Systemic Monitoring

Systemic monitoring in the NICU typically includes cardiac telemetry, frequent scheduled noninvasive BP measurements (by automatic cuff inflation) or continuous invasive arterial BP recording, pulse oximetry, and core body temperature. Continuous arterial BP monitoring is accomplished by inserting an indwelling cannula into a medium-caliber artery (e.g., radial arterial line). The invasiveness of the procedure is justified by the precise real-time information it provides. Continuous arterial BP monitoring is especially recommended in patients treated with induced hypertension (e.g., symptomatic vasospasm after SAH), cases requiring very strict BP control to avoid hemorrhagic complications (e.g., ruptured arteriovenous malformations), and patients with hypotension (e.g., sepsis), compromised cerebral perfusion pressure (CPP) (e.g., TBI with raised intracranial pressure [ICP]), or autonomic instability (e.g., Guillain-Barré syndrome). Arterial lines provide the additional advantage of eliminating the need for repeated arterial punctures to measure arterial blood gases. However, although generally safe, placement of an arterial line may be complicated by local infection, leading to bacteremia and thrombosis with risk of digital ischemia. Careful attention to proper technique and adherence to strict sterile conditions during placement and manipulation of the catheter are mandatory (Tegtmeyer et al., 2006).

The most accurate method of measuring core body temperature is a pulmonary artery catheter thermistor, but since most patients in the NICU do not require pulmonary artery catheter insertion, bladder or rectal probes are most frequently used. Bladder and rectal probes correlate well with pulmonary artery catheter thermistor readings, but there is a lag in the detection of temperature changes by the probes. The site of temperature recording becomes particularly important in patients treated with cooling measures. Thus, monitoring esophageal temperatures is recommended when using certain intravascular cooling devices (De Georgia et al., 2004).

Central venous catheters allow monitoring of central venous pressure while providing access for fluid and drug administration. They are, however, a frequent source of infection. Rigorous sterile techniques at the time of catheter insertion, cutaneous antisepsis with chlorhexidine (rather than povidone-iodine), topical application of antiinfective ointment or a chlorhexidine-impregnated dressing to the insertion site, and catheters with an antiinfective surface may reduce the risk of catheter-related bloodstream infection (Safdar et al., 2002). The role of pulmonary artery catheters in ICUs is shrinking as studies consistently demonstrate that their use is associated with higher rates of complications (Sandham et al., 2003; Wheeler et al., 2006).

Brain Monitoring

Neurological examination may lack sensitivity in critically ill patients who have depressed levels of consciousness due to the brain disease or from the effect of sedative medications. Brain monitoring methods developed and refined over the past 25 years may provide additional valuable information in these cases. These techniques offer real-time data, unlike imaging modalities that represent only “snapshots” of the patient’s condition at certain points in time. Therefore, brain monitoring techniques are better suited to dynamic assessment of the changes in neurological status of critical patients.

Multiple brain monitoring methods are now available. They are most useful when applied in combination, a practice known as multimodality monitoring (Diedler and Czosnyka, 2010). It is important to be aware, however, that the endpoints of most studies validating the use of brain monitoring methods modalities have been surrogate physiological measures rather than actual assessments of patients’ functional outcome. In fact, there is no class I evidence proving that the use of multimodality brain monitoring results in improved clinical outcomes. Currently, the clinical application of brain monitoring techniques is restricted to large centers, especially those treating numerous TBI patients.

Methods for cerebral monitoring are divided into three main categories according to their spatial resolution: global, regional, and local brain monitoring (Table 45.1). Global brain monitoring techniques measure ICP, CPP, electrical potentials, and venous oxygen saturation. Regional and local brain monitoring methods include cerebral blood flow (CBF), cerebral blood flow velocities (BFV), brain tissue metabolism, temperature, and oxygenation.

Intracranial Pressure Monitoring

The intracranial space is occupied by three constituent compartments: the brain (accounting for 80% to 90% of the intracranial volume), the blood, and the cerebrospinal fluid (CSF). Because the skull is rigid, any expansion of one of these compartments must be compensated by a reduction in size of the others (a physiological principle known as the Monro-Kellie doctrine). When this compensation is insufficient, the ICP rises. Small increases in intracranial volume can be initially accommodated with little or no effect on the ICP, but as more volume is added, intracranial compliance falls until it reaches a critical point beyond which a minimal increase in volume causes an exponential rise in ICP. This pressure-volume relationship is depicted in Fig. 45.2. In other words, the initial physiological response to an increase in brain volume is a reduction in the CSF and venous blood volumes by shifting these fluids out of the intracranial space (except in the cases of hydrocephalus and venous thrombosis). Once these compensatory mechanisms are exhausted, the system becomes noncompliant, and further increments in brain volume compromise arterial blood flow and eventually lead to herniation of brain tissue.

Fig. 45.2 Relationship between pressure and volume changes in the intracranial compartment. CSF, Cerebrospinal fluid; ICP, intracranial pressure; P, pressure; V, volume.

Normal ICP in a supine individual is less than 10 mm Hg when measured at the level of the foramen of Monro (typically referenced to the tragus). Levels exceeding 20 mm Hg define raised ICP. Knowing the actual ICP is a prerequisite to determining CPP, which is defined by the relationship between ICP and mean arterial pressure (MAP) as follows:

It has also been argued that the main purpose of ICP monitoring is maintenance of normal CPP (normal, >70 mm Hg) because the latter may be more related to secondary ischemic injury (Rosner et al., 1995). The relative importance of ICP and CPP as main targets of therapy remains a matter of debate.

ICP is pulsatile and has systolic and diastolic components. In addition to the value of mean ICP, these components must also be evaluated carefully. The normal ICP waveform consists of a 3-peaked wave (Fig. 45.3). P1, the first and generally the tallest peak, is also known as the percussion wave and corresponds to the transmitted systolic BP; P2 (the tidal wave) and P3 (the dicrotic wave) are normally smaller peaks, and the notch between them corresponds to the dicrotic notch of the arterial waveform. As ICP increases, P2 and P3 rise and eventually surpass P1. Ultimately, with continued elevation of ICP, the waveform loses distinct peaks and assumes a triangular morphology. Intracranial pathology leading to sustained elevations of ICP may produce plateau waves, also known as A-waves of Lundberg (see Fig. 45.3). These waves reflect a sudden dramatic rise in ICP to levels of 40 to 100 mm Hg, often lasting 5 to 20 minutes. Plateau waves indicate critically low intracranial compliance leading to marked changes in ICP, even with very small variations in intracranial volume. Although their pathophysiology is not fully elucidated, experimental observations suggest that plateau waves may be generated by brief episodes of systemic hypotension leading to exaggerated cerebral vasodilation in patients with abnormal vasomotor reactivity (Rosner and Becker, 1984).

Fig. 45.3 Intracranial pressure tracings in the setting of normal and reduced compliance. Plateau wave (A wave of Lundberg) is seen in the center of the figure.

ICP may be monitored using intraparenchymal, intraventricular, epidural, or subdural devices (Brain Trauma Foundation, 2007). Intraventricular monitoring remains the gold standard because of its precision. It consists of a ventricular catheter connected to an external transducer which allows continuous ICP readings. Advantages of this technique are providing reliable ICP measurements and allowing external drainage of CSF. Hence, ventricular monitoring is indicated in patients with hydrocephalus and often preferred in those with refractory intracranial hypertension. Major drawbacks are higher risk of infection (rate of ventriculitis is 3% to 8% and increases with duration of the ventriculostomy) (Flibotte et al., 2004; Holloway et al., 1996; Martinez-Manas et al., 2000), risk of bleeding at the time of catheter placement (especially in patients with underlying coagulopathy or recent use of antithrombotics), and system malfunction (dampening of the waveform may be caused by apposition of the catheter tip against the ventricular wall or obstruction of the catheter by a blood clot or an air bubble). Risks may be minimized by careful placement of the catheter and maintenance of the system under strict sterile conditions, use of antibiotic prophylaxis (e.g., cefotaxime 2 g every 6 hours from the time of catheter insertion until 24 to 48 hours after its removal) (Flibotte et al., 2004), and withdrawal of the catheter as soon as possible (Holloway et al., 1996). Exchange of the catheter every 5 days, although a common practice, does not appear to decrease the risk of infection (Holloway et al., 1996; Lozier et al., 2002); in fact, repeated catheter insertions have been found to be associated with higher risk of ventriculitis (Arabi et al., 2005).

