Continuing Medical Education Article

Hypertonic saline (7.2%) in 6% hydroxyethyl starch reduces intracranial pressure and improves hemodynamics in a placebo-controlled study involving stable patients with subarachnoid hemorrhage* Gunnar Bentsen, MD; Harald Breivik, MD, DMSc, FRCA; Tryggve Lundar, MD, DMSc; Audun Stubhaug, MD, DMSc

LEARNING OBJECTIVES On completion of this article, the reader should be able to: 1. Describe the effects of hypertonic saline (7.2% saline in 6% hydroxyethyl starch 200/0.5) on intracranial pressure. 2. Compare the effects of hypertonic with normal saline on cerebral perfusion pressure. 3. Use this information in a clinical setting. All authors have disclosed that they have no financial relationships with or interests in any commercial companies pertaining to this educational activity. Lippincott CME Institute, Inc., has identified and resolved all faculty conflicts of interest regarding this educational activity. Visit the Critical Care Medicine Web site (www.ccmjournal.org) for information on obtaining continuing medical education credit.

Objective: To compare the effects of a bolus infusion of hyper- was ؊5.6 (range, ؊0.8 to ؊12.2) mm Hg after 64 (range, 40 to tonic saline hydroxyethyl starch with the effects of normal saline 115) mins. Mean difference in cerebral perfusion pressure change placebo) on intracranial pressure (ICP) and cerebral perfusion pres- between the groups (HSS ؊ normal saline) was 5.4 mm Hg (95%) and mean difference in ,(002. ؍ sure in patients with spontaneous subarachnoid hemorrhage. confidence interval, 2.2 to 8.6; p Design and Setting: Prospective, randomized, single-blinded, cardiac index change, measured as the area under the curve for placebo-controlled study in a university hospital. the whole study period, corresponded to 0.2 L·min؊1·m؊2 (95% .(025. ؍ Patients: A total of 22 mechanically ventilated patients with confidence interval, 0.03 to 0.4; p spontaneous subarachnoid hemorrhage with stable ICP between Conclusions: In this placebo-controlled study involving spon- 10 and 20 mm Hg. taneous subarachnoid hemorrhage patients with normal to mod- Interventions: During the course of 30 mins, 2 mL/kg of either erately increased ICP, 2 mL/kg HSS reduced ICP and increased 7.2% saline in 6% hydroxyethyl starch 200/0.5 (HSS) or of normal cerebral perfusion pressure significantly. Maximum effect was saline was infused. The effects were observed for another 180 mins. reached at twice the infusion time of 30 mins. There were also Measurements and Main Results: Mean change in ICP after intervention (⌬ICP) calculated from the average of all observa- beneficial hemodynamic effects with increased cardiac index in the HSS group. (Crit Care Med 2006; 34:2912–2917) (tions was ؊3.3 (SD 2.6) mm Hg in the HSS group vs. ؊0.3 (SD 1.3 mm Hg in the normal saline group. Mean difference between the KEY WORDS: brain edema; hypertonic solutions; intracranial hy- -groups (HSS ؊ normal saline) was ؊3.0 mm Hg (95% confidence pertension; intracranial pressure; saline solution; hypertonic; sub Mean peak change after HSS arachnoid hemorrhage .(004. ؍ interval, ؊4.9 to ؊1.1; p

ypertonic saline solutions 5, 6), and rebound increase in intracra- One of the limitations of these studies are an alternative to manni- nial pressure (ICP) (2, 7–11). Hypertonic has been the lack of placebo control, tol in the treatment of intra- saline has attenuated intracranial hyper- which makes it impossible to prove how cranial hypertension (1, 2). tension in a number of clinical trials (12– much of the measured effect can be at- RepeatedH administrations of mannitol are 17). In studies comparing hypertonic sa- tributed to the hypertonic saline inter- associated with adverse effects such as line with mannitol, the results generally vention. The administration of osmoth- acute renal failure (3–5), hypovolemia (3, favor hypertonic saline (18–22). erapy is often one of several interventions applied almost simultaneously to unsta- ble patients. The fact that ICP is danger- *See also p. 3037. Copyright © 2006 by the Society of Critical Care ously high at the time of intervention Consultant (GB), Professor (HB), Associate Profes- Medicine and Lippincott Williams & Wilkins makes the use of a placebo unethical. We sor (AS), Department of Anesthesiology and Intensive DOI: 10.1097/01.CCM.0000245665.46789.7C Care, Professor (TL), Department of Neurosurgery, Rik- have previously, in a prospective observa- shospitalet-Radiumhospitalet Medical Center, Faculty tional study of spontaneous subarachnoid of Medicine, University of Oslo, Oslo, Norway. hemorrhage (SAH) patients with ICP of

