Capillary leakage in post-cardiac arrest survivors during therapeutic hypothermia - a prospective, randomised study

The study was approved by the Regional Committees for Medical Research Ethics, the Data Inspectorate, the Directorate for Health and Social Affairs and the Norwegian Medicines Agency. Deferred consent was used, and the patients' families were entitled to withdraw the patients at any time. All patients included were informed about the study when they were able to receive the information and signed a written informed consent form.

The study was performed on 19 patients with witnessed out-of-hospital cardiac arrest (OHCA) and carried out between September 2005 and March 2007 at Haukeland University Hospital (Bergen, Norway), an 1,100-bed hospital serving 600,000 people. All inclusion/exclusion criteria are presented in Table 1. The fluid intervention was initiated immediately after admission to the emergency room and continued for the first 24 hours.

On admission, the patients were allocated by means of stratified randomisation to one of two fluid regimens administered via infusion pumps: Ringer's Acetate and saline 9 mg/ml (control), or hypertonic colloid,7.2% NaCl with 6% Hydroxyethyl starch 200/0.5 (HyperHAES^® Fresenius Kabi, Germany) (HH). Fluid was administered to achieve the treatment goals listed in Table 2. HH was limited to 500 ml per 24 hours (20 ml/hr). Further needs for fluid in the HH group were met by Ringer's Acetate/saline 9 mg/ml. The control group received Ringer's Acetate and saline 9 mg/ml by turn during the observation period, in accordance with the standard treatment in the medical intensive care unit (MICU).

Patients with ST elevation, a new left bundle branch block or cardiogenic shock were referred immediately for coronary angiography and subsequent percutaneous coronary intervention (PCI).

Before admission to the MICU, after cardiac intervention and if the patient did not have an intra-aortic-balloon pump (IABP), magnetic resonance imaging (MRI) of the brain was planned (1.5 Tesla, conventional morphological and diffusion sequences). Repeated MRI was scheduled after 24 and 96 hours.

Cardiac arrest data were recorded according to the Utstein style [16]. In the MICU, monitoring was performed (IntelliVue, Philips, Eindhoven, the Netherlands) with continuous ECG, arterial pressure and continuous cardiac output registration (PiCCO^®, Pulsion Medical System AG, Germany). Fluid balance was measured as the total amount of fluid administered intravenously and enterally in relation to output measured by hourly diuresis and 24-hour faecal loss. Systemic vascular resistance (SVR) was calculated 0, 8, 16 and 24 hours after admission to the MICU. Vasopressors (dopamine, noradrenaline, and adrenaline) were administered if the mean arterial blood pressure was <60 mmHg and the fluid load proved ineffective, guided by PiCCO measurements. Dopamine was replaced with noradrenaline if tachycardia occurred (>100 beats min^-1) or if dopamine failed to achieve the required blood pressure. Sedatives (midazolam, alfentanil) were administered to achieve a motor activity assessment score of 0 (MAAS). If necessary, vecuronium was administered to prevent shivering. Ventilation was provided by Evita XL (Dräger Medical, Lübeck, Germany), using a bi-positive airway pressure mode. Cooling was initiated outside the hospital for all patients who had return of spontaneous circulation (ROSC) and remained unconscious. At the scene, cooling was performed using icepacks placed on the neck, armpits and groin. A Coolgard catheter (Alsius, California, USA) was installed in the right femoral vein in the PCI lab and activated in the MICU, cooling the patient at a rate of 1°C per hour. The target temperature was set at 33°C and measured in the urine bladder. After 24 hours of cooling, rewarming at a rate of 0.5°C per hour was stopped at 35.0°C.

Blood samples were taken from the artery line after 0, 8, 16 and 24 hours, and analysed at the Laboratory of Clinical Biochemistry at Haukeland University Hospital. Colloid osmotic pressure (COP) was measured at 0, 8, 16 and 24 hours in serum and in interstitial fluid that was sampled using the wick method, installed for 60 minutes [17‐20]. A sterile, multi-filament nylon wick was soaked in Ringer AC. Using a sterile technique and a needle, the wick was placed subcutaneously in the midaxillary line. Three wicks were installed at intervals of 3 cm and covered by plastic film (Tegaderm, 3M Inc., Canada), to prevent evaporation. COP was measured by means of a transducer (Gould-Statham, Spectramed, USA), recorded and amplified with an EasyGraph 240 (Gould Inc., USA).

The randomisation was stratified with respect to initial heart rhythm. Numbered envelopes were distributed from the MICU and opened when the physician in the emergency room enrolled a patient, filling in the inclusion criteria. The allocation was generated by the authors. The sample size was determined by power calculations on the basis of a required volume load of 8000 ml crystalloids during the first 24 hours and a standard deviation of 500 ml. A power of 80% and a significance level of 0.05 for a two sample t-test suggested that it would be sufficient to have three patients in each group if HH reduced the required volume by 30% to 5600 ml. Due to lower power in non-parametric tests, a higher number was chosen. The unconscious patients, as well as the neuroradiologist, were blinded to the treatment. The two treatment groups were descriptively compared at baseline. Fluid load, urine output and fluid balance were compared using an exact Mann-Whitney test. Mixed effects models were used for group comparisons of repeated measurements of variables [21]. Time from baseline was entered as a categorical covariate, as well as any differences in developments in the two groups, and there were assumed to be no group differences at baseline. The nlme package in R (R Foundation for Statistical Computing, Vienna, Austria) was used for linear mixed effects models; SPSS version 15.0 (SPSS Inc., Chicago, IL, USA) was used for other statistical analyses, and SPSS Sample Power for power calculation. Numbers were presented as mean (standard error), or median (low-high). A p-value <0.05 was considered significant. For categorical covariates with more than two categories, both overall p-values for the variable and p-values for individual contrasts are reported.

Twenty-four patients were randomised. Five were excluded due to lack of witnessed arrest, inclusion in another study, probable respiratory cause of the cardiac arrest, and age >80 years (Fig 1). Ten patients (two female) were randomised to HH, and nine (one female) to the control fluid regimen. The initial heart rhythms and baseline characteristics are presented in Table 3. There were no substantial differences between the groups as regards the aetiology of the arrest. The first temperature recorded at the hospital was 34.5 (1.4) °C. Survival after one year was 79%, with no significant difference between the groups (Table 3).

During the first 24 hours in the hospital, the HH group required significantly less fluid than the control group to meet the treatment goals. Fluid calculations are presented in Table 4. The HH group received 6.02 ml/kg (4.63 - 7.69) of HH during the first 24 hours.

COP in plasma showed a significant decline in both groups (Fig. 2a). The reduction was more rapid in the control than in the HH group, but the nadir levels were the same in both groups. The corresponding levels of interstitial COP showed the same pattern (Fig. 2b). The drop in COP was significant at all times, except for the HH group at eight hours. The planned MRI at 0, 24 and 96 hours was performed on seven patients. MRI was performed at 0 and 96 hours on two patients, and on one patient at 96 hours. The ten patients were equally distributed between the two groups. Divergence from the plan was due to technical problems. MRI did not reveal vasogenic cerebral oedema in any of these patients.

SVR dropped significantly in both groups (Fig. 3). The cardiac index (CI) was 2.2 l/min/m² (0.2) on admission to the MICU. At 24 hours, before rewarming, the CI was higher in both groups and significantly higher in the control group (Fig. 4). MAP and CVP did not differ significantly between groups (Fig. 5). There were no differences in dose and type of vasopressors between the groups. All patients needed vasopressors, primarily dopamine in accordance with the MICU guidelines.

All laboratory data are listed in Table 5. Serum osmolality differed significantly, with an increase in the HH group and a decrease in the control group (p < 0.001). Serum sodium and chloride increased in both groups. Two patients who received HH later developed renal failure.