Intraparenchymal fiberoptic monitors are also quite accurate. When compared with intraventricular catheters, the measurements provided by intraparenchymal monitors differ on average by ± 2 to 5 mm Hg. Advantages of this monitoring system include simple and safe insertion technique, easy maintenance, relative lack of substantial drift (even after several days), and low risk of infection. Disadvantages include high cost, technical complications (e.g., breakage of the optical fiber), and most importantly, inability to drain CSF. Epidural and subdural monitors are less reliable and therefore rarely used but are a valuable option for patients with severe coagulopathy (e.g., liver failure with cerebral edema), given their lower risk of hemorrhagic complications (Vaquero et al., 2005).

ICP should be monitored in patients with severe TBI and a GCS sum score below 9 and an abnormal computed tomography (CT) scan, or a normal CT scan with two or more of the following criteria: age older than 40, unilateral or bilateral motor posturing, and systolic BP less than 90 mm Hg (Brain Trauma Foundation, 2007). It is difficult to extrapolate the value of these guidelines to patients with diagnoses other than trauma, owing to lack of specific data on ICP monitoring in those other conditions. Some experts advocate monitoring ICP in comatose patients with a large intracranial mass lesion (hematoma, abscess, large infarctions, etc.) causing radiologically documented tissue shift. Patients with SAH, ICH, or cerebellar ischemic or hemorrhagic strokes producing acute hydrocephalus typically have their ICP monitored once a ventriculostomy catheter has been placed primarily for drainage purposes.

Jugular Bulb Oximetry

Jugular bulb oximetry measures the oxygen saturation of venous blood returning from the brain (normal 50% to 65%) by means of a fiberoptic catheter (Feldman and Robertson, 1997). The main goal of jugular venous oxygen saturation (Sjvo₂) monitoring is to provide a continuous measure of the changing balance between cerebral oxygen delivery and cerebral oxygen consumption. Simultaneous determination of Sjvo₂ using the jugular bulb catheter and arterial oxygen saturation (Sao₂) allows for the calculation of the intracranial arteriovenous oxygen difference (AVDo₂) (normal 24%-42%). Cerebral oxygen consumption can be calculated as the product of AVDo₂ and CBF. The cerebral oxygen extraction rate (O₂ER) is derived from the ratio of cerebral oxygen consumption to cerebral oxygen delivery.

Jugular venous desaturations denote relative reductions of global cerebral oxygenation. Sjvo₂ below 50% for 15 minutes or more are deemed indicative of ischemia. Sjvo₂ monitoring has been mostly tested in patients with severe TBI. In these patients, jugular venous desaturations have been shown to correlate with the occurrence of secondary brain insults and poor outcome (Gopinath et al., 1994; Robertson et al., 1995). High Sjvo₂ should not simply be equated with hyperemia; it may also be associated with poor outcome in comatose patients, possibly indicating lack of oxygen utilization after extensive neuronal death (Cormio et al., 1999). Favorable experience with jugular bulb oximetry has been reported in patients with SAH and ICH (Heran et al., 2004; von Helden et al., 1993), but interpreting Sjvo₂ may be difficult in patients with severe hemispheric ischemic strokes (Keller et al., 2002). This technique is also used to monitor cerebral oxygenation during neurosurgical procedures.

Therapeutic interventions in response to information provided by jugular bulb oximetry have been proposed, including adjustment of the degree of hyperventilation, timing and intensity of osmotherapy, adjustment of MAP, and treatment of anemia (Macmillan and Andrews, 2000). There is no proof, however, that these interventions improve functional outcome. As shown by the negative results observed in studies testing therapies guided by pulmonary artery catheters, the clinical value of aggressive interventions aimed at optimizing physiological parameters must be proven before we incorporate these into clinical practice.

Advantages of the jugular bulb catheter as a monitoring modality include the practicality of continuous bedside monitoring, the capability of confirming the oximeter reading by drawing blood through the catheter, and the numerous physiological parameters that can be derived from the Sjvo₂ to ascertain cerebral oxygen balance. Disadvantages of the catheter include its susceptibility to positioning artifacts and the complications associated with catheter insertion, including carotid puncture, infection, accidental misplacement, and jugular thrombosis (Coplin et al., 1997; Coplin et al., 1998; Latronico et al., 2000).

Electroencephalography

Continuous bedside electroencephalography (EEG) monitoring is based on four of its major neurobiological features (Jordan, 1995): (1) its close relationship to cerebral metabolic rate; (2) its sensitivity in detecting hypoxic-ischemic neuronal dysfunction at an early stage; (3) its obvious primacy as a monitor of seizure activity; and (4) its value in cerebral localization. Continuous EEG recording has been advocated as a valuable routine tool to monitor critically ill neurosurgical and neurological patients.

Despite the fact that the technical aspects of EEG application in the NICU do not differ greatly from the standard routine EEG, some factors are relatively unique to the ICU setting. The main differences are the abundance of electrical artifact sources (ventilators, intravenous [IV] pumps, dialysis machines, suctioning equipment) and the inability of the patient to cooperate, secondary to various degrees of encephalopathy. In addition, continuous bedside EEG monitoring requires EEG interpreters available to view the recording frequently throughout the day and specially trained nurses capable of recognizing meaningful changes in the tracing.

Status epilepticus is the most common indication for EEG monitoring, because the clinical ascertainment of ongoing seizure activity is often obscured by the effect of sedatives and analgesic agents. The EEG is essential for monitoring the effects of treatment, especially when barbiturates or general anesthetics are administered to achieve a burst-suppression pattern. Detection of nonconvulsive seizures and nonconvulsive status epilepticus (NCSE) can only be accomplished by EEG monitoring. Timely diagnosis of NCSE is important because delayed recognition may be associated with increased mortality (Young et al., 1996).

Nonconvulsive seizures were reported in up to one-third of unselected NICU patients, frequently involving the presence of NCSE (Jordan, 1995). NCSE has also been noted in 8% of comatose patients in medical ICUs (Towne et al., 2000), and nearly a third of septic patients with encephalopathy may have electrographic seizures without obvious clinical manifestations (Oddo et al., 2009). Continuous EEG monitoring has documented nonconvulsive seizures after severe TBI, ischemic stroke, poor-grade SAH, ICH, and after termination of generalized convulsive status epilepticus (DeLorenzo et al., 1998; Dennis et al., 2002; Vespa et al., 1999). These events might exacerbate excitotoxic injury in vulnerable brains and have been associated with high mortality (Young et al., 1996). But while their prognostic value is fairly well established, the impact of aggressive treatment of nonconvulsive seizures on clinical outcome remains to be determined (Hirsch, 2004).

Continuous EEG monitoring has also been used as an aid for early detection of ischemia in patients with SAH at high risk for vasospasm. Although early experience is promising (Claassen et al., 2006; Vespa et al., 1997), it is too early to recommend continuous EEG for this indication. Intracortical EEG (based on the use of deep electrodes) may be substantially superior to scalp EEG for detecting changes related to secondary neurological insults in patients with various forms of acute brain injury (Waziri et al., 2009). Furthermore, recurrent cortical spreading depolarizations may exacerbate local brain hypoxia in patients with TBI or SAH (Bosche et al., 2010), but the value of monitoring for these changes with intracortical EEG remains to be conclusively determined.

Other applications of EEG in the ICU, especially in comatose patients, include evaluating metabolic encephalopathy (EEG serves to substantiate the diagnosis by showing diffusely slow, low-amplitude activity, and often triphasic waves, but does not distinguish between different causes of the condition), recognizing psychogenic unresponsiveness, and confirming brain death (Wijdicks, 2001). After cardiac arrest, near-complete suppression, burst-suppression, nonreactive alpha or theta rhythms (alpha or theta coma), status epilepticus and generalized periodic complexes are considered malignant patterns (Rossetti et al., 2007; Synek, 1990). Although valuable for the prognostication of anoxic-ischemic encephalopathy, EEG data should not be interpreted in isolation in these patients (Wijdicks et al., 2006).

Evoked Potentials

Evoked potentials have a more restricted role in the NICU (Moulton et al., 1998). The median nerve somatosensory evoked potential (SSEP) has been mostly used; technical details are discussed in Chapter 32A. Bilateral absence of the N20 response 1 to 3 days after cardiopulmonary resuscitation accurately predicts poor chances of recovery of awareness (Zandbergen et al., 2006). Unfortunately, presence of these responses after anoxic brain injury lacks meaningful prognostic value.

Continuous monitoring of brainstem evoked potentials and SSEP is now technically feasible. However, the very few studies conducted using these modalities failed to demonstrate any value in early recognition of secondary insults.

Regional Cerebral Blood Flow Monitoring

A major focus in neurointensive care is to ensure that patients maintain adequate CBF. Normal CBF in adult individuals ranges from 45 to 60 mL/100 g/min, and it is higher in the gray matter than in the white matter. Values below 10 mL/100 g/min are considered indicative of ischemia. Determinants of CBF include the status of brain metabolism, Paco₂, systemic BP, hematocrit, and cardiac output. Most of these determinants can be therapeutically manipulated by interventions such as the use of sedatives, changes in the ventilator setting, volume expansion, administration of vasoactive agents, blood transfusions, and inotropic medications. When interpreting information offered by CBF monitoring techniques, it is essential to understand the concept that CBF may be inappropriately low (i.e., metabolic demands exceed supply of blood flow, resulting in ischemia), appropriately low (i.e., metabolic demands are reduced and result in a coupled reduction in blood flow and oxygen consumption), inappropriately high (i.e., cerebral hyperemia), or appropriately high (i.e., situations of increased metabolic demand, such as seizures or fever). There are regional and local techniques for CBF monitoring. Regional modalities include: (1) bedside xenon-133 IV injection technique, (2) stable xenon CT scan, (3) single-photon emission tomography (SPECT), (4) positron emission tomography (PET), (5) perfusion-weighted imaging by magnetic resonance imaging (PWI-MRI), and (6) CT perfusion scans.

The main disadvantage of most of these techniques is that they require transportation of the patient outside of the ICU to the location of the scanner. Consequently, they only provide information about the status of CBF at certain points in time, while CBF is a highly dynamic variable that may fluctuate extensively over time. The bedside xenon-133 technique is the only regional CBF monitoring modality that permits repeated testing in the NICU. However, it requires injection of small doses of the radioactive isotope. The xenon-CT technique involves transporting the patient to the CT scanner and administering non-radioactive xenon gas by inhalation. The inhaled gas can create a euphoric sensation, thus making this technique less desirable in agitated patients. SPECT, PET, MR perfusion, and CT perfusion are valid options for assessing brain perfusion at a certain point in time. PET also allows measurement of the oxygen extraction fraction, which is a reliable indicator of hemodynamic failure and early ischemia when elevated. MRI scanning provides greater anatomical information and the advantage of displaying areas of ischemia on diffusion-weighted imaging. CT perfusion is becoming increasingly available and offers quantifiable perfusion data. However, cumulative exposure to radiation limits the number of CT perfusion scans that can be safely performed for monitoring purposes.

Local CBF monitoring techniques include laser Doppler flowmetry and thermal diffusion flowmetry. Laser Doppler flowmetry is based on assessing the Doppler shift of low-power laser light captured by the moving red blood cells (red cell flux). It produces the continuous real-time flow output, which is linearly related to CBF, thus providing reliable information on local perfusion with excellent dynamic resolution. The main disadvantages of this technique, however, are its invasiveness (requires insertion of the probe via a burr hole), its susceptibility to movement artifact, its small sample volume (1 to 2 cubic millimeters), and the qualitative nature of the information provided (this technique does not allow quantification of CBF, and only relative changes can be assessed). Thermal diffusion flowmetry is used to estimate cortical blood flow by measuring changes in the temperature gradient between two cold plates within a probe applied to the cortex. Advantages include its simplicity and continuous measurement without using ionizing radiation. However, this technique does not provide absolute measures of CBF, and it has not been sufficiently standardized to be recommended for clinical practice.

Transcranial Doppler (TCD) ultrasonography and various brain oxygenation monitoring techniques represent indirect measures of CBF monitoring.

Transcranial Doppler Ultrasonography

TCD ultrasonography is a noninvasive technique used to evaluate mean CBF velocity in the large intracranial arteries at the level of the circle of Willis. TCD is easy to learn and use, noninvasive, and safe. It measures CBF velocity rather than CBF, and the linear relationship between CBF and BFV depends on the angle of insonation. Still, TCD provides a wealth of useful clinical information including the presence or absence of blood flow, its velocity (systolic, diastolic, and mean), and direction. It also allows calculation of the pulsatility index (PI = peak systolic velocity minus end diastolic velocity divided by mean BFV), which represents the downstream resistance to blood flow. Increases in BFV are observed in patients with cerebral vasospasm, hyperventilation (which produces vasoconstriction), and anemia. Cerebral vasospasm may be distinguished from hyperdynamic status by measuring the hemispheric index or Lindegaard ratio (ratio of middle cerebral artery to extracranial internal carotid artery mean BFV) (Lindegaard et al., 1989). A ratio greater than 3 is considered indicative of vasospasm; a low ratio is more suggestive of hyperemia. TCD also allows assessment of vasomotor reactivity (Ng et al., 2000). Impairment of vasomotor reactivity is a well-established poor prognostic factor in patients with TBI and may portend the occurrence of symptomatic vasospasm in patients with SAH (Czosnyka et al., 1997; Frontera et al., 2006b). TCD may also be used as a confirmatory test for the diagnosis of brain death (severely diminished mean cerebral BFV associated with absent diastolic flow, reversed flow, and severely elevated PI).

The diagnosis of cerebral vasospasm in patients with SAH remains the main indication of TCD monitoring in the NICU. The criteria for vasospasm in the middle cerebral artery territory is a mean BFV greater than 120 cm/sec with a hemispheric index greater than 3, or an increment greater than 50 cm/sec within a 24-hour period (Suarez et al., 2002). A specialized headset allows continuous monitoring of BFV and may be a useful adjunct in monitoring patients at high risk for vasospasm. TCD monitoring in patients with cerebral vasospasm has good correlation with angiographic vasospasm and is comparable to conventional angiography in the prognostication of delayed ischemia in these patient, although neither technique is uniformly diagnostic (Rabinstein et al., 2004).