2912 Crit Care Med 2006 Vol. 34, No. 12 Ͼ20 mm Hg, demonstrated an ICP re- 200/0.5 solution (HyperHAES, Fresenius Kabi four and six. The examiners were unaware of duction from a mean of 25 mm Hg to a AG). The observation period lasted from 10 the randomization method and unaware of the mean of 11 mm Hg after infusion of a mins before until 210 mins after the start of block size. Group allocations for all patients hypertonic saline solution (23). We could the infusion. Need for rescue treatment was were stored in sealed envelopes marked with also report an increase in cerebral perfu- defined by treatment failure limits for ICP and consecutive patient numbers only. To avoid CPP, which were an ICP of Ͼ20 mm Hg and a examiner bias, the envelopes remained un- sion pressure (CPP). This was done in a CPP of Ͻ60 mm Hg. Unless these limits were opened until inclusion was decided on and controlled intensive care setting. All reached during the observation period, the baseline data logging was established. No pa- other factors that influence ICP and CPP ventilation variables were kept unaltered, the tients were withdrawn after inclusion. The were kept stable during the trial. The aim infusion rates for vasopressors, analgesics, randomization code was not revealed until the of the present study was to validate these sedatives, and fluids were kept stable, the re- study was completed and all data had been findings with an even more laboratory- sistance in the external ventricular drainage entered and validated. like trial in which the patients were stable (EVD) was unchanged, and the patients were Statistics. The mean value of each patient’s before inclusion and hypertonic saline neither stimulated nor moved. measured variables from the 5 mins of regis- was compared with a solution that was Data Acquisition. ICP (intraparenchymal tration before the infusion served as the base- not expected to influence ICP and CPP. device, Codman, Raynham, MA), CPP, heart line. For data acquired electronically every 30 rate, arterial and central venous pressures secs, the mean values of all 5-min periods The present study also addresses the (zeroed at the level of the right atrium of the throughout the 210-min observation period question as to when one can expect to heart), and peripheral oxygen saturation were calculated and used for analysis. As rec- reach the maximum effect of a hypertonic (Siemens AG, Munich, Germany) were regis- ommended for statistical analysis of serial saline infusion. tered electronically every 30 secs (LabView, measurements by Matthews et al. (26), we National Instruments, Austin, TX). The pri- calculated the area under the curve for the PATIENTS AND METHODS mary author manually removed false arterial different variables in each patient and stan- blood pressure values due to blood sampling. dardized by the length of the study, 210 mins, This was a single-center study, performed This typically affected three consecutive re- to get the mean change for the whole period. in the intensive care unit of a university hos- cordings (90 secs). The use of an intraparen- For variables for which time intervals between pital that treats about 170 acute SAH patients chymal device for ICP monitoring is routine successive observations were constant, the annually. The Regional Ethics Committee for practice in our hospital in patients with severe simple mean of the observations was used. To Medical Research and the Norwegian Medi- SAH. This allows continuous ICP and CPP assess differences between the groups, un- cines Agency approved the protocol for the monitoring during EVD, even in situations in paired Student’s t-tests were used, with Welch study. All patients would be unconscious dur- which the EVD catheter is blocked by blood correction when there was unequal variance ing inclusion. A thorough evaluation of the clots. No additional sensors were used for the or Mann-Whitney tests when normality tests failed (GraphPad InStat version 3.05, Graph- risks and possible benefits to the patients was purpose