Fluid management of the neurological patient: a concise review

Osmolality of plasma and brain interstitial fluid and CSF are equal under normal circumstances [8]. Hypotonic fluids cause water shifts to the brain because the blood–brain barrier (BBB) is water permeable whereas hypertonic fluids are well known for their ability to cause brain dehydration, both when the BBB is intact and is disrupted [9, 10]. Neurons can compensate for such fluid shifts by active solute depletion to the extracellular compartment to cause reactive ‘shrinkage’, and the BBB endothelial and other highly specialized cells within the so-called neurovascular unit will operate similarly to expel water to the intravascular compartment [11]. However, BBB disruption locally abolishes its ability to control homeostasis of electrolytes, water and other solutes, and fluid shifts will become more dependent on local pressure differences between the intravascular and extravascular compartment than osmotic tension. In contrast to peripheral tissues, where endothelium is highly permeable to electrolytes and oedema formation is more or less proportional to the infused volume of isotonic fluids, electrolytes do not distribute freely through an intact BBB. This is a key mechanism protecting the brain from oedema even when very high amounts of isotonic fluids are administered [11].

Cerebral oedema is stratified depending on location (intracellular or extracellular) and BBB disruption. Cytotoxic oedema is the cellular oedema of neurons or astrocytes and is the result of mainly sodium and water shifts into the cells after an insult with ATP depletion and mitochondrial dysfunction [8, 12]. Vasogenic oedema represents both water and albumin shifts through disrupted endothelial tight junctions. An intermediate type of oedema is ionic oedema, resulting from compensatory solute and water shifts from the vascular compartment to the interstitium through an intact BBB after the formation of cytotoxic oedema has decreased interstitial osmolality.

Autoregulation concerns the capacity of the blood vessels in the brain to sustain CBF by vasodilation or vasoconstriction over a wide range of systemic blood pressures, and in a more general sense may be regarded as the capacity of brain vessels to regulate blood flow in response to changes in metabolic needs. The connection between volume status and intact autoregulation relates to increased CBF to preserve oxygen delivery in response to fluid loading and decreased haematocrit or to maintaining constant CBF through vasodilation when blood pressure drops due to hypovolemia.

Perfusion pressure determinants are both upstream and downstream pressures, with upstream pressures being arterial and downstream pressures being venous. Both lower arterial pressures and higher venous pressures will theoretically result in lower perfusion pressures, albeit with different consequences (i.e. low flow versus tissue oedema) [13]. Increased central venous pressure (CVP) may impede venous outflow from the brain and contribute to increased intracranial pressure (ICP) or cerebral oedema. However, increased CVP will in principle not be transferred to the intracranial compartment so long as intracranial venous structures are collapsed under the influence of ICP before exiting the cranium, and ICP cannot be affected by the extracranial CVP that is generally much lower than the ICP (waterfall effect) [14]. Consequently, venous pressure transferral back to the intracranial contents is possible when ICP is low compared with either CVP or positive-end expiratory pressure (PEEP) in mechanically ventilated patients with brain trauma [15, 16], or when several adverse circumstances act simultaneously to antagonize brain compliance (e.g. hypotonic fluid loading, high CVD, recent brain injury with oedema) as has been shown in animal experiments, but investigations have yielded contradictory results [17, 18]. Although high PEEP may influence ICP on the ‘venous side’ via pressure back-transferral, it may also and independently influence ICP on the ‘arterial side’ depending on whether autoregulation is intact (e.g. when intact, PEEP impedes venous return, resulting in arterial hypotension with cerebral vasodilation and ICP surges) [16].

Contemporary recommendations for routine fluid and intravascular volume management are available from several guidelines and consensus conferences [6, 7, 19‐21]. The 2007 Brain Trauma Foundation guidelines [22] do not provide specific recommendations on fluid management reflecting the pressure-oriented approach. The guideline and consensus recommendations are presented in Table 1. In SAH, euvolemia is recommended to prevent delayed cerebral ischaemia (DCI), routine hypervolemia is not recommended and hypotonic fluids and volume contraction are to be avoided. Furthermore, haemodynamic monitoring to guide fluid management is not advised routinely. Vigilant fluid balance assessment is advised to guide fluid administration but aggressive fluid administration aimed at hypervolemia is considered harmful. The consensus statement on multimodality monitoring in neurocritical care [19] recommends haemodynamic monitoring in patients with haemodynamic instability. Guidelines on ischaemic stroke highlight the importance of isotonic rather than hypotonic fluids and avoidance of hypovolemia and dextrose solutions [20, 21].

The current guidelines on fluid management in brain injury recommend using fluid balances to guide volume status (Table 1). A non-systematic overview of pertinent contemporary studies in brain-injured patients is provided in Additional file 1 [3, 23‐45]. Not all of the reports in this overview studied fluid balance or fluid intake as the primary aim, but because fluid amounts were clearly reported some relevant information could be extracted.

