We studied fluid requirements and oedema formation in survivors of OHCA in a prospective, randomised design. The HH patients received significantly less fluid than the control patients (4750 ml vs. 8010 ml, p = 0.019). Both groups had a significant drop in SVR, and demonstrated increased extravasation through the drop in COP. The extravasation did not show as vasogenic brain oedema.
The strength of the study lies in its design and the multiple determination of leakage. The weakness of our design is that the treating physicians were not blinded. This could have caused a tendency to replace fluid with vasopressors. However, there were no differences between the groups regarding the use of these drugs. Furthermore, as sedation can cause vasodilatation, the use of sedation may influence the use of fluid and vasopressors. The lowest doses of sedation were used in all patients to achieve MAAS 0-1. The number of patients in our study is not sufficient to determine whether fluid load can affect neurological outcome/survival.
A large cohort study recently reported on the challenging aspects of therapeutic hypothermia . In spite of a positive fluid balance, many patients appear to be hypovolemic and have high fluid requirements . Our reported fluid balance is slightly higher than the balance reported by Sunde and colleagues, who found a positive balance of 3455 ml (1594) during 24 hours with similar treatment goals . Laurent and colleagues used 3500-6500 ml during the first 24 hours to maintain an adequate filling pressure in normothermic cardiac arrest patients . Whether reduced fluid load is of benefit to these patients remains unknown.
To our knowledge, there are no papers describing repetitive MRI in the initial treatment of OHCA patients. Järnum and colleagues performed MRI on 20 cardiac arrest patients who remained unconscious 72 hours after normothermia . They found hypoxic-ischemic cerebral oedema in two patients during neuropathological examination post mortem. None of the patients in our study had a vasogenic cerebral oedema on the MRI, which indicated an intact blood-brain barrier. Animal studies have shown that asphyxia is more likely to cause a disrupted blood-brain barrier [24–26]. The lack of vasogenic oedema may be the result of cardiac origin of the arrest, the fact that arrests were witnessed and short time before initiation of CPR.
We found reduced fluid leakage to the interstitial space in the HH patients compared with controls. Maintenance of intravascular COP is one important factor in determining fluid flux across the capillary membrane. The decline in COP in plasma was probably due to hemodilution, which is also reflected in a reduction in haemoglobin and erythrocyte volume fraction.
The reduction in COP in interstitial fluid is probably caused by the escape of fluid with a lower COP through the capillaries. Since we observed a simultaneous reduction in COP both in plasma and interstitial fluid, the increased extravasation cannot be explained by the differences in the COP gradient between the groups. However, the change may be attributed instead to capillary leakage, which has also been demonstrated in several animal studies [5, 27, 28]. This is supported by Nordmark et al., who found a decreased intravascular volume during hypothermia after cardiac arrest . COP is important in capillary fluid exchange, but is a minor component of the total osmotic pressure. The significant difference between the groups regarding serum osmolality may partly explain the observed differences in fluid loads. This emphasises the importance of also taking the total osmotic pressure into consideration when choosing i.v. fluid. Sodium concentration in the HH group differed significantly from the controls after 24 hours and reflected the content of sodium in the HH solution. This may influence fluid shifts and lead to osmotic dehydration, with shrinkage of cells and the prevention of endothelial oedema .
Both groups demonstrated a comparable and significant reduction in SVR, suggesting a similarity between septic and post-cardiac arrest patients . As hypovolemia leads to an increased SVR, our finding may reflect a volume 'overload' . Hypothermia and infusion of vasopressors should induce vasoconstriction and centralise circulation. However, intravenous fluid and inflammation counteract vasoconstriction , and the overall result was a significant decline in SVR in both the study and the control group, also observed by Laurent et al. .
Small volume resuscitation with hypertonic saline during CPR is described as feasible and safe , and, in a study of critically ill ICU patients, HH was infused without negative effects on renal function . The VISEP study  showed impaired renal function in sepsis patients resuscitated with hydroxyethyl starch 200/0.5, and there have been discussions concerning the safety of these solutions in critically ill patients. Two of our patients who received HH developed renal failure, one due to arterial embolism, while the other developed failure weeks later. We consider the kidney failure in these two patients to be unrelated to HH; its contribution cannot be excluded, however.
Despite lower body temperature, CI was higher at 24 hours than on admission to the MICU. The increase was significant in the control group. Laurent and collaborators made the same observation when they monitored more than 160 OHCA patients with pulmonary artery catheter . The improvement in CI in our study represents adequate fluid load and reduced stunning of the heart. In a recent study, Jacobshagen et al. also demonstrated an improved ventricular function over time in patients after cardiac arrest .
The clinical implication of the present study is that post-cardiac arrest patients can be liberally infused with crystalloids during the first 24 hours without cerebral oedema resulting. They also have a high fluid requirement, which is partly because of increased extravasation, measured by means of colloid osmotic pressures, systemic vascular resistance and fluid calculations. Both fluid regimens stabilise hemodynamics. The reduced fluid load achieved by the application of HH should be further investigated in cardiac arrest caused by asphyxia, where a disrupted blood-brain barrier is more likely. The lack of vasogenic brain oedema in these patients is encouraging. This supports a liberal use of crystalloids, especially due to an increased need for intravascular volume and the possible side effects of colloids. Furthermore, the impact on neurological outcome and survival should be examined.