of this study. Pad Software, San Diego, CA). therefore especially important, as underlined Arterial blood gases, pH, hemoglobin, and sodium (ABL 725, Radiometer, Denmark) in the Helsinki Declaration. Our search of the were measured before and 30, 90, and 210 relevant literature indicated minimal risk of RESULTS mins after start of the infusion. Cardiac index, adverse effects. Promising clinical results jus- intrathoracic blood volume index, and ex- tified further clinical trials with this critically From April 2002 through October 2004, travascular lung water index were registered 22 patients with acute, spontaneous SAH ill patient population. Informed consent was before and 30, 90, and 210 mins after start of impossible to obtain. According to Norwegian infusion by use of the PiCCO system (PiCCO, were included, 11 in each group. A total of legislation, one cannot obtain a legally valid Pulsion Medical Systems, Munich, Germany) 21 patients had hemorrhaged because of consent to medical research by proxy. Thus, in (24, 25). The catheters were introduced via the a ruptured aneurysm, and one patient agreement with our ethics committee, close femoral artery and the internal jugular or sub- was diagnosed with a fusiform dilation of relatives were given oral and written informa- clavian vein. the left vertebral artery. Except for this tion about the study when present in the in- Primary Outcome Variables. Primary out- last patient, all patients either were coiled tensive care unit. come variables were: 1) changes in ICP, 2) or clipped (Table 1). All patients were Study Protocol. To perform a study with a difference between the two groups in ⌬ICP mechanically ventilated. They were all placebo group, we could not address the target calculated from the mean of all observations treated with nimodipine and all but one patient population for osmotherapy, namely, in each patient during the observation period, with vasopressors. A majority of the pa- Ͼ and 3) the difference between the groups in those with ICP of 20–25 mm Hg. We there- tients had ongoing EVD (i.e., drainage of fore included intensive care patients with an maximum ⌬ICP. acute, spontaneous SAH, with stable ICP in Secondary Outcome Variables. Secondary cerebrospinal fluid from one or both ce- the range of 10–20 mm Hg. They needed to be outcome variables were: 1) changes in CPP; 2) rebral lateral ventricles). Hemodynamics Ͼ18 yrs of age, sedated, mechanically venti- changes in cardiac output, intrathoracic blood and ICP were stable before inclusion in lated, have stable hemodynamics, and serum volume, and extravascular lung water; and 3) the study. There were no significant de- sodium of Ͻ160 mmol/L. If the inclusion cri- changes in serum sodium levels. mographic differences between the teria were met, the patients either received a Estimate of Number of Patients Needed for groups (Table 1). normal saline solution (NS group) or a hyper- the Study. Study size was calculated based on All patients included were reported ac- tonic saline solution (HSS group) according to difference between the groups in maximal ef- cording to the intention-to-treat princi- a computer-generated, randomized list. As a fect on ICP. We defined a difference of 4 mm ple. No patients were withdrawn after in- Hg as clinically relevant within the current continuous intravenous infusion for 30 mins, clusion. One patient in the NS group 2 mL/kg of the study solution was adminis- ICP range. With a SD of 2, we needed ten tered into a central venous catheter using an patients in each group to achieve a power of needed rescue treatment at 96 mins after infusion pump. The placebo was a 0.9% saline 90% and alpha of 1%. the start of infusion due to increased ICP solution (Fresenius Kabi AG, Bad Homburgh Randomization. A person not otherwise in- and reduced CPP. This patient was then v.d.h., Germany), and the hypertonic solution volved in the study prepared a computer- given 2 mL/kg HSS during the course of was a 7.2% saline in 6% hydroxyethyl starch generated list with alternating block sizes of 20 mins. ICP decreased from 24 mm Hg