Local Cerebral Oxygenation Monitoring Techniques

Brain tissue oxygen probes and near-infrared spectroscopy allow assessment of local oxygenation. Brain tissue oxygen may be measured by invasive probes such as the Licox catheter. Apart from tissue Po₂, this catheter allows measurement of brain temperature. Brain tissue Po₂ measures the diffusion of dissolved plasma oxygen across the blood brain barrier (rather than CBF, arterial delivery of oxygen, or brain metabolism) in a relatively small area of brain tissue (approximately 15 mm³) (Rosenthal et al., 2009). Factors that determine brain tissue Po₂ include Pao₂, arterial Paco₂, systemic BP, and CBF. Normal brain tissue Po₂ values range from 25 to 30 mm Hg. The major disadvantages of brain oxygen probes include their invasiveness, limited spatial resolution, and susceptibility to artifacts (due to inappropriate calibration and head movement, among other factors). Its use has been recommended by experts in various major centers for patients with severe head injuries and poor-grade SAH (Maloney-Wilensky et al., 2009). It is best used when applied in the setting of multimodality monitoring, along with jugular oximetry and perhaps microdialysis (Andrews et al., 2008) (Fig. 45.4).

Fig. 45.4 Example of multimodality monitoring in a patient with traumatic brain injury. Notice the evidence of local and regional hypoxia with elevation of intracranial pressure leading to reduction in cerebral perfusion. CPP, Cerebral perfusion pressure; FV, flow velocities on transcranial Doppler; ICP, intracranial pressure; Ptio₂, oxygen pressure in brain tissue; Sjo₂, oxygen saturation in jugular blood.

Near-infrared spectroscopy (NIRS) is based on the property of a near-infrared light (700–1000 nm) to pass through tissues while being both scattered and absorbed. The absorption of a near-infrared light is proportional to the local concentration of certain chromophores, most notably hemoglobin. Thus, the absorption of near-infrared light changes according to the oxygenation state of hemoglobin. The probes illuminate up to a volume of 10 mL of brain tissue. All measurements are expressed as absolute concentration changes from a baseline zero at the start of the measurement. Normal values of oxygenated hemoglobin are reported to be 60% to 80%, and ischemic threshold is estimated to be below 47% saturation (Casati et al., 2006). However, the reliability of this technique has been questioned. It is susceptible to extraneous light, motion artifact, and signal drift. The measurement may also become unreliable when obtained through intracranial hematomas or through blood in the CSF.

Microdialysis

The basic concept of microdialysis involves inserting a fine catheter into the brain parenchyma, then perfusing the catheter with a physiological solution such as Ringer’s lactate, thereby facilitating the exchange of molecules between the perfusate and the extracellular fluid across a dialysis membrane located within the catheter tip. The dialysate is sampled under sterile conditions at hourly or other regular intervals and put through a microdialysis analyzer at the bedside. Insertion artifacts make measurements unreliable for the first hour after placement (Bellander et al., 2004).

Microdialysis allows monitoring of brain pH, lactate and pyruvate, glucose, glycerol, glutamate, urea, and potentially other soluble molecules of interest (Bellander et al., 2004; Nilsson et al., 1999; Vespa et al., 1998). Changes in lactate concentration, lactate/pyruvate ratio, and glutamate concentration have been used as indices of cerebral ischemia. A lactate/pyruvate ratio greater than 25 is probably the best indicator of ischemia (Andrews et al., 2008; Bellander et al., 2004). Rises in glycerol are believed to reflect phospholipid breakdown as a result of cell membrane damage. Because of this, cerebral microdialysis has been employed in the NICU to monitor for cerebral vasospasm and delayed cerebral ischemia in SAH, to identify secondary insults after severe brain trauma, and to follow extracellular glutamate concentration peri-ictally in patients with epilepsy.

Several aspects of microdialytic analysis remain controversial, such as where to place the catheter (Andrews et al., 2008), whether the lactate/pyruvate ratio alone or in combination with other parameters is a better indicator of early cerebral ischemia, and why there has been no correlation between microdialysis measures and clinical outcome in some studies. Other problems are the invasiveness and labor intensiveness of the technique. It currently represents a valuable research tool, but its widespread clinical use cannot yet be recommended.

Analgesia and Sedation

Analgesia and sedation are essential practices in neurointensive care. It is always challenging to avoid confounding the neurological examination while keeping the patient comfortable. Distinguishing agitation from the psychomotor manifestations of pain may be a difficult task in acutely ill neurological patients. Anxiety and pain lead to stress responses characterized by a hyperdynamic circulation, increased metabolic rate, and hyperkinesis. When pain is the cause of restlessness, the abnormal behavior may be more refractory to sedative medications. In those cases, appropriate and timely use of analgesics may result in correction of the abnormal behavior.

Whenever a patient is agitated in the NICU, an organic cause for the agitation should be sought. Confusion and restlessness are often seen in patients with acute strokes (especially those involving the right parietal lobe or the left posterior cerebral artery territory), early after SAH, in TBI with bifrontal damage, and in certain postictal states. On the other hand, patients with neuromuscular respiratory failure may become extremely anxious as they fail to achieve adequate ventilation and gas exchange. Drug withdrawal and side effects from medications are other common causes of agitation in the NICU. Metabolic and endocrine derangements should also be investigated in agitated critically ill patients.

The ideal sedative agent in the NICU would be one that can achieve sedation rapidly and allow for fast and reliable reversal of its effect. Propofol is one of the most commonly used drugs because it fulfills the criteria to a great extent. Propofol crosses the blood-brain barrier within minutes of administration and is a markedly effective hypnotic agent. Another advantageous pharmacokinetic property of propofol is that its clearance is not significantly altered by liver or renal failure, which is often a problem with benzodiazepines and opiates. Propofol lacks amnestic properties, so adequate analgesia should always be ensured in patients receiving this drug.

Awakening is typically seen within minutes of discontinuation of the infusion, but time to awakening may be significantly prolonged when the drug has been used in large doses for several days, because propofol is redistributed to the fat tissue from where it is only slowly released. Hypotension is the most common side effect of propofol infusion, especially when administered as a bolus. Falls in BP are more frequent and pronounced in patients who are hypovolemic. Other adverse effects include caloric overload (1 mL of propofol contains 1 kilocalorie), hypertriglyceridemia, and withdrawal myoclonus (often confused with seizures). Propofol can also be used to treat elevated ICP and status epilepticus (Parviainen et al., 2006). However, administration of high doses of propofol for prolonged periods of time (i.e., >4-5 mg/kg/h for more than 48 hours) can cause the propofol infusion syndrome (Kam and Cardone, 2007). This is a serious complication characterized by metabolic acidosis, rhabdomyolysis, refractory bradycardia, myocardial depression, and when most severe, cardiac arrest (Iyer et al., 2009). Even strict surveillance for these manifestations may fail to prevent this life-threatening complication. Consequently, propofol should be used with great caution for the treatment of recalcitrant intracranial hypertension and status epilepticus, indications in which high doses of the medication are often necessary for up to several days to achieve the therapeutic goal.

Midazolam and lorazepam are the two most commonly used benzodiazepines in the NICU. The advantages of midazolam are its rapid effect and short duration of action (half-life 1.9 hours); it has only one active metabolite. Clearance is fast, but accumulation may occur after 3 days of continuous infusion. Patients who receive midazolam for several days can be expected to exhibit delayed awakening. Clearance of midazolam is diminished by hepatic and renal failure. Lorazepam has a much longer half-life (14 hours), which leads to a much slower emergence from sedation. However, continuous infusion of lorazepam can produce severe metabolic acidosis from propylene glycol toxicity (Arroliga et al., 2004). The main side effect of benzodiazepines is respiratory depression in patients who are not mechanically ventilated. They can also induce hypotension in patients with reduced intravascular volume. Risk of withdrawal symptoms is small. Benzodiazepines are effective in the treatment of status epilepticus, but pharmacoresistance emerges over time and requires progressive increase in the rate of infusion of the drug. They do not have a significant effect on ICP.

Unlike propofol, benzodiazepines have an effective antidote in flumazenil, which is a benzodiazepine receptor antagonist with little or no agonist activity and a half-life of 0.5 to 1.3 hours. Its administration is free of negative cardiovascular effects but may be complicated by the occurrence of seizures. The risk of seizures after administration of flumazenil is relatively small, except in patients with a history of epilepsy or with a significantly reduced seizure threshold. Therefore, caution should be exercised when administering this medication to patients with acute brain disease.

Dexmedetomidine is a selective α₂-adrenergic receptor agonist. It produces effective sedation while preserving patient’s alertness. Patients sedated with this medication are often easily aroused, so adequate neurological assessment may be performed without the need to temporarily discontinue the sedative infusion. Upon initial administration, this drug may produce transient mild hypertension followed by a temporary drop in BP. It has a short elimination half-life of approximately 2 hours, though this may be prolonged in patients with liver failure. Caution is recommended when using this medication in patients with severe bradycardia or abnormal cardiac conduction, in patients with a severely depressed cardiac ejection fraction, and in patients who are hypovolemic or hypotensive at the time of infusion.

Haloperidol is the drug of choice for patients with signs of psychosis. Intravenous doses of haloperidol may achieve successful control of agitated psychotic behavior within 20 minutes. The drug is fairly safe, but its use can be complicated by the appearance of extrapyramidal signs and (rarely) by neuroleptic malignant syndrome. Haloperidol should be used with caution in patients with prolongation of the QT interval. New-generation antipsychotics (atypical antipsychotics such as risperidone, olanzapine, and quetiapine) are also useful in the management of agitation and delirium. However, they can only be administered by the enteric route, and their therapeutic effect is slower, hence they should not be used to control acute severe agitation.

Opiates represent the mainstay of analgesic treatment in acutely ill neurological patients. Options include morphine, fentanyl, sufentanil, alfentanil, and codeine. They all produce analgesia, reduced level of consciousness, and respiratory depression. Hypotension may occur in hypovolemic patients or when using high doses of these medications. Fentanyl is preferred to morphine because it provokes fewer cardiovascular side effects and does not produce histamine release. Codeine is a much less potent agent, and its role is limited in the NICU. The action of opioids may be reversed by using naloxone, a competitive antagonist. Hypertension and cardiac arrhythmias are potential side effects of naloxone use. For milder forms of pain, nonsteroidal antiinflammatory agents (e.g., ketorolac), tramadol, and acetaminophen may be used. Table 45.2 summarizes key pharmacokinetic and pharmacodynamic information for the most commonly used sedative and analgesic agents in the NICU.

Airway and Ventilatory Assistance

Acutely ill neurological patients typically develop respiratory failure because of inability to oxygenate or to sustain their ventilatory needs. The most common causes for oxygenation failure in NICU patients are cardiogenic and neurogenic pulmonary edema. However, sudden hypoxia should always raise the suspicion for pulmonary embolism (PE). Aspiration pneumonia occurs frequently because of inability to protect the airway in patients with depressed level of consciousness and impaired cough reflex. Atelectasis may be a cause of hypoxia in patients with neuromuscular weakness. Hypercapnia is the hallmark of patients with ventilatory failure. Ventilatory insufficiency from upper airway collapse is encountered in patients with neuromuscular respiratory failure or coma.

At the time of endotracheal intubation, the two main complications in neurological patients are a rise in ICP and exacerbation of hypoxia. Rapid-sequence intubation is the safest approach for patients with increased ICP (Wijdicks and Borel, 1998). Rapid-sequence intubation proceeds in three phases: (1) preoxygenation to prevent worsening hypoxia during intubation—this can be achieved by providing effective bag-valve-mask (AMBU) ventilation; (2) pretreatment with drugs to mitigate the hemodynamic changes that may increase ICP upon intubation (e.g., lidocaine, thiopental); and (3) sequential administration of a potent sedative (e.g., propofol) and, when necessary, a rapidly acting nondepolarizing neuromuscular blocking agent (e.g., rocuronium, vecuronium). Succinylcholine should be avoided because it may increase ICP due to widespread muscle fasciculations, increased central venous pressure, and hypercarbia, and because it can produce dangerous hyperkalemia in patients with underlying muscle disease. In cases of TBI, it is also essential to maintain in-line stabilization of the cervical spine. When cervical spine injury is suspected, fiberoptic-assisted intubation is preferred.

The essential goal of mechanical ventilation is to assist the patient to achieve adequate gas exchange. There are two basic forms of mechanical ventilation: volume control and pressure control. Volume-control ventilation delivers a consistent preset volume of air with each ventilator breath. Pressure-control ventilation delivers a preset amount of pressure to the patient, with varying degrees of volume, depending on the amount of resistance in the system.

The modes of volume-control ventilation most frequently used in neurological and neurosurgical patients are assist/control (A/C) and synchronized intermittent mandatory ventilation (SIMV). In A/C ventilation, the ventilator will always deliver the preset air volume. In control mode, breaths are initiated by the machine and not influenced by the patient. The rate of these controlled breaths is determined in the ventilatory settings. In assist mode, the ventilator will deliver extra breaths of the same predetermined tidal volume every time the patient generates sufficient negative pressure during an attempted inspiration. In SIMV, the ventilator delivers breaths with full preset volume up to a prescribed rate. If the patient’s inspiratory effort exceeds such preset rate, all additional breaths initiated by the patient (spontaneous breaths) will have a volume determined by the extent of the negative inspiratory pressure produced by the patient. The volume of this spontaneous breath may be adjusted by setting a support function on the ventilator called pressure support. Thus, pressure support is used in conjunction with SIMV to augment the patient’s negative inspiratory force and increase the efficiency of the independent breaths produced by the patient. SIMV is well tolerated by patients and avoids deconditioning of respiratory muscles. Its main disadvantages include the possibility of developing high peak airway pressures, high rate of gas delivery in the early phase of inspiration (which may not be tolerated by agitated patients), and insufficient treatment of hypoxia in severely hypoxemic patients.

Pressure-control ventilation differs from volume control mode in that inspiratory and expiratory airway pressures are consistently regulated at the expense of variation in the delivered volume. Pressure-control ventilation is used most frequently in patients who are sedated and paralyzed. It requires setting the fraction of inspired oxygen (Fio₂), the ventilatory rate, and the pressure difference between inspiration and expiration.

In pressure-support ventilation (PSV), all breaths are triggered by the patient. The ventilator delivers a particular level of pressure support each time a breath is initiated; this pressure is delivered at the onset of inspiration. When the flow rate reaches 20% of its initial value, gas flow is terminated. This mode is often fairly comfortable for the conscious patient, as it closely approximates the flow characteristics of a normal breath. Since patients on a pressure-support ventilator may become hypopneic, they should be closely monitored with the help of adequately set apnea alarms.

Regardless of the ventilatory mode used, oxygenation depends upon the Fio₂ and the level of positive end-expiratory pressure (PEEP) provided. Increasing the Fio₂ increases the oxygen available for absorption by the pulmonary capillaries. Very high levels of Fio₂ (>0.5) may result in pulmonary oxygen toxicity. PEEP allows for tapering of the Fio₂ in many cases. The basic goal of PEEP is to prevent microatelectasis by keeping alveoli from collapsing at the end of expiration. This improves the efficiency of gaseous exchange by maximizing recruitment of lung units. The main danger of PEEP use is increasing intrathoracic pressure to levels that compromise venous return. This may result in hypotension unless intravascular filling pressures are augmented by volume expansion. High levels of PEEP may also produce tension pneumothorax. Finally, it is important to monitor the ICP when positive airway pressure is applied. Patients with decreased intracranial compliance may develop increases in ICP as intrathoracic pressure rises and imposes resistance to venous return. However, for the most part, relatively high levels of PEEP are well tolerated by euvolemic patients with intracranial hypertension.

