Elective intubation for neurologic deterioration after stroke

Neurologic deterioration, often accompanied by cerebral edema, is a complication of severe ischemic stroke. If uncontrolled, cerebral swelling raises intracranial pressure (ICP) and will usually lead to brain herniation, with death ensuing from compression of vital brainstem centers controlling respiration, automatic functions, and arousal. Treatment for imminent or incipient herniation includes intubation and mechanical ventilation, the purposes of which are respiratory support to ensure adequate oxygenation. Hyperventilation is usually instituted along with intubation in order to reduce the partial pressure of carbon dioxide in arterial blood, causing constriction of cerebral blood vessels, with consequent reduction of cerebral blood volume and thus intracranial volume, and concomitant reduction of ICP.

Although intubation and hyperventilation are routine--though heroic--measures in patients who deteriorate from cerebral swelling after stroke, their efficacy in reducing mortality and, more importantly, in salvaging patients who go on to make a meaningful recovery have never been assessed. Hyperventilation in particular may have adverse effects by reducing perfusion to penumbral brain regions. In other words, this form of "cerebral resuscitation" has never been submitted to the same sort of analysis of morbidity and mortality recently applied to "cardiac resuscitation" ^[1]. Consequently, when faced with deciding whether or not to intubate a patient, or when counseling the family or patient executing an advance directive, the clinician has no hard data for guidance. The recent well-publicized case of Richard Nixon, who died of cerebral edema without being intubated 4 days after suffering a massive embolic left middle cerebral artery (MCA) stroke, is an illustrative example.

Whether or not to intubate is usually a critical point in decision-making in the setting of deterioration after stroke, because the decision not to intubate and support ventilation will almost always lead to death within the next 24 to 48 hours, whereas mechanical ventilation may result in only vegetative survival. The number of patients salvaged who have meaningful neurologic recovery is unknown, but is probably small. Furthermore, the variables that may be useful in identifying these recovering patients, such as age, comorbidities, type of stroke, stroke location and size, and time and rate of neurologic deterioration, are unknown. Finally, the process and effect of the decision on the family have not been addressed.

Because of the pivotal importance of the decision to intubate, and the lack of existing data on outcome, we decided to assess the mortality, morbidity, outcome, and current status of those patients who were recently electively intubated and mechanically ventilated for infarct-related neurologic deterioration at our stroke center. We identified variables predicting good outcome and assessed the process of decision-making as well as the effect of this decision on the family. Our hope is that these data will help guide decisions regarding the desirability of intubation and ventilation as therapeutic modalities and will stimulate further study of decision-making and outcome in these desperate situations.

Methods. We reviewed the last 250 acute carotid distribution ischemic infarct cases seen by the Department of Neurology Stroke Service, which screens all stroke admissions to our 608-bed teaching hospital. We identified a total of 20 patients who were electively intubated in an effort to control neurologic deterioration. "Elective" means that only patients intubated after CT and neurologic assessment were included; those patients who were intubated by paramedics or immediately upon arrival in the emergency room prior to CT were not included in this analysis. The first 11 patients were identified by a retrospective chart review of the previous 100 carotid distribution stroke cases. The remaining nine patients were identified by an ongoing prospective review (150 cases). If the patient was alive, the patient's primary caregiver was defined as the person who lives in the same household as the patient with primary responsibility for the patient's welfare and was head of the household.

In most cases, intubation took place after neurologic consultation. "Neurologic deterioration" was defined as a decline in neurologic status thought to be due to massive infarction of the cerebral hemisphere with or without brainstem compression from cerebral edema, not due to associated metabolic conditions, hypotension, seizure, or sedative drugs, and severe enough to require intubation in order to prevent hypoxemia or hypercarbia. Patients intubated because of primary respiratory problems such as aspiration, pneumonia, or pulmonary embolus were not included. All patients received "standard" neurologic ICU management in the same unit managed by the same physicians and nurses, but no set protocol was followed. All patients were hyperventilated to some extent and most received mannitol in standard doses. ICP monitors were not placed. Only one patient had surgical decompression (extensive hemicraniectomy).