Crit Care Med 2006 Vol. 34, No. 12 2913 to 13 mm Hg and remained stable for the Table 1. Demographic data next 3 hrs. CPP increased to above our treatment threshold of 60 mm Hg. We HSS NS p Value did not strengthen the results by extrap- Age in years, mean (SD) 50.1 (10.5) 55.2 (10.8) .28a olating the ICP and CPP values measured Men/women 3/8 1/10 .58b at 96 mins to the remaining observation SAPS II score, mean (SD) 40.5 (11.1) 47.0 (12.1) .20a period. Instead, baseline values were used Hunt and Hess score, median (range) 5 (3 to 5) 5 (2 to 5) .42c a for the period of 97 to 210 mins in this Days since insult, mean (SD) 3.9 (2.9) 2.6 (1.7) .23 EVD 10/11 9/11 .53b patient. Only 5 of 946 scheduled measure- IVH 8/11 8/11 1.0b ments of the different variables were lack- Number of patients coiledd 7/11 5/11 .67b ing because of technical failure. Missing values were replaced by the mean of the HSS, 7.2% saline in 6% hydroxyethyl starch; NS, normal saline; SAPS, Simplified Acute Physiology preceding and the following measurement. Score; EVD, external ventricular drainage; IVH, intraventricular hemorrhage. a b c d Changes in ICP. The reduction in ICP Unpaired t-test; chi-square test; Mann-Whitney test; remaining patients were clipped, except for from baseline, measured as average ⌬ICP one patient in the HSS group for whom no option was available for securing the source of bleeding. during the study period, was significantly greater in the HSS group compared with Table 2. Physiologic data and differences in physiologic data between the groups the NS group (p ϭ .004) (Table 2). The HSS, Mean (SD) NS, Mean (SD) Mean Differencea same variable’s temporal development is shown in Figure 1. The mean maximal ICP, mm Hg, baseline 15.1 (2.9) 15.5 (2.5) Ϫ0.4 (Ϫ2.8 to 2.0), p ϭ .74 change in ICP from baseline in the HSS ⌬ICP, mm Hg, average Ϫ3.3 (2.6) Ϫ0.3 (1.3) Ϫ3.0 (Ϫ4.9 to Ϫ1.1), p ϭ .004b group was Ϫ5.6 (range, Ϫ0.8 to Ϫ12.2) during study period mm Hg and came after 64 (range, 40 to CPP, mm Hg, baseline 71.6 (7.9) 70.6 (5.4) 1.0 (Ϫ5.0 to 7.0), p ϭ .73 ⌬ ϭ 115) mins. As the curve in Figure 1 CPP, mm Hg, average 5.6 (4.2) 0.2 (2.8) 5.4 (2.2 to 8.6), p .002 during study period shows, this was not a meaningful value to MAP, mm Hg, baseline 87.0 (7.6) 84.5 (2.9) 2.4 (Ϫ3.2 to 8.1), p ϭ .36b calculate for the NS group. Mean ICP for ⌬MAP, mm Hg, average 2.3 (3.9) 0.1 (2.8) 2.2 (Ϫ0.8 to 5.2), p ϭ .15 the last 5 mins of the study were still less during study period than baseline in 8 of 11 patients admin- HR, mm Hg, baseline 71(13) 67 (14) 4 (Ϫ8 to 16), p ϭ .48 istered HSS. The last three had a mean ⌬HR, mm Hg, average 2.0 (3.0) 0.9 (3.4) 1.1 (Ϫ1.8 to 4.0), p ϭ .44 during study period ⌬ICP between 0 and ϩ1 mm Hg com- Ϫ Ϫ CI, L⅐min 1⅐m 2, 4.0 (0.8) 3.7 (0.5) 0.3 (Ϫ0.3 to 0.9), p ϭ .36 pared with baseline. baseline Changes in CPP. The average ⌬CPP ⌬CI, L⅐minϪ1⅐mϪ2, AUC 0.27 (0.23) 0.06 (0.18) 0.21 (0.03 to 0.39), p ϭ .025 during the study period was significantly standardized by time higher in the HSS group (p ϭ .002) (Ta- ITBI, mL/m2, baseline 883 (192) 932 (174) Ϫ49 (Ϫ212 to 114), p ϭ .54 2 b ble 2, Fig. 2). The average ⌬CPP after ⌬ITBI, mL/m , AUC 17.3 (38.3) 14.2 (78.3) 3.1 (Ϫ53.3 to 59.5), p ϭ .91 HSS was 5.6 mm Hg higher than base- standardized by time ELWI, mL/kg, baseline 8.0 (2.3) 9.2 (3.5) Ϫ1.2 (Ϫ3.8 to 1.4), p ϭ .36 line, whereas ICP decreased 3.3 mm Hg. ⌬ELWI, mL/kg, AUC Ϫ0.03 (0.4) 0.5 (1.0) Ϫ0.6 (Ϫ1.3 to 0.2), p ϭ .12b This indicates that HSS has a separate standardized by time effect on blood pressure. Average change in mean arterial pressure was 2.3 mm Hg HSS, 7.2% saline in 6% hydroxyethyl starch; NS, normal saline; ICP, intracranial pressure; CPP, after HSS, but this was not statistically cerebral perfusion pressure; MAP; mean arterial pressure, HR, heart rate; CI, cardiac index; ITBI, intrathoracic blood volume index; ELWI, extravascular lung H2O index. different from the change in mean arte- a b rial pressure in the NS group. There was Results from unpaired t-test, mean (HSS - NS) difference (95% confidence interval); p value with Welch correction. also no significant difference in heart rate (Table 2). Changes in Cardiac Index, Intratho- at the 30-min recording. At that point, the study. In the HSS group, a maximum racic Blood Volume Index, and Extravas- change in cardiac index was 0.5 (range, serum sodium increase of 5.6 (range, 4 to cular Lung Water Index. These data were Ϫ0.1 to 1.5) L·minϪ1·mϪ2 in the HSS 7) mmol/L was measured at 30 mins (i.e., at collected at baseline and at 30, 90, and group. Intrathoracic blood volