Severe acute stroke, i.e. acute ischemic stroke (AIS) and intracerebral hemorrhage (ICH), has been recognized to deserve management in an intensive care unit (ICU), or better the dedicated neurocritical care unit (NCCU), and considerable progress in such management has been made over the last 15 years. The results of older studies had put the usefulness of giving patients with severe stroke access to NCCU treatment and mechanical ventilation into question based on a very poor reported prognosis with mortality rates between 50% and 80% [1, 2]. Today, however, treatment options such as endovascular thrombectomy, decompressive surgery, intraventricular thrombolysis, minimally invasive hematoma evacuation, or targeted temperature management, to name but few, have changed the perspective of these patients, and they deserve rapid, adequately aggressive, and consequent emergency and intensive care. This chapter addresses the intensive care management of severe ischemic and hemorrhagic stroke with an account of the current data and some less evidence-based suggestions from the author’s own experience. The chapter will give an overview and highlight the critical care aspects of the different stroke types; for all other basic aspects and the workup in the emergency room or prehospital phase, although crucial, reference to other chapters is recommended. Although they are severe cerebrovascular disorders as well, the conditions subarachnoid hemorrhage and cerebral venous and sinus thrombosis will not be covered here.
General Aspects and Principles of Intensive Care for Stroke
For space limitations, the following part will only briefly touch on these less evidence-based ICU aspects of stroke care and mainly give recommendations, for a good part based on the author’s own experience and opinion. More details as well as discussion of studies and data, largely omitted here, can be found in guidelines such as those in Jauch et al. , Powers et al. , Torbey et al. , Hemphill et al. , and Steiner et al. .
What is a “Severe” Stroke?
Most forms of stroke can and should be managed on stroke units. However, if in AIS the main stem of a brain-supplying vessel such as the distal internal carotid artery (ICA), the proximal middle cerebral artery (MCA), or the basilar artery (BA) is occluded, the affected brain territory is large and the resulting deficit substantial. Also, verterobasilar stroke can lead to impairment of vital functions such as circulation, breathing, or airway protection. Similar consequences can be encountered in large or brainstem ICH. Also, AIS and ICH can both have cerebral (e.g. edema) or systemic sequelae (e.g. arrhythmias) that cannot be sufficiently handled on stroke units. There is no official definition of “severe” stroke, but the following criteria may be characteristic (one or more): substantially disabling deficit (National Institutes of Health Stroke Scale (NIHSS) score >15), immediate dependency (modified Rankin Scale [mRS] score 4–5, see below), potential compromise of vital functions, association with extracerebral complications, features of space-occupying effect or brainstem affection on computed tomography (CT).
When Should a Patient with Stroke be Transferred to the NCCU?
Patients with stroke that display the features above may at first be managed in the emergency room or on the stroke unit or be admitted to the NCCU for prophylactic close monitoring. While it is customary in some centers to admit all AIS patients to the NCCU who received thrombolysis or thrombectomy, many centers manage these patients very successfully on the stroke unit. The key point is that stroke patients must never be without a monitor in the acute phase. If stroke patients develop the features outlined in Box 21.1, however, they should certainly be transferred to the NCCU.
Treatment options such as endovascular thrombectomy, decompressive surgery, intraventricular thrombolysis, minimally invasive hematoma evacuation, or targeted temperature management have changed the perspective of patients with severe acute stroke and intracerebral hemorrhage and they deserve rapid, adequately aggressive, and consequent emergency and intensive care. If stroke patients develop the features listed in Box 21.1, they should be transferred to the neurocritical care unit.
Large infarction/hematoma with mass effect on cerebral imaging
Post-endovascular treatment if this involved intubation and mechanical ventilation
Need for neurosurgical operations or invasive interventions
Instability during thrombolysis or thrombolysis-related bleeding
Compromise in airway protection with risk of aspiration
Respiratory failure and need of intubation
Substantial hemodynamic instability
Progressive decline in level of consciousness
Overall, it is important to aim for cerebral tissue oxygenation and not for arbitrary levels of oxygen in the blood. Toxic levels of oxygen in stroke patients who are not hypoxemic may cause tissue damage resulting from free oxygen radical formation, lipid peroxidation, and a not completely understood process called hyperoxia-related cerebral vasoconstriction that may theoretically even lead to secondary ischemia [8, 9].
Impaired level of consciousness, decreased respiratory drive, loss of protective reflexes, and dysphagia may all lead to life-threatening respiratory situations in patients with severe stroke. Advanced modern therapeutic ICU options for severe stroke call for early life-saving intubation and initiation of mechanical ventilation. Certainly, this should take into account the patient’s or his/her family’s wishes and the overall clinical situation, but initating invasive ventilation in an acute and possibly unclear emergency situation when such a statement cannot be obtained does not constitute an ethical obstacle to later withdraw treatment efforts, if adequate.
In patients with vertebrobasilar strokes intubation is frequently necessary if the brainstem is involved and decline of level of consciousness, dysphagia, or loss of protective reflexes ensue. The decision for intubation should be based on (1) a Glasgow Coma Scale (GCS) score ≤8 or progressive decline in level of consciousness, (2) clinical or monitoring signs of respiratory failure, (3) loss of protective reflexes, (4) signs of increased intracranial pressure (ICP), (5) infarct size >2/3 of MCA territory on imaging, (6) coexistence of pulmonary edema or pneumonia, or (7) imminent surgical or invasive procedure.
Principally, early extubation should be aimed for after acute diagnostics and therapies have been performed. Extubation can be difficult as impaired level of consciousness and a high prevalence of dysphagia may lead to extubation failure and re-intubation which is associated with increased morbidity and mortality in ICU patients. NCCU patients often experience post-extubation dysphagia, and the re-intubation rate can be as high as 40%. Classical predictors of successful extubation from general critical care are unreliable in the brain-injured ICU patient and this certainly applies to most stroke patients in the NCCU as well. For instance, cooperation is often not found simply due to aphasia or apraxia, and dysphagia is much more frequent. As such, classical extubation triggers can only be used for cautious orientation in stroke patients. A retrospective study in 47 intubated MCA stroke patients suggested that a composite GCS score ≥8 trends towards extubation success (with a mean eye score of 4 in those who could be extubated versus 2.5 in those who could not) . Another recent prospective study, PRINCIPLE, investigated predictors of re-intubation in 93 critically ill AIS patients planned for extubation. Thirty-six percent of these needed re-intubation, and predictors were reduced level of consciousness for initial intubation, a higher Airway Care Score, episodes of raised ICP, length of NCCU stay, and the need for antibiotic treatment . Obviously, extubation must not be attempted if sufficient respiratory and airway protection criteria are not present, but if these are, it often can be achieved even though cooperation is not established. On the other hand, failure to detect dysphagia may result in unnecessary re-intubation. It appears reasonable to attempt extubation after successful spontaneous breathing trials, in the absence of relevant oropharyngeal saliva collections and absence of relevant demand of suctioning, together with the presence of a cough reflex and tube intolerance, if the patient is free of analgesia and sedation, even if communication and cooperation cannot be established. Stand-by re-intubation measures should be taken, and the patient should remain under monitored observation for at least 24 hours post-extubation .
Tracheostomy is frequently necessary in the ICU patient if timely extubation is not feasible. While the procedure has to be applied to c. 10–15% in the general ICU, the rate is 20–30% in the NCCU. Tracheostomy, and particularly early tracheostomy, has hardly been studied in stroke patients specifically. A few retrospective studies in ICU stroke patients have suggested advantages for early tracheostomy . A more recent randomized trial on early tracheostomy in mixed cerebrovascular ICU patients demonstrated safety, feasibility, and reduction of sedation need . The question of an outcome benefit is currently under investigation in the multicenter randomized trial SETPOINT2 . Two studies have recently identified predictors of tracheostomy need in ICH, such as ganglionic location, hematoma size, and intraventricular invasion [16, 17]. A score to predict longer duration of mechanical ventilation and tracheostomy need in stroke patients, the SETscore, may be helpful in guiding the decision  (Table 21.1).
|Area of assessment||Situation||Points|
GCS on admission <10
Ischemic infarct >2/3 MCA territory
ICH volume >25 ml
|General organ function/procedure||(Neuro)surgical intervention|
Additional respiratory disease
APS (of APACHEII) >20
Estimation of at least 2 weeks of ventilatory support and tracheostomy need, if score sums up to >10.
APS – Acute Physiology Score; LIS – Lung Injury Score.
In essence, it appears reasonable to consider tracheostomy, which is a safe procedure in the ICU setting and experienced hands, in stroke patients failing extubation or if extubation is not feasible by 7–14 days from intubation.
Impaired level of consciousness, decreased respiratory drive, loss of protective reflexes, and dysphagia may all lead to life-threatening respiratory situations in patients with severe stroke that call for early life-saving intubation and initiation of mechanical ventilation.