Clinical course, neurologic outcome, and financial and psychosocial effect of the decision to intubate were determined by chart review, follow-up examination, and telephone or personal interview of the family/caregivers, using case report forms and self-completed questionnaires. The case report form was used to attempt to identify clinical, laboratory, and radiographic features related to outcome. Two self-completed questionnaires were developed ^[2]. One was completed by the next of kin of patients who died, in order to determine the process by which the decision was made to intubate the patient and the family's current perception of the appropriateness of their decision in light of the outcome. The other was completed by the patient and/or family in cases where the patient survived. In addition to the questions specified above, it incorporated the Quality of Life Questionnaire ^[3] and the Relatives Stress Scale ^[4] to characterize the patient's current level of function as perceived by the caregiver and to assess possible psychologic morbidity among caregivers secondary to the stresses of caring for a patient who has suffered a severe brain insult.

The Relatives Stress Scale ^[4] has been previously designed to measure the level of stress of people who take care of a family member who has had a serious illness. It consists of 15 items (questions), each graded from 0 to 2 points (not at all = 0, somewhat = 1, very much = 2) as follows:

(1) Do you ever feel you can no longer cope with the situation?

(2) Do you ever feel you need a break?

(3) Do you ever get depressed by the situation?

(4) Has your own health suffered at all?

(5) Do you worry about accidents happening to the patient?

(6) Do you ever feel that there will be no end to the problem?

(7) Do you find it difficult to go away on vacation?

(8) How much has your social life been affected?

(9) How much has the household routine been upset?

(10) Is your sleep interrupted by the patient's needs?

(11) Has your standard of living been reduced?

(12) Do you ever feel embarrassed by the patient?

(13) Are you prevented from having visitors?

(14) Do you ever get upset and angry with the patient?

(15) Do you ever feel frustrated with the patient?

Our investigators conducted telephone and personal interviews with patients and their caregivers to assess the patient's current status by means of the Modified Rankin Disability Scale ^[5] and the Barthel Index ^[6].

The Modified Rankin Disability Scale ^[5] has a total of six degrees from no symptoms to severe disability (0 to 5) as follows: 0 = No symptoms. 1 = No significant disability despite symptoms; able to carry out all usual duties of daily living. 2 = Slight disability; unable to carry out some previous activities, but able to look after own affairs without assistance. 3 = Moderate disability; requiring some help, but able to walk without assistance. 4 = Moderately severe disability; unable to walk without assistance and unable to attend to own bodily needs without assistance. 5 = Severe disability; bedridden, incontinent, and requiring constant nursing care and attention.

The Barthel Index ^[6] is an 11-item form that gives graded scores (in multiples of 5) to each item for a total possible score of 0 to 100, where full credit is not given for an activity if the patient needs even minimal help or supervision. A score of 0 is given when the patient cannot meet criteria as defined. The items with a maximum score of 15 are moving (transfer) and walking on level surface; with a maximum score of 10 are feeding, getting on and off the toilet, ascending and descending stairs, dressing/undressing, continence of bowels, controlling bladder; and with a maximum score of 5 are personal toilet, bathing self, propelling a wheelchair (not scored if patient gets a score for walking).

In addition to routine demographics such as age and gender, and the location and cause of stroke, we determined the Glasgow Coma Scale (GCS) score from chart review at the time intubation was carried out, the length of time after stroke symptom onset when intubation was carried out, the interval between stroke onset and the beginning of clinical deterioration that led to intubation, the interval between onset of the clinical deterioration and the time of actual intubation, the duration of hospital stay following intubation for the patients that died, and the length of survival and location of residence for all patients leaving the hospital alive. Among these survivors, in addition to their degree of independence or disability based on Rankin and Barthel scores, we recorded the average monthly out-of-pocket (ie, not covered by third party payor) cost of care. We determined the Relatives Stress Scale score for the caregivers of survivors, and we determined the number of caregivers of survivors and nonsurvivors that would or would not intubate a critically ill family member or themselves again, knowing what they know now.

Informed consent was obtained from all subjects. All data collected have been maintained in confidentiality in the offices of the Stroke Service of the Department of Neurology.

Because of the small sample size, our analysis is descriptive rather than statistical. Data were analyzed when appropriate by t tests comparing clinical variables in fair or good outcome (Rankin 0 to 4 inclusive and residence at home or rehab) versus poor outcome (all other) patients.