Weaning from mechanical ventilation is usually achieved in critically ill neurological patients by decreasing the rate of mandatory breaths on SIMV or by using PSV. In fact, both methods can be combined in practice. A patient on SIMV can have the set rate decreased as clinical improvement occurs. If the patient has adequate spontaneous tidal volumes and no apneas, he or she may be switched to PSV, which consists of pressure support at the onset of inspiration and PEEP to prevent alveolar collapse and improve oxygenation. Subsequently, the amount of pressure support may be weaned until extubation is deemed safe.

In patients with acute brain disorders, level of consciousness may a limiting factor when considering extubation. Despite successful weaning, the stuporous patient may be considered unsafe for extubation because of concerns about airway safety. Keeping patients intubated once they have fulfilled the ventilatory criteria for extubation is a common but questionable practice. In patients with TBI, this practice may be associated with a higher risk of ventilator-associated complications (Coplin et al., 2000). Thus, safety of extubation in patients with adequate respiratory function but persistently depressed level of consciousness is a problem that demands further research (Manno et al., 2008).

Patients who fail extubation and those who are considered unsafe for an extubation trial require tracheostomy. The timing of a tracheostomy varies according to the patient’s primary condition. Local airway complications increase with longer duration of endotracheal intubation. In addition, tracheostomy is more comfortable for patients than endotracheal intubation and provides better access for effective pulmonary toileting. The most common indications for tracheostomy in the NICU are persistent stupor or coma, severe impairment of cough reflex, and prolonged neuromuscular respiratory failure. Percutaneous tracheostomy has become the standard procedure in most ICUs. It is important to bear in mind that tracheostomies are reversible. Also, specially modified tracheostomy tubes that allow patients to vocalize and communicate are now available.

Pulmonary Complications

The main respiratory complications in critically ill neurological patients are pneumonia (either induced by aspiration or ventilator-associated), PE, atelectasis, and pulmonary edema (either cardiogenic or neurogenic). Aspiration is common in patients with depressed level of consciousness, seizures, or bulbar weakness. Patients who have been intubated for over 48 hours may develop ventilator-associated pneumonia, manifested by increased amount of thick secretions, fever, leukocytosis, new radiographic abnormalities, and increased Pao₂:Fio₂ ratio. Aspiration pneumonia should prompt coverage for anaerobes and gram-negative organisms. Coverage for ventilator-associated pneumonia will depend on the organisms and susceptibility most prevalent on each ICU.

Sudden development of unexplained hypoxia should be considered possible PE until proven otherwise. Patients with critical neurological illness are especially predisposed to the development of venous thromboembolism because of prolonged immobility. Tachypnea is often prominent in patients with PE. However, quadriparetic patients with high cervical lesions cannot develop this response, and oxygen desaturation associated with tachycardia may be the only manifestation in these patients. The differential diagnosis in cases of acute tachypnea and oxygen desaturation includes plugging of the airway by secretions. However, these patients typically also develop hypercapnia due to hypoventilation. If hypoxia is not resolved by airway suctioning and the situation remains unexplained after an emergent chest radiograph (to exclude new infiltrates, pneumothorax, or lobar collapse), the patient should undergo specific studies to rule out PE. At present, CT angiograms are the diagnostic modality of choice. The possibility of deep venous thrombosis (DVT) should also be investigated by venous Doppler of the lower extremities. The treatment of venous thromboembolism may be particularly challenging in acute neurological patients. Patients with large ischemic strokes or intracranial hemorrhages are at increased risk for complications from IV heparin. When systemic anticoagulation is deemed strictly contraindicated, insertion of a vena cava filter may be a reasonable alternative. Patients with massive PE may require endovascular maneuvers to mechanically remove the clot or intraarterial infusion of a thrombolytic agent.

Atelectasis is very common in patients receiving mechanical ventilation. Large areas of atelectasis or lobar collapse may produce profound hypoxia. Mucous plugging of the airway is common among critically ill patients. Increasing levels of PEEP are often used to treat collapsed lung regions. Physical measures including suctioning, postural drainage, and external percussion may be effective, but bronchoscopic suction and lavage are necessary in severe cases.

Interpretation of pulmonary edema is more complex in critically ill neurological patients than in the general population of ICU patients. While most cases of pulmonary edema will be due to cardiac failure, neurogenic pulmonary edema may occur after acute SAH, TBI, and other neurological catastrophes associated with massive surge of central sympathetic output. Neurogenic pulmonary edema is successfully treated using high levels of PEEP. Cardiogenic pulmonary edema should be treated by ameliorating cardiac workload (through diuresis and vasodilatation) and providing adequate levels of supplemental oxygen.

Cardiovascular Care and Blood Pressure Management

Cardiac disorders are common in critically ill neurological patients, and they may precede or accompany the neurological illness. They are often related to the massive catecholamine release associated with the acute brain insult (Banki et al., 2005). The most common forms of cardiac complications in the NICU are acute coronary syndrome, cardiac arrhythmias, and congestive heart failure.

Acute Coronary Syndrome

Electrocardiographic (ECG) and clinical abnormalities suggestive of myocardial ischemia are fairly common in patients with acute brain injury (e.g., large ischemic stroke, SAH, large intraparenchymal hematoma, TBI with contusions, status epilepticus). Typical ECG abnormalities in patients with acute brain damage include symmetrically inverted T waves (Fig. 45.5) and sometimes ST-segment elevation across all the precordial leads. Elevation of serum troponin levels should be considered indicative of myocardial injury, whereas elevation of serum creatinine kinase is much less specific in patients with acute brain damage (Woodruff et al., 2003). Yet, troponin elevation is seen in patients with SAH as an expression of ventricular dysfunction secondary to the neurogenic (adrenergic-induced) injury (Deibert et al., 2003).

Fig. 45.5 Electrocardiographic changes in a patient with acute aneurysmal subarachnoid hemorrhage. Notice diffuse repolarization abnormalities in the precordial leads.

It is always difficult to define optimal hemodynamic goals in patients with coexistent myocardial ischemia and acute neurological conditions that require maintenance of adequate CPP, such as acute ischemic stroke and SAH at risk for vasospasm. In these patients, lowering the BP to the levels commonly used as goals in most patients with acute myocardial ischemia may further compromise cerebral perfusion and precipitate infarction in areas of ischemic penumbra. Anticoagulation or IV glycoprotein IIb/IIIa inhibitors may be contraindicated early after an extensive ischemic stroke, in patients with a large intraparenchymal hematoma, or shortly after a neurosurgical procedure. Percutaneous coronary angioplasty and stenting may be considered, but limitations on the use of aspirin and clopidogrel after the intervention may increase the risk of acute in-stent thrombosis. Induced diuresis is indicated to reduce afterload in patients with depressed left ventricular ejection fraction, but it should be closely monitored; hypovolemia may induce cerebral ischemia in patients with vasospasm or areas of ischemic penumbra.

Fluid and Electrolytes

Acute renal failure in acutely ill neurological or neurosurgical patients is most commonly iatrogenic. Mannitol can rapidly cause prerenal azotemia when adequate hydration is not provided to compensate for the fluid lost from osmotic diuresis. This complication can be reliably avoided by adjusting the fluid intake to prevent negative fluid balance, while monitoring serum osmolality. If serum osmolality exceeds 320 mOsm/kg, mannitol infusion is typically withheld to protect renal function. If continuation of osmotherapy is indispensable, mannitol may be continued with relatively low risk of kidney failure, so long as concomitant aggressive hydration is provided. Hypertonic saline may be a valuable alternative in these cases; it is a safer choice than mannitol in patients with chronic renal insufficiency. Radiocontrast-induced nephropathy can be prevented by preemptive hydration, N-acetylcysteine, and bicarbonate infusion (Merten et al., 2004; Tepel et al., 2000). Acute interstitial nephritis from drug toxicity (e.g., antibiotics) and (less commonly) pyelonephritis (in patients with chronic indwelling catheters) are also causes of acute renal failure in the NICU.