Tracheostomy is frequently necessary in the ICU patient if timely extubation is not feasible. While the procedure has to be applied to c. 10–15% in the general ICU, the rate is 20–30% in the NCCU (Table 21.1).
Modern lung-protective ICU ventilation following principles such as application of low tidal volumes at higher respiratory frequencies or application of positive-end-expiratory pressure (PEEP) aiming for the “open lung” have not been studied sufficiently in NCCU-dependent stroke patients. It still appears reasonable to apply them, until data support other modes of ventilation. Goals of ventilation should be optimal arterial oxygenation (arterial partial pressure of oxygen [PaO2] >70 mmHg) and normalization of the arterial partial pressure of carbon dioxide (PaCO2) between 35 and 40 mmHg, as hypocarbia may lead to cerebral vasoconstriction and risk of secondary ischemia, while hypercarbia may cause cerebral vasodilation and a subsequent increase of ICP. It appears desirable to aim for early adoption of patient-controlled modes of ventilation to achieve training of respiratory muscles, and to apply standardized respirator weaning.
Only small observational studies have addressed ventilation in severe stroke and largely suggested that escalative ventilation strategies, including application of higher PEEP in oxygenation failure, are hardly detrimental. More prospective research on stroke-adapted ventilation is urgently warranted.
Goals of ventilation should be optimal arterial oxygenation (arterial partial pressure of oxygen [PaO2] >70 mmHg) and normalization of the arterial partial pressure of carbon dioxide (PaCO2) between 35 and 40 mmHg.
Patients with ICU-dependent stroke often need analgesia and sedation to achieve freedom from pain, anxiety, and agitation. Addtionally, sedation and analgesia may facilitate medical goals such as lowering ICP, enabling procedures and operations, or terminating seizures. The way to use analgesia and sedation optimally has hardly been studied in NCCU patients, including those with AIS. Although common analgesics and sedatives such as different opioids, midazolam, propofol, ketamine, etc. have been the subject of small studies in brain-injured patients (mainly with traumatic brain injury [TBI] or subarachnoid hemorrhage [SAH]), this was largely limited to physiologic instead of outcome effects. There currently exist no data to allow for preference of any analgesic or sedative agent over the other in neurocritical care . Likewise, the application of sedation and pain scores, sedation and analgesia protocols, or sedation monitoring devices has only been addressed in small studies on brain-injured ICU patients, but not in stroke patients, specifically. Daily wake-up trials in NCCU populations (mainly with TBI and SAH) were associated with potentially negative effects, such as transient rises in ICP and stress hormone levels and cerebral desoxygenations, and this may apply to stroke patients as well.
Aiming for the lowest level of analgesia and sedation that still provides systemic and cerebral hemodynamic stability, as well as patient comfort, seems desirable. Overall, the goal should be to wean the patient from sedation as soon as adequate. Neuromonitoring of at least ICP and cerebral perfusion pressure (CPP) is recommended to guide sedation. Daily wake-up trials should be abandoned or postponed at signs of physiological compromise or clinical discomfort. Intensivists should use the agents customary at their units, choose agents according to the patient’s comorbidities and agent-specific side-effects, and pay particular attention to avoidance of extreme blood pressure variations.
There are currently not enough data to allow for preference of any analgesic or sedative agent over the other in neurocritical care. Intensivists should use the agents customary at their units, choose agents according to patient’s comorbidities and agent-specific side-effects, and pay particular attention to avoidance of extreme blood pressure variations.
Dysphagia as a result of stroke lesions compromising the swallowing act at different stages affects 30–50% of stroke patients in the acute phase. Strongest impairment is caused by brainstem stroke. Screening for dysphagia has been reported to decrease peumonia in the general stroke population. Dysphagia screening tests such as the gugging-swallowing-test (GUSS) have been studied and found useful in stroke patients, but these studies contained hardly any patients with more severe stroke. The swallowing-provocation-test (SPT), testing the involuntary part of swallowing by means of a thin oropharyngeal catheter, might overcome problems with vigilance or cooperation. After initial screening, dysphagia can be confirmed and differentiated by endoscopic swallowing tests that do not necessarily demand the patient’s cooperation. Especially the fiberoptic endoscopic evaluation of swallowing (FEES) can be done in severely affected and uncooperative patients, and it has been found reliable and predictive in studies on acute stroke patients.