Results. Out of the 250 infarct cases that were reviewed, we identified 20 patients who were electively intubated for deterioration after stroke (table 1).

Key data of the 20 patients who were subjected to final analysis are shown in Table 1and Table 2. The mean age was 59 years (range, 31 to 77 years) and 8 (40%) of the patients were female. Six patients survived to be discharged. The remaining (mean age, 60 years) died in the hospital after an average post-intubation hospital stay of 6.4 days (range, 1 to 32 days). Of the survivors, three patients (nos. 1, 2, and 4) returned home and another (no. 3) is still in rehab. All four reached some level of independence (Rankin 2 to 4) and in our statistical analyses were considered to have a good outcome. The survivors who were dependent and died and all those who expired before discharge were considered to have a poor outcome.

Forty percent of patients had cardioembolic stroke and 30% had large vessel disease as the underlying stroke mechanism. The etiology remained undetermined in 30%. CT revealed left hemispheric insults in 13 patients (65%) and right-sided infarcts in seven. All 20 patients had massive cortical and subcortical infarcts (MCA territory in 19 and anterior cerebral artery (ACA) plus MCA territory in one). Evidence of edema, mass effect, and midline shift were eventually noted on CT in all patients. However, 6 of 20 patients deteriorated within the first 6 hours, and at the time they required intubation their CTs showed extensive MCA hypodensity but no significant midline shift. Although the mechanisms responsible for this rapid early deterioration remain unclear, it was thought that these patients were deteriorating and needed intubation primarily because of their stroke and not because of associated conditions, using criteria already described in the Methods section.

No association could be established between outcome and age, gender, infarct cause, infarct location, or comorbidities. Seven patients had evidence of associated cardiac disease (cardiomyopathy, coronary artery disease, mitral valve prolapse, atrial fibrillation) or pulmonary disease (chronic obstructive pulmonary disease, pleural effusion). Two of the six survivors had significant cardiac disease, and two of these patients had additional chronic obstructive pulmonary disease as well.

The interval between stroke onset and neurologic deterioration ranged from 1.5 hours to 7 days (mean, 34.7 hours), but, as mentioned above, six of the patients deteriorated particularly early (within 6 hours of stroke onset). Once deterioration began, 10 patients showed a rapid worsening of their neurologic status (intubation required within 1 hour of the time deterioration was first noted) (table 1). Another three patients required intubation during the next hour. The GCS score at intubation was 7 to 12 (mean, 9.25) in the four good outcome patients, which was higher than for patients with poor outcome (3 to 12; mean, 5.9; p < 0.05).

The average duration of intubation was 5.8 days (range, 1 to 24 days) in poor outcome patients. It was 3 days or less in the three nonoperated infarct survivors with good outcome (table 1).

Among infarct survivors, the patient with the worst outcome (patient 6) had a tracheostomy and was transferred to a chronic care hospital Table 1and Table 2. He subsequently died 6.5 months after discharge. Another died after 5 months in a nursing home (patient 5). Patient 1 was mostly independent, living at home, and able to leave for medical and social visits for 4 years before dying of heart disease. Patient 2 is mostly dependent but living at home and cared for by her family. Patient 3, who had a prolonged hospital course after a hemicraniectomy, is still improving in a rehab hospital. Patient 4 made an excellent recovery and is fully independent, able to drive, and still improving.

At present, the stroke survivors have lived an average of 16 months (range, 5 to 48 months). The primary caregiver was either the spouse (two patients) or child (one patient). The estimated out-of-pocket (after third party payment) monthly cost of medical care ranged from $0 to $2,000 among the survivors.

The stress score for the caregiver of survivors ranged from 1 to 20 on a scale of 0 to 30. Two of the four caregivers questioned in the survivor group turned out to be moderately stressed (scores of 14 and 20, respectively).

Eighty-three percent of the caregivers questioned in the survivor group and 75% of the caregivers in the nonsurvivor group said that despite being aware of the final outcome, quality of life, and expenditure involved, they would still repeat the decision to intubate the patient. Fifty-three percent of the caregivers said that they would themselves want to be intubated in a similar situation.