The mean fluid intake was around 3–4 L/day in SAH patients who were treated with normovolemia or received fluid management based on volumetric haemodynamic monitoring versus 4–5 L/day in patients managed with hypervolemic treatment which often included CVP or pulmonary artery occlusion pressure (PAOP)-directed management. Fluid balances generally did not differ between both treatment groups and varied around neutral balance (−0.5 to +1 L) even in a study where mean daily fluid intake was >8 L [28]. Only one study [30] included weight-normalized fluid intake (ml/kg/day). Positive fluid balances have been associated with (angiographic) vasospasm, longer hospital length of stay and poor functional outcomes [27, 37] (see Additional file 1). Higher fluid intake has been associated with more cardiovascular side effects and DCI/delayed ischaemic neurologic deficit (DIND)/infarctions [25, 27, 28, 30, 31, 34, 35]. One may argue that the adverse prognostic value of aggressive fluid loading may reflect more intense treatments in more severely affected patients rather than causal associations because many of these studies are observational cohort studies undoubtedly prone to confounding.

In the trial on prophylactic hypervolemia after aneurysm clipping after SAH by Lennihan et al. [46] the hypervolemic group had a mean fluid intake of up to 4.5 L/day versus around 3.7 L/day in the normovolemia group, with similar daily net fluid balances in both groups (between +0.7 and −0.7 L/day). Hypervolemia did not confer any benefit with regard to CBF or clinical outcomes. The trial by Egge et al. [47] randomized SAH patients between prophylactic hypertensive hypervolemic haemodilution (triple-H) and normovolemia, and reported fluid intake of approximately 3 L/day in the normovolemic group versus 4–5 L/day in the triple-H group (no exact data were provided in the publication). There were no differences in clinical endpoints, but more complications with triple-H (extradural haematoma, haemorrhagic diathesis, congestive heart failure and arrhythmia). For fluid balances (in contrast to fluid intake) such a trend for DCI/DIND/vasospasm was less clear, although two studies reported more adverse outcomes (not restricted to DCI) associated with positive versus negative fluid balances. Data from three other RCTs (of which two were by the same group) [25, 34, 35], a propensity matched analysis on prospective data from a RCT in SAH patients [31] and a RCT on echocardiography-guided fluid resuscitation in trauma patients [43] corroborated the association between more aggressive fluid loading and adverse outcomes (DCI/DIND, cardiovascular side effects, pulmonary oedema, functional outcome and mortality) in both SAH and TBI patients. In addition, a population-based study (n = 5400) reported a temporal association between increased fluid intake and mortality when administered in the pre-DCI period in SAH patients (days 1–3 after the bleed), although it seemed to be beneficial in the DCI risk period (days 4–14) [30]. The data from the RCTs, the propensity matched analysis and the population-based study suggest that there may indeed be a causal link between aggressive fluid loading beyond euvolemia and adverse neurological outcomes, since major confounding is much less likely in these studies. However, tailoring treatment in individual patients remains important, which is exemplified by an investigation in SAH patients showing that increased fluid intake was associated with DIND whereas net negative fluid balances seemed harmful, but only in patients with severe vasospasm [31]. In line with this study and the fact that frank hypovolemia is to be avoided in brain-injured patients, a study in TBI patients found an association of negative fluid balances (< −594 ml) with poor outcome [42]. The ICP and CPP values did not differ between outcome groups, which may indicate that fluid management might impact on outcomes despite successful pressure-targeted management in TBI [42]. Studies showing harm from more positive fluid balances and higher fluid intake and studies specifically targeting fluid management with isotonic fluids are scarce in TBI compared with SAH [42, 45, 48].

A recent review summarized current knowledge on risks and benefits of different types of fluids that are used in traumatic brain injuries [49], and therefore this will not be dealt with in depth here. Some key points regarding the fluid compounds in brain-injured patients are as follows: (1) isotonic fluids are the mainstay of maintenance fluid therapy [50]; (2) synthetic colloids may be harmful after SAH [31, 51] and have not been thoroughly investigated in TBI; (3) contrasting evidence on albumin exists in TBI—its use has been associated with both harm (SAFE study [52]) and benefit [53], but consensus exists that it should generally not be used in TBI and in SAH there is currently insufficient evidence on definite benefit from albumin [54]; (4) in SAH, standard fluid management with saline may have alternatives with more balanced solutions resulting in more stabile electrolytes, less fluid intake and less activation of the pituitary axis stress hormones (cortisol, TSH) [55]; and (5) sodium lactate may hold promise as an alternative fluid to saline solutions in routine fluid management in severe TBI because a recent pilot RCT showed improved ICP control, better electrolyte profile and decreased fluid intake, and its use may have interesting metabolic benefits for the injured brain and its susceptibility to secondary injuries [40]. Of note with regard to the SAFE study, equipoise exists regarding whether the adverse effects of albumin on ICP were related to the relative hypotonicity of the 5 % solution or leakage of albumin through a disrupted BBB creating oncotic shifts that promote oedema [56].