Studies on best timing of nasogastric tube (NGT) placement in critically ill stroke patients are lacking. Similarly, predictors of the need for percutaneous endoscopic gastrostomy (PEG) tube placement have only been studied in mixed ischemic stroke populations. In a recent Cochrane analysis on the general acute and subacute stroke population, NGT and PEG did not differ in terms of case fatality, but PEG was more secure, resulted in fewer treatment failures, and reduced gastrointestinal bleeding, as well as a higher feed delivery . In terms of gastric ulcer prophylaxis, recent data support ranitidine use in reduction of nosocomial pneumonia versus proton pump inhibitors.
With this lack of studies focusing on food intake, delivery, and gastrointestinal function in NCCU patients with stroke, it may be best to carefully transfer insights from the general ICU. It is often possible to assess swallowing by standard screening tests in the very early phase (days 1 and 2, patients may still be in the emergency room or the stroke unit) to guide NGT placement. Once the level of consciousness starts to decline, however, the patient should be kept nil by mouth and the NGT placed at a very low threshold to avoid aspiration. The NGT can then be used for feeding over the next few weeks of NCCU treatment. During this time, gastrointestinal transport stimulation and gastric ulcer prophylaxis should follow principles from the general ICU. In the later phase of the disease, i.e. after termination of sedation and weaning from the respirator, swallowing capacity can be reassessed, preferrably by use of an endoscopic method such as the FEES. The results thereof should be incorporated in the decision to place a PEG.
A nasogastric tube can be used for feeding over the first few weeks. After termination of sedation and weaning from the respirator, swallowing capacity can be assessed, preferrably by use of an endoscopic method such as the fiberoptic endoscopic evaluation of swallowing (FEES). The results thereof should be incorporated in the decision to place a PEG.
Fluid Status and Nutrition
Stroke patients in the NCCU should be kept euvolemic, since hypovolemia may compromise cerebral perfusion by hypotension and high blood viscosity, while hypervolemia may promote formation of brain edema. The average demand for an adult patient is 2–3 l/day (1 ml/kg per hour), but individual factors such as higher perspiration under invasive ventilation and hyperthermia as well as cardiac and renal function have to be taken into account. After decades of ICU controversy regarding colloidals versus cristalloidals in volume management, more recent studies have cast doubt on the benefits of colloids in many situations and populations. Hypotonic fluids may promote edema and should be avoided. In essence, cristalloids (normal saline) appear adequate for fluid management of the AIS patient in most situations.
In terms of nutrition, sedated and ventilated stroke patients show a good correlation with the total energy expenditure (TEE) and the basal energy expenditure (BEE) as estimated by the Harris-Benedict equation. The guiding principle in general ICU nutrition is towards a preference for enteral over parenteral nutrition. Nutrition should be tailored with the goals to meet caloric demand, while avoiding over-feeding and hyperglycemia in AIS patients, using the Harris-Benedict equation to predict basal energy demand. It is probably reasonable to follow general ICU principles in stroke patients, i.e. to (1) calculate energy (caloric) demand and standardize caloric requirements based on height, weight, age, and sex, (2) balance nutrition to represent the components carbohydrates, proteins, and fat (taking into account “nutritional” components of some infused drugs) and be supplemented, (3) start nutrition on day 2 from admission, and (4) prefer enteral over parenteral nutrition whenever possible.
Stroke patients in the NCCU should be kept euvolemic and nourished corresponding to their energy demand.
Both hyperglycemia and hypoglycemia are associated with increased morbidity and mortality in severe stroke. So far, it remains unclear whether systemic glucose itself is the decisive pathophysiologic factor of these observations or just an indicator of other compromising mechanisms. Likewise, the impact of controlling glucose and the best method to do so have remained controversial in the brain-injured ICU population. A recent systematic review and meta-analysis on 16 RCTs in more than 1 200 mixed NCCU patients, including severe stroke, found no mortality benefits of intensive insulin therapy, but did find the association with a higher risk of hypoglycemia, particularly jeopardizing in the ischemic brain. However, a very loose glycemic control was also associated with worse neurologic recovery. The authors concluded that intermediate glucose control (insulin therapy aiming for 140–180 mg/dl) may be most appropriate for this patient population  and this is also recommended in current stroke guidelines .
In stroke intermediate glucose control (insulin therapy aiming for 140–180 mg/dl) is recommended.
Anemia is associated with worse outcome in ischemic stroke. It is very common in the majority of ICU patients from the third day of admission. A more recent retrospective study on critically ill stroke patients revealed that almost all of these acquire marked anemia during their NCCU course, which is then associated with longer ICU stay and duration of ventilation . However, the optimal hemoglobin level in critically ill stroke patients is unclear, as is the optimal red blood cell (RBC) transfusion policy.