Discussion. We recognize that the number of patients electively intubated who form the basis for this study is small and the results and conclusions are preliminary, requiring confirmation by prospective studies involving larger numbers of patients. Ideally, an analysis should include a multivariable regression to determine those features that are predictive of good outcome. We did not have a large enough sample for such an analysis, but we believe our data identify certain areas that should be explored in subsequent studies of cerebral resuscitation, such as the time of onset and rate of neurologic deterioration prior to the requirement for intubation, the role of surgical intervention (hemicraniectomy) on outcome, and the duration of intubation before a good or poor outcome becomes evident. Some variables commonly used at present to assist in decision-making such as age and infarct side were not clearly predictive.

This study was not designed to evaluate the efficacy of intubation or intensive management for acute ischemic stroke. Not only was ICU management (including ICP monitoring, medical treatment for increased ICP, and surgical decompression) not standardized, but there was also no control group.

The main finding of our study is that nihilism is not always completely justified in the 8% of patients with acute carotid distribution infarction who require intubation because of neurologic deterioration after admission. Some stroke patients (four of 20 in this series) will survive and recover to a state of semi- independence. Most of the time, even in patients who die or are left vegetative, families will not regret the decision made to intubate the patient and would make the same decision again for themselves as well. These results are similar to those reported in patients and families choosing home ventilation for amyotrophic lateral sclerosis ^[2].

We also found an interesting dichotomy in our patients who deteriorated after stroke and needed intubation: in most patients this deterioration occurred as expected about 2 to 3 days after stroke onset, but six patients deteriorated within the first 6 hours (the "malignant MCA syndrome" described by Schwab et al ^[7]). Furthermore, in patients requiring intubation, we found that less severe depression of consciousness (higher GCS score) indicated a greater likelihood of a good outcome, probably because such patients have less severely increased ICP.

Most patients (13 of 20 in the present study) will have a rapid rate of deterioration even if the onset of deterioration occurs 2 to 3 days after stroke onset. It is common to admit patients to critical care units who are at risk for deterioration from increasing ICP and watch them carefully for deterioration before instituting aggressive therapeutic measures. Although surgical intervention (hemicraniectomy) might be useful, if one were to delay surgical relief of increased intracranial pressure until the patient began to deteriorate, in most cases it would be impossible to accomplish the surgery before herniation and probable irreversible brainstem damage had occurred. Therefore, it makes more sense to identify those patients who are likely to deteriorate and institute such aggressive therapy prophylactically. Such identification is beyond the scope of this study, but ICA occlusion or dissection, and hypodensity and edema of the caudate, may be predictive of subsequent malignant edema ^[7].

There have been recent reports of hemicraniectomy in patients deteriorating from increased intracranial pressure after massive right cerebral infarction ^[8]. Although surgical intervention is well accepted in the case of cerebellar infarct, ^[9] it is infrequently performed but may still be effective for some patients with hemispheric infarction, ^[8] as suggested by one of our cases. Decompressive hemicraniectomy in selected patients should be subjected to randomized prospective evaluation.

Besides the neurologic status of the patients, we considered the financial consequences of the decision to intubate. We reported the out-of-pocket expenses for each family that were not covered by third party payors rather than recording the total cost of care. Admittedly, the latter figure would include not only the cost to the patient but also the cost to the health care system. However, we did not have the resources in this small study to obtain accurate information on total cost of hospitalization and post-hospital care. In most cases, the families were more easily able to estimate their out-of-pocket expenses. For patients who died in hospital, we believed the best indication of cost was the number of extra hospital days after intubation that were required before death. Nonsurvivors lived on average 6.4 days (range, 1 to 32 days) in the hospital after intubation prior to their death, but patients with good outcome treated nonsurgically required only 3 days of intubation.

In conclusion, our data suggest that approximately 8% of patients deteriorate to the point of needing intubation after acute ischemic stroke. Families can be told that complete recovery will not occur, but that there is a roughly 20% (four of 20 in our series) chance that there will be improvement to a state of normal or near normal cognition and semi-independence, and that a trial of intubation and aggressive medical management is reasonable if the patient or family would accept that outcome. In most patients who will achieve this outcome, improvement should be evident within 72 hours; if improvement does not occur in 72 hours, further treatment is probably futile.

Acknowledgment

This work was carried out during a World Health Organization (WHO) Fellowship (Geeta Khwaja, MD). The statements made and opinions expressed are not necessarily those of the WHO.