When renal failure is established, it is essential that the neurointensivist cooperate with the consulting nephrologist to maximize the safety of renal replacement therapy (dialysis). Sudden fluid shifts and changes in BP that would be inconsequential in other patients may have dramatic detrimental effects in patients with cerebral edema or cerebral hypoperfusion. In patients with renal failure, it is also important to closely monitor the free levels of anticonvulsive drugs. These patients not only have heightened risk of developing toxic complications from decreased elimination of the drug, but rapid clearance of the anticonvulsant during dialysis may increase the risk of seizures.

Hyponatremia is the most common electrolyte imbalance encountered in critically ill neurological patients. The two most common mechanisms of hyponatremia in these patients are cerebral salt-wasting syndrome (CSWS) and the syndrome of inappropriate secretion of antidiuretic hormone (SIADH) (Rabinstein and Wijdicks, 2003a). Both mechanisms produce hypotonic hyponatremia with high concentration of urinary sodium (secondary to increased sodium excretion in CSWS and increased water reabsorption in SIADH). In fact, determination of extracellular fluid volume remains the only reliable distinguishing feature between these two conditions: SIADH is a state of volume expansion, while CSWS is a state of volume depletion (Table 45.4). The practical importance of this concept needs to be highlighted because fluid restriction—adequate therapy for SIADH—may be enormously deleterious in patients with CSWS, as is the case in SAH. Symptomatic acute hyponatremia requires tightly controlled infusion of hypertonic saline. Excessively rapid correction of profound chronic hyponatremia may precipitate severe osmotic myelinolysis. The rate of correction should not exceed 10 mmol/L over any 24-hour period to avoid this potentially devastating complication (Laureno and Karp, 1997). Fig. 45.6 presents an algorithm for the diagnosis and management of hyponatremia in critically ill neurological patients.

Table 45.4 Clinical and Laboratory Features of CSWS and SIADH

↔, Absent or minor variable change; ↓, decreased; ↑, increased; CSWS, Cerebral salt-wasting syndrome; CVP, central venous pressure; PCWP, pulmonary capillary wedge pressure; SIADH, syndrome of inappropriate secretion of antidiuretic hormone.

Fig. 45.6 Algorithm for the diagnosis and management of hyponatremia in critically ill neurological patients.

Hypernatremia in the NICU is most often produced by therapeutic interventions (e.g., mannitol without sufficient fluid replacement, infusion of hypertonic saline) or diabetes insipidus (DI). Focal brain lesions (most frequently tumors or trauma) or surgery involving the sellar/suprasellar region are the typical causes of DI. Profound DI is also seen at the time of brain death. Diagnosis of DI hinges on the finding of polyuria (characteristically > than 250–300 ML/h for ≥ 2 consecutive hours) with very dilute urine (specific gravity <1.010, urine osmolality <250 mOsm/kg). Treatment demands aggressive fluid replacement. Central DI responds rapidly to administration of vasopressin or desmopressin acetate (DDAVP). Vasopressin is short-acting (2-4 hours), and the recommended dose is 2 to 5 units subcutaneously or intramuscularly every 4 hours. Desmopressin acetate has a longer duration of action and should be administered cautiously in postsurgical patients because, in those cases, DI tends to resolve spontaneously within days or even hours of its presentation. The recommended dose is 0.5 to 4 µg IV or subcutaneously every 12 hours. Serum and urine osmolalities and serum electrolytes should be checked every 2 to 4 hours in every patient with DI.

Nutrition and Metabolic Derangements

Adequate nutrition is essential for the recovery of critically ill patients, including those with primary acute neurological conditions. Depressed level of consciousness and abnormal swallowing function are very prevalent in NICU patients, who consequently often require tube feedings. Meanwhile, gastroparesis is also common and may increase the risk of aspiration in patients receiving enteral nutrition. This potential risk demands close monitoring of gastric residuals and positioning of the feeding tube in the distal part of the stomach or first portion of the duodenum. Agents that promote gastric motility (e.g., metoclopramide) may be added in the most severe cases (Booth et al., 2002).

Daily caloric requirement is calculated using the Harris Benedict equation to estimate basal energy expenditure (BEE):

For both equations, W is body weight in kilograms, H is height in centimeters, and A is age in years. This is often operationally translated into approximately 25 to 30 kcal/kg/day. Nutritional requirement may be adjusted to the particular disease and nutritional status indicators. For instance, sepsis may require increasing nutritional support by 30%, whereas high-caloric feeding should be avoided in Guillain-Barré syndrome and myasthenia gravis. Adequacy of nutritional support is better assessed using prealbumin (half-life of 2-3 days) rather than albumin (half-life of 20 days). Working closely with a nutritionist specialized in critical care patients is highly advisable to make adequate adjustments.

Enteral feeding is preferred whenever possible to help maintain the integrity of the intestinal mucosal lining. It is recommended to start feeding patients early (ideally within 48 hours of admission); early feeding has been associated with a trend toward better survival and less disability in patients with TBI (Yanagawa et al., 2002). The optimal timing of percutaneous gastrostomy in neurological patients has not been sufficiently studied. There is some evidence that gastrostomy should be performed in patients with dysphagia from stroke persisting after 14 days (Norton et al., 1996).

Hyperglycemia is the most frequent metabolic derangement in critically ill neurological patients. There is solid evidence that hyperglycemia activates neurotoxic oxidative and inflammatory responses after acute ischemia. In patients with acute ischemic stroke, hyperglycemia has been associated with increased risk of hemorrhagic transformation and hyperacute worsening and lower rates of recanalization after thrombolysis (Alvarez-Sabin et al., 2004; Leigh et al., 2004; Ribo et al., 2005). It has also been found to correlate with infarct expansion and worse functional outcome (Baird et al., 2003; Bruno et al., 2002). Similarly, functional recovery is poorer in hyperglycemic patients with ICH and SAH (Frontera et al., 2006a; Passero et al., 2003). Intensive insulin therapy to maintain strict normoglycemia is no longer recommended for critically ill patients, after this practice led to increased mortality in a large randomized trial (Finfer et al., 2009). However, strict blood sugar control could decrease the rate of critical illness polyneuropathy (Van den et al., 2005). In the NICU, intensive insulin therapy carries the risk of inducing neuroglycopenia (with the ensuing risk of energy failure), which may occur in patients with acute brain insults even with serum glucose concentrations within the usual normal range (Godoy et al., 2010; Oddo et al., 2008). Therefore, the treatment of hyperglycemia must be particularly cautious in patients with acute brain disease, and serum glucose concentrations below 100 to 110 mg/dL should be avoided.

Fever and Infections

Fever in a patient with acute brain disease demands prompt diagnostic investigation to determine its cause and symptomatic treatment to avoid the deleterious impact of hyperthermia on the injured brain. Experimental models have consistently shown that even mild hyperthermia worsens cerebral damage after ischemia or trauma (Baena et al., 1997; Dietrich et al., 1996; Kim et al., 1996). Fever has been associated with poor functional outcome in patients with ischemic infarction (Reith et al., 1996; Wang et al., 2000), ICH (Schwarz et al., 2000), SAH (Oliveira-Filho et al., 2001), and TBI (Jiang et al., 2002). Increased metabolic expenditure, exacerbation of excitotoxicity, and elevated ICP may be responsible for the detrimental effects of hyperthermia (Rossi et al., 2001; Thompson et al., 2003).