A comprehensive literature search by delegates from a 2010 SAH consensus conference that selected studies on clinical monitoring and volume status (n = 16) highlighted several important findings [57]. First, bedside assessment of volume status is not accurate because sensitivity and positive predictive values for hypovolemia and hypervolemia were less than or equal to 0.37 and 0.06 respectively. These data seem to call into question the effectiveness of vigilant fluid balance management in establishing euvolemia. Second, blood volume measurements to guide fluid management seem feasible and may contribute to the prevention of hypovolemia, but these results are from a small study and blood volume measurements are not widely available. Third, transpulmonary thermodilution (TPT) techniques seem feasible to guide fluid management after SAH. The concluding remarks of this literature search focused on fluid ‘imbalance’, but stressed hypovolemia as a more stringent problem after SAH than hypervolemia. A recent systematic review on advanced hemodynamic monitoring in brain-injured patients (SAH, cardiac arrest, TBI, stroke [58]) showed that such monitoring is widely applied using many different protocols based on local experience. Many other—sometime contradictory—associations between haemodynamic parameters and clinically relevant outcomes were found, but the authors concluded that more research is necessary. The publication showed that the relation between systemic haemodynamics and cerebral perfusion and oxygenation was scarcely studied [58].

In line with the consensus statement on multimodality monitoring in neurocritical care (Table 1 [19]) the goal of fluid management is optimization of cerebral perfusion and oxygenation and minimizing secondary brain insults. Importantly, adequate fluid management in brain injury should preferably be guided by some measure of brain function as a reflection of adequacy of cerebral perfusion and oxygenation, since these are the actual endpoints of fluid titration.

There is broad consensus that hypovolemia should generally be avoided in acute brain injuries. Hypovolemia in this context may be defined as an intravascular volume that is insufficient to sustain minimally adequate cerebral perfusion and oxygenation. Euvolemia may be defined as an intravascular volume that sustains the required cerebral perfusion for adequate brain oxygenation. Defining ‘hypervolemia’ in brain injury is less straightforward. Of note, the distinctive feature of hypervolemia versus hypovolemia or euvolemia is the fact that it concerns what is outside the circulation (i.e. the extravascular space), which makes its assessment and definition much more difficult. For comparison, clinical examples outside neurocritical care are oliguria in fluid-overloaded septic and decompensated heart failure patients representing venous congestion [66]. Obviously, these situations with oliguria do not require fluid loading, since venous congestion will then increase and the ‘congestive kidney failure’ worsen. An increase in CVP will promote tissue oedema, resulting in a dilution of the capillaries and increased tissue diffusion distances for oxygen to the cells. This definition of hypervolemia derived from systemic circulation conflicts with the general use of ‘hypervolemia’ within the older SAH literature, since this designation has been associated with potential benefit for ‘clinical vasospasm’ (DCI) in some classic studies that assumed beneficial effects of ‘hypervolemia’ on blood rheology and prevention of hypovolemia [67, 68]. Further, because definitions of ‘hypervolemia’ as a therapeutic strategy have not been uniform in previous studies, comparability of these studies is hampered [69].

A practical approach to fluid management in brain-injured patients may include: maintenance fluid volumes routinely administered, the type(s) of fluids allowed and their tonicity; and triggers for more advanced haemodynamic monitoring. Monitoring may include invasive methods (e.g. TPT-guided) or less invasive methods (e.g. oesophageal Doppler) [65]. Further, fluid management based on fluid responsiveness [70], other dynamic hemodynamic measures (e.g. pulse pressure variation) or volumetric measures of preload (e.g. GEDI) [25] may be favoured over filling-pressure measures such as PAOP [71].

An algorithm has been used with success by the author in critically ill SAH patients to significantly reduce fluid intake whilst maintaining sufficient cardiac output and indices of cardiac preload (Fig. 2). This algorithm serves as an example of how the basic tenets already described may be materialized and made practical. Maintenance fluids should generally be aimed at 30–40 ml/kg/day of isotonic crystalloids (normal saline 0.9 %), with SAH patients generally needing around 40 ml/kg/day due to higher tendency of polyuria compared with most other brain-injured patients. Triggers for application of haemodynamic monitoring with TPT have been defined in the algorithm, including subsequent haemodynamic goals and ‘stopping rules’. Because the target organ concerns the brain, consciousness assessed with the Glasgow Coma Scale (GCS) is included in the algorithm assuming that a perfectly awake patient will constitute a patient with adequate CBF. The protocol is usually adhered to for up to 5 days. Related co-morbidities and circumstances that are quite frequent in brain-injured patients (diabetes insipidus, cerebral salt wasting, osmotic therapies for increased ICP) are not within the scope of this review and the reader is referred to existing literature [50, 72].