Theoretically, optimizing the oxygen-carrying capacity should play a decisive role in the compromised brain. The physiological benefits of RBC transfusion suggested in neuromonitoring studies in TBI and SAH patients should be assumed for stroke patients as well. It is thus questionable if 7 g/dl hemoglobin (a widely accepted level in general ICU patients) is optimal for these patients. However, transfusion was more often found associated with worse outcome in neurocritical care patients, either by being an indicator of higher disease severity or indeed by causing harm . In anemic NCCU patients with stroke, RBC transfusion was not associated with improvement . Until more research clarifies optimal hemoglobin levels and transfusion strategies, it may be reasonable to use RBC transfusion below a hemoglobin level of 7 g/dl or if systemic or cerebral monitoring – if available – suggests a compromised arteriovenous oxygen extraction. RBC transfusion should also be triggered by specific situations such as planned surgery, hemodynamic status, and bleeding. Anemia or the progression thereof should be avoided by reducing blood sampling to the necessary minimum, treating infections and renal disease, preventing volume overload, promoting appropriate diuresis, etc.
Parameters indicating a compromised arteriovenous oxygen extraction or hemoglobin levels below 7 g/dl appear reasonable RBC transfusion triggers.
Deep Vein Thrombosis Prophylaxis
The incidence of DVT in stroke patients is about 3%, despite an average DVT prophylaxis rate of more than 95%. The CLOTS trial on the prevention of DVT and pulmonary embolism (PE) in 5 632 acute stroke patients screening systematically for DVT and PE revealed an incidence of 11.4% of DVT in the early phase (7–10 days) and an additional 3.1% in the late phase (25–30 days) of the hospital stay . Of the stroke patients with DVT, 35% were symptomatic and 5% developed PE. The authors concluded that DVT prophylaxis has to be started early and continued for at least 4 weeks. This study confirmed and extended previous studies on this subject, but contained only a difficult-to-determine proportion of patients with severe stroke. Follow-up trials showed that thigh-length graduated compression stockings are not beneficial with regard to prevention of DVT or PE when compared to patients in whom stockings were avoided. Furthermore, more skin ulcers, necrosis, and even leg ischemia were found in the stocking group. Intermittent pneumatic compression has recently been investigated in CLOTS3, enrolling almost 3 000 immobile stroke patients, and reduced the risk of DVT effectively at the price of significantly more skin breaks, but showed a trend to better survival . Heparin is another effective preventive strategy for DVT in patients with acute stroke and the concomittant bleeding risk is outweighed by the benefits of thromboembolic prevention . Low molecular weight heparin (LMWH) appears to be advantageous over unfractionated heparin. A small study in patients with ICH found that heparin could be started safely within 2 days from onset without augmenting cerebral bleeding .
If stroke patients in general are at considerable risk of DVT and PE, the extremely immobile, ventilated, and unconscious stroke patient in the ICU should be at a much higher risk, even if this was hardly systematically investigated, so far. A mode of DVT prophylaxis, installed early and continued at least as long as immobilization is present, appears warranted. As stockings are of no proven benefit in stroke patients and as patients in the ICU are more prone to dermatological problems and often need access to their legs for nursing care and positioning or application of devices, stockings should be avoided and the use of intermittent pneumatic compression should be considered with caution. Although these patients do certainly have a higher bleeding risk than patients with smaller AIS or ICH, this does not outweigh the benefits from DVT and PE prophylaxis. Therefore, these patients should receive DVT prophylaxis from admission to the ICU and for the time of immobilization. LMWH should be used for DVT prophylaxis. Early mobilization should be strived for in NCCU-dependent AIS patients with a stable systemic and cerebral physiology.
Deep venous thrombosis prophylaxis with LMWH should be started early and continued for at least 4 weeks.
Patients with increased thromboembolic risk, e.g. due to atrial fibrillation (AF) or a prosthetic heart valve, need re-initiation of their oral anti-coagulation, which is usually paused in the event of a large ischemic stroke. However, the best point in time for this re-institution is highly controversial, as prevention of embolism goes together with the risk of ICH. Recommendations on the dilemma in anti-coagulated patients suffering from AIS or ICH range from 2 days to 4 weeks. The largest, yet retrospective, study in almost 1 200 patients with ICH under oral anti-coagulation examined re-initiation of oral anti-coagulation within such a time frame and found a clear benefit of prevention of thrombembolic events over bleeding risk . Whether these findings can be transferred to all ICU ICH or AIS patients is questionable and prospective research in that setting is necessary.
There is presently not much conclusive evidence to solve this very challenging problem. Recommedations regarding the best timing differ considerably, as do the customs in different ICUs. The situation is confused by a number of options which have all been used: re-initiation after 7, 14, 21, or more days, bridging with heparin or not, very different aPPT/aPTT targets when using heparin and re-anti-coagulation with warfarin or new oral anti-coagulants have all been chosen and make the situation confusing. The following considerations may serve as a strategy: (1) the thromboembolic risk is low in the first 2 weeks, even in the AF patients with a prior stroke or patients with prosthetic valves; (2) the risk of bleeding is quite high in large stroke; (3) oral intake will be compromised for weeks; (4) invasive procedures such as hemicraniectomy, tracheostomy, or PEG placement do not comply well with anti-coagulation. Hence, it appears reasonable that oral anti-coagulation should be re-initiated after 4–6 weeks in patients with severe stroke, provided all invasive procedures are completed. In selected patients with an extraordinarily high thromboembolic risk (e.g. prosthetic valve with transesophageal echocardiography [TEE] evidence of intracardiac thrombus), an earlier individual therapy (e.g. with a modest aPPT/aPTT heparin strategy) may be chosen.
The best point in time for re-institution of oral anti-coagulation in patients with increased thromboembolic risk is highly controversial, as prevention of embolism goes together with the risk of intracerebral hemorrhage. In ICU care, it appears reasonable that oral anti-coagulation in patients with severe stroke should be re-initiated when all invasive procedures are completed often 4–6 weeks after the acute stroke.
The majority of patients with severe stroke, known to be associated with immunosuppression, will present with risk factors for pneumonia on admission or shortly therafter. Given this high risk and association with morbidity, early initiation of preventive strategies, such as early intubation and/or NG tube placement, bundles of ventilation-associated pneumonia prevention, and standard hygienic measures seem prudent – even if high-quality evidence for this is lacking. Any infection must be treated early, aggressively, and according to modern standards of ICU infection control. Data are too weak, so far, to justify prophylactic antibiotics in stroke.
BP mangement in the NCCU should follow similar principles applied in the ER for early stroke treatment [3, 6]. Patients often have an impaired cerebral autoregulation, thus systemic alterations in BP might passively be transduced to cerebral perfusion, which has to be kept in mind. Systemic hypotension may lead to secondary ischemia, and hypertension may lead to to secondary hemorrhagic transformation or edema promotion. Only neuromonitoring, e.g. ICP and CPP or measures of autoregulation, may help to guide systemic blood pressure treatment. However, actualization of this process remains theoretical and largely unresolved. Often in the sedated stroke patient, preventing hypotension by vasopressors and volume will dominate the early phase, while anti-hypertensive measures might have to be applied later when the patient is allowed to wake up and weaned from the ventilator. Studies in very different settings and mainly in the early phases of disease have suggested that a systolic blood pressure (SBP) >140 mmHg in AIS and <140 mmHg in ICH may be a reasonable target, but this is just a very raw estimate that may well vary individually.
About one-third of patients presenting with stroke are hyperthermic. Fever was associated with worse outcome in numerous studies, and putative effects involve increase of metabolic demand, release of excitatory neurotransmitters, formation of free radicals, promotion of apoptosis, etc. Fever can be part of a non-specific stress response, be caused by infection (c. 50%, e.g. peumonia, urinary tract infection, infective endocarditis), or be caused by central hypothalamic dysregulation. Obviously, infections should be searched for and treated vigorously with anti-infectives. Pharmacological measures to lower temperature and achieve normothermia have neither been very efficient nor produced convincing outcome benefits. The largest randomized trial on high-dose acetaminophen for normothermia in AIS had to be abandoned due to lack of funding after 1 400 patients had been recruited. It found no significant differences in pre-specified outcomes between the groups that had a mean body temperature difference of 0.26°C; only in a post-hoc analysis, there was a beneficial effect for patients with a baseline temperature between 37° and 39°C . Possibly, application of a consequent stepwise normothermia protocol involving not only acetaminophen or metamizol, but also convection mehods, ice-packs, ice-cold saline infusions, or at best feedback surface and endovascular cooling devices, may achieve benefits, but that remains to be proven. Since measures to achieve normothermia appear fairly safe, their application to keep stroke patients at a target temperature of 36.5–37.5°C in the acute phase of their NCCU stay may be reasonable.
Hypothermia as a long-standing neuroprotective principle has been backed up by countless experimental studies and is thought to work by reduction of metabolic and oxygen demand, decrease in excitotoxic neurotransmitter release, inhibition of free radical formation, stabilization of the blood–brain barrier, and anti-edematous, anti-inflammatory, and many other mechanisms. It can be subdivided into mild (up to 33°C), moderate (29–33°C), and deep (below 29°C) hypothermia. It can be maintained by different surface and endovascular, systemic, and regional cooling devices and induced by very diverse methods such as cooling packs, ice-cold saline infusions, nasal and sinus cooling, etc. Small clinical studies on hypothermia in severe AIS and ICH have had mixed – at times encouraging – results, but a recent Cochrane analysis comes to the conclusion that there is at present too little evidence to routinely advocate therapeutic hypothermia in stroke . Hypotherermia may be part of an escalative ICP-lowering strategy (see below). Potential side-effects of hypothermia are cardiac arrhythmias, electrolyte derangements, impaired coagulation, immunosuppression and infection, and rebound edema/raised ICP on re-warming. All of these appear to be less frequent under mild compared to moderate hypothermia. Shivering, the physiological response to cooling, is associated with high systemic and cerebral energy demand and has to be fought by wrapping hands and feet of the patient (“counterwarming”), and drugs such as meperidine, buspirone, or clonidine. Re-warming should not be allowed faster than 0.1°C/hour, as this can be associated with rebound raise in ICP. Current randomized trials are investigating hypothermia in AIS and ICH further, and will hopefully produce evidence on its impact, indication, patient selection, target temperature and duration, and optimal technical application.
Seizures can occur as part of stroke onset (early) or in the course of severe stroke (late), triggered by hemorrhagic transformation, expansion, or edema formation. Overall, the incidence is probably around 10%. Seizures raise the cerebral energy demand and promote secondary brain injury. Non-convulsive status epilepticus is a frequent cause of unexplained coma in the NCCU patient, which must be actively looked for and proven by the electroencephalogram (EEG). There are no data supporting prophylactic anti-convulsants. Manifest seizures and particularly status epilepticus have to be treated vigorously according to customs from other fields of neurology and neurocritical care [3, 31].
Any infection must be treated early, aggressively, and according to modern standards of ICU infection control. BP management in the NCCU should follow similar principles applied for early stroke treatment. Pharmacological measures to lower temperature and achieve normothermia have neither been very efficient nor produced convincing outcome benefits. A recent Cochrane analysis comes to the conclusion that there is at present too little evidence to routinely advocate therapeutic hypothermia in stroke.
Brain Edema and Raised ICP
Large hemispheric infarction leads to the same pathophysiologcal changes in the infarcted brain tissue and its vincinity as in any other types of AIS. However, due to the size of infarction, the consequences of these mechanisms are much greater and often fatal. In the first minutes to hours after stroke, massive release of excitatory neurotransmitters, such as glutamate, cause excitotoxicity, while at the same time the energy breakdown of the cells, due to cessation of oxygen and glucose supply, causes failure of the active membrane pumps and channels to stabilize the ionic membrane potential. The energy disbalance is further increased by peri-infarct depolarizations ensuing within the first hours after stroke. Together, this leads to uncontrolled influx of cations into brain cells dragging water and causing, together with other factors, cytotoxic edema. Once the capacity of cell swelling is overwhelmed, water will enter the interstitial compartment. Since these mechanisms are not limited to neurons, but also involve neuroglia and endothelial cells, they promote disruption of the blood–brain barrier and eventually cause vasogenic edema, also driven by inflammatory processes extending from hours to days after AIS. In the rigid skull, the enlarging edema can hardly be compensated by reductions of the blood and cerebrospinal fluid (CSF) compartments, hence compression of as-yet unaffected brain tissue or vessels. Eventually rise in ICP, and herniation will follow. In most patients with hemispheric stroke, secondary brain edema develops in a fairly gradual and temporally predictable fashion, with a radiologically evident space-occupying effect on day 2, peaking at days 4 or 5. In some patients, however, relevant brain edema may occur with a delay or not at all, for unknown reasons. In posterior circulation stroke, relatively small edema can cause dangerous tissue compression and CSF flow blockage due to limited space in the posterior fossa of the skull. Basic measures to prevent brain edema comprise restriction of free water, avoidance of hypo-osmolar fluids, avoidance of excess glucose administration, avoidance of hypertension after reperfusion therapies, minimization of hypoxenia and hypercarbia, caution in application of drugs causing cerebral vasodilation, and treatment of hyperthermia. Recently, edema-inhibitory effects of oral anti-diabetics have been made the subject of a current trial (GAMES ), which showed safety and feasibility of glyburide in large hemispheric infarction (LHI), as well as encouraging results for secondary endpoints such as midline shift, but not (yet) advantages in functional outcome or reduction of necessity of decompression. The follow-up trial CHARM is ongoing. Another theoretical approach to edema prevention, hypothermia, awaits further prospective research.
If clinically relevant edema occurs and is detected by clinical signs (decline in level of consciousness, augmentation of neurological deficit, nausea and vomiting, anisocoria), ICP monitoring, or cerebral imaging, pharmacological treatment should be applied. Steroids are not efficient in AIS-related edema and may have deleterious side-effects for the NCCU patient .
Although secondary brain damage after ICH shares some molecular mechanisms with that after AIS, certain distinct features should be acknowledged. The primary impact of ICH is mechanical, with disruption of the cerebral cellular architecture within seconds. Very early, depending on the initial hematoma size, compression can compromise adjacent brain tissue and vessels and lead to secondary mechanical and ischemic damage in the surroundings. In about 30% of patients, secondary hematoma expansion (HE) will occur within the first 24 hours, either due to ongoing or recurring bleeding for mostly unknown reasons . HE with eventually enlarged final hematoma volume and/or invasion of the ventricles (IVH) is a consistent factor to worsen the prognosis. Perihematomal edema (PE) evolves at a slower pace, contributes to total lesion volume and mass effect, and peaks during the second week from onset in most cases. Factors that promote PE come from the hematoma itself or its degradation products, or are a direct response of the coagulation cascade. They include thrombin, hemoglobin, and iron, and are associated with inflammation and free radicals. Pre-clinical and first clinical studies have been started on these treatment targets, such as chelating iron by deferoxamine, but these studies are either ongoing or have not shown functional benefits. Reported attempts to prevent or ameliorate HE have either failed or been limited to single, small studies, with the exception of two treatment approaches. One is by hemostatic stabilization employing substances such as recombinant factor VII, tranexamic acid, or platelet infusion. The other is early and aggressive blood pressure lowering (see below).
Osmotherapy, following the principle to osmotically draw fluids from the brain tissue into the blood to be removed, has been applied for more than 60 years. Most experience and supportive data exist for mannitol and hypertonic saline. Many other substances have been used, but are probably dispensable. Some authors have criticized osmotherapy for stroke-related edema therapy, postulating that in the context of a damaged blood–brain barrier, the substances may actually rather draw fluids from healthy tissue to infarcted tissue and thus increase mass effect and midline shift. Data on this theory are inconsistent, however, and observations and data on the ICP-decreasing effect of osmotherapy seem to prevail, if osmotherapeutics are applied in pulsatile fashion. Osmotherapy alone is very often not sufficient to treat stroke-related brain edema, however.
Every NCCU patient with severe stroke who is prone to mass effect and/or edema formation, and particularly if clinical observation is limited by ventilation and sedation, should have the ICP measured by parenchymal probe or external ventricular drain (if hydrocephalus is feared or present) and the CPP calculated (CPP = mean arterial pressure [MAP] – ICP). If ICP exceeds 25 mmHg for more than 5 minutes, immediate measures to lower it (and/or raise CPP >65 mmHg as by augmentation of MAP) should be undertaken and cerebral imaging realized to detect the cause of the ICP increase. Causes will be edema with mass effect and midline shift (hemispheric ischemic stroke with edema, enlarging hematoma and/or edema) or consequent hydrocephalus (cerebellar stroke, intraventricular hemorrhage [IVH]) in most of the cases, but may also be secondary hemorrhage, seizures, reduced venous return, etc. There do not exist specific data on the optimal ICP-lowering therapy in severe stroke, so it is recommended to follow a stepwise approach common in other brain injuries (Box 21.2).
Raised ICP is caused by cytotoxic and vasogenic edema in acute ischemic stroke. In patients with intracerebral hemorrhage, secondary hematoma and perihematomal edema evolving at a slower pace contribute to ICP. Every NCCU patient with severe stroke who is prone to mass effect and/or edema formation, and particularly if clinical observation is limited by ventilation and sedation, should have the ICP measured by parenchymal probe or external ventricular drain. For ICP-lowering therapy, see Box 21.2.