Fever is very prevalent among NICU patients. Although disturbances in central thermoregulation occur frequently in patients with acute brain disorders (Rabinstein et al., 2007), infections are a common cause of fever in the NICU and should always be excluded (Commichau et al., 2003). Pneumonia, urinary tract infection, and bloodstream infection are the most frequent infectious complications (Dettenkofer et al., 1999). Ventriculitis must always be ruled out in patients with ventriculostomy. The appearance of fever in an NICU patient should be evaluated with cultures of blood, urine, respiratory secretions (ideally collected by bronchoalveolar lavage), and CSF (always in patients with ventriculostomy and when deemed clinically indicated in others). Chest radiograph and urine microscopic analysis are also pertinent. CT scan of the sinuses may be added to the diagnostic evaluation in febrile patients who have been intubated for several days. CT scan of the abdomen and pelvis is sometimes necessary to detect abscesses, pancreatitis, or cholecystitis. The skin should be thoroughly searched for signs of cellulitis or phlebitis. Osteomyelitis and discitis must be included in the differential diagnosis of fever after spine surgery.

When an infection is suspected, empirical antibiotic therapy is reasonable. It should cover for the most likely responsible organisms, depending on the patient’s risk factors and the local microbiological resistance patterns of the ICU. Empirical antibiotics may be discontinued after 3 days if no infection is documented. It is always prudent to consider changing indwelling catheters (central venous, arterial, bladder, ventricular) in persistently febrile patients.

Drug reactions, DVT, ethanol withdrawal, pancreatitis, and gout are relatively common causes of noninfectious fever in the NICU. Drug fever is most frequently caused by phenytoin in neurological patients. Signs of anticonvulsant hypersensitivity syndrome (rash, lymphadenopathy, hepatomegaly, eosinophilia, elevation of liver transaminases) must be readily recognized, since failure to discontinue the culprit medication promptly may have the devastating consequence of Stevens-Johnson syndrome (Schlienger and Shear, 1998). Detailed physical examination, venous Doppler, and measurement of liver and pancreatic enzymes should be performed in NICU patients with fever of unclear cause. In chronically immobile patients (especially those with spinal cord injury), heterotopic ossification may be a cause of persistent fever; it may be suspected by marked elevation of the C-reactive protein and sedimentation rate and confirmed by bone scintigraphy.

Central fever remains a diagnosis of exclusion. It occurs most frequently in patients with SAH, and in those patients it is associated with increased risk of vasospasm (Oliveira-Filho et al., 2001; Rabinstein et al., 2007). Patients with central fever often have prolonged hyperthermia with failure to return to normal body temperature, as opposed to the spikes of fever followed by normothermia typically observed with infections. Among patients with TBI, high fevers may be accompanied by other manifestations of paroxysmal sympathetic hyperactivity (tachycardia, hypertension, diaphoresis, dystonia) (Rabinstein, 2007).

Multiple measures can be used to normalize body temperature in febrile patients. Antipyretic medications (acetaminophen, ibuprofen) are sufficient in milder cases. However, mechanical cooling methods must be added in patients with more severe or refractory hyperthermia. Ice packs, air- or water-circulating cooling blankets, and effective cooling vests are alternative methods of conductive cooling (Mayer et al., 2004; Seder and Van der Kloot, 2009). Endovascular cooling devices may offer greater control of temperature modulation but require placement of a central venous catheter (De Georgia et al., 2004; Seder and Van der Kloot, 2009). Patients should be monitored for the appearance of shivering, which can be treated with warming gloves, buspirone, meperidine (in patients without high risk for seizures), magnesium infusion, or dexmedetomidine in patients who are awake, but it may necessitate neuromuscular paralysis when severe.

Hematological Complications

The risk of DVT is increased in immobilized patients. The incidence of clinical DVT after acute ischemic stroke ranges between 1% and 5%, and clinical PE occurs in 0.5% to 3.5% of these patients (Kamphuisen et al., 2005). However, the incidence of subclinical DVT is much higher when assessed by ultrasound, venography, or nuclear scans (Kamphuisen et al., 2005). The risk of DVT is also increased after craniotomy (Hamilton et al., 1994), ICH (Lacut et al., 2005), and in patients with severe neuromuscular weakness. The main diagnostic test for DVT is noninvasive bedside vascular ultrasound (venous Doppler). Physical examination is relatively insensitive to detect DVT in acutely ill hospitalized patients. When PE is suspected, spiral CT angiogram of the chest should be performed.

Early mobilization should be promoted in all patients. Options for prevention of thromboembolic complications in immobilized patients include mechanical methods (compressive stockings, intermittent pneumatic compression) and antithrombotics. Current evidence supports the use of subcutaneous anticoagulants for most patients with acute ischemic stroke (Adams, Jr. et al., 2007; Kamphuisen et al., 2005) and after craniotomy (Iorio and Agnelli, 2000). Unfortunately, similar data are not available to guide the management of thromboprophylaxis in patients with ICH or large ischemic cerebral infarction. Enoxaparin (40 mg daily) was found to be superior to unfractionated heparin (5000 units twice daily) in one randomized controlled trial of patients with acute ischemic stroke (Sherman et al., 2007). In cases of documented DVT and high risk of hemorrhagic complications from anticoagulation, placement of a Greenfield filter in the inferior vena cava is a valuable alternative.

The optimal hematocrit level in critically ill neurological patients has not been adequately studied and probably varies according to the underlying primary disease process. Mild hemodilution may improve the rheological properties of the cerebral circulation, but excessive anemia may compromise oxygen delivery. Transfusions are only indicated for general critically ill patients when the hemoglobin concentration is lower than 7 g/dL. However, the appropriateness of this conservative practice in patients with acute brain damage (who may be particularly sensitive to local or regional hypoxia) remains to be established.

Thrombocytopenia in the NICU is most commonly associated with exposure to heparin or other drugs (e.g., valproic acid, antibiotics). Heparin-induced thrombocytopenia may be diagnosed by the presence of circulating serum antibodies to platelet factor 4. Discontinuation of heparin results in prompt normalization of the platelet count. In patients with heparin-induced thrombocytopenia who still have indication for continuing therapeutic anticoagulation, a direct thrombin inhibitor (recombinant hirudin or argatroban) should be used instead of heparin.

Cerebral Protection after Cardiac Arrest

Therapeutic hypothermia has been shown to improve clinical outcomes after witnessed out-of-hospital ventricular fibrillation arrest (Bernard et al., 2002; the Hypothermia after Cardiac Arrest Study Group, 2002), and induction of hypothermia has become standard of care for the management of this condition. In fact, many centers have adopted the practice of inducing hypothermia after all cardiac arrests, regardless of the initial rhythm. Therapeutic hypothermia serves primarily as a neuroprotective modality, allowing more patients to recover awareness and improving functional outcomes among survivors.

Protocols for induction of hypothermia after cardiac arrest may vary across centers (e.g., methods of induction ranging from cooling blankets and ice packs to more sophisticated surface or intravascular cooling devices), but there is uniformity of criteria on its main principles: (1) cooling should be started as soon as feasible and continued for 24 hours, (2) target temperature should be ideally maintained at 33°C, (3) shivering is treated with paralytic agents, and (4) rewarming should take place over 12 to 18 hours.

The degree of hypothermia used for treatment of cardiac arrest is usually well tolerated. Electrolyte imbalances, especially hypokalemia during induction and hyperkalemia upon rewarming, are the most common side effects. Hypotension may result from increased diuresis due to peripheral vasoconstriction. Bradycardia is frequent but usually asymptomatic. Life-threatening arrhythmias are rare, even in patients with myocardial ischemia. The risk of pneumonia is frankly increased when the duration of hypothermia exceeds 48 hours but not in cardiac arrest patients cooled for 1 day. Hyperglycemia from decreased insulin sensitivity, reduced intestinal motility, and mild coagulopathy (very infrequently leading to clinically significant bleeding) are also side effects of hypothermia.