital cardiac arrest. Lancet. 1995; state: clinical and neuropathologic correlation. Neu-
proved outcome prediction in unconscious cardiac ar- 346:417-421. rology. 1981;31:991-997.
rest survivors with sensory evoked potential com- 48. Kentsch M, Stendel M, Berkel H, Mueller-Esch 55. Hung TP, Chen ST. Prognosis of deeply coma-
pared with clinical assessment. Crit Care Med. 2000; G. Early prediction of prognosis in out-of-hospital tose patients on ventilators. J Neurol Neurosurg Psy-
28:721-726. cardiac arrest. Intensive Care Med. 1990;16:378- chiatry. 1995;58:75-80.
41. Madl C, Grimm G, Kramer L, et al. Early predic- 383. 56. Kano T, Shimoda O, Morioka T, Yagishita Y,
tion of individual outcome after cardiopulmonary re- 49. Snyder BD, Ramirez-Lassepas M, Lippert DM. Hashiguchi A. Evaluation of the central nervous func-
suscitation. Lancet. 1993;341:855-858. Neurologic status and prognosis after cardiopulmo- tion in resuscitated comatose patients by multilevel
42. Sasser H. Association of clinical signs with neu- nary arrest, I: a retrospective study. Neurology. 1977; evoked potentials. Resuscitation. 1992;23:235-248.
rological outcome after cardiac arrest [dissertation]. 27:807-811. 57. Levy DE, Bates D, Caronna JJ, et al. Prognosis in
Pittsburgh, Pa: University of Pittsburgh; 1999. 50. Steen-Hansen JE, Hansen NN, Vaagenes P, nontraumatic coma. Ann Intern Med. 1981;94:293-
43. Snyder BD, Loewenson RB, Gumnit RJ, Hauser A, Schreiner B. Pupil size and light reactivity during car- 301.
Leppik IE, Ramirez-Lassepas M. Neurologic progno- diopulmonary resuscitation: a clinical study. Crit Care 58. Brain Resuscitation Clinical Trial I Study Group.
sis after cardiopulmonary arrest, II: level of conscious- Med. 1988;16:69-70. Randomized clinical study of thiopental loading in co-
ness. Neurology. 1980;30:52-58. 51. Thompson RJ, McCullough PA, Kahn JK, matose survivors of cardiac arrest. N Engl J Med. 1986;
44. Snyder BD, Hauser A, Loewenson RB, Leppik IE, O’Neill WW. Prediction of death and neurologic 314:397-403.
Ramirez-Lassepas M, Gumnit RJ. Neurologic progno- outcome in the emergency department in out-of- 59. Brain Resuscitation Clinical Trial II Study Group.
sis after cardiopulmonary arrest, III: seizure activity. hospital cardiac arrest survivors. Am J Cardiol. 1998; A randomized clinical study of a calcium-entry blocker
Neurology. 1980;30:1292-1297. 81:17-21. (lidoflazine) in the treatment of comatose survivors of
45. Snyder BD, Gumnit RJ, Leppik IE, Hauser A, Loe- 52. Willoughby JO, Leach BG. Relation of neurologi- cardiac arrest. N Engl J Med. 1991;324:1225-1231.

What is needed most in architecture today is the very
thing that is most needed in life—Integrity.

—Frank Lloyd Wright (1869-1959)

©2004 American Medical Association. All rights reserved. (Reprinted) JAMA, February 18, 2004—Vol 291, No. 7 879

Downloaded from www.jama.com at Ottwa Hospital on August 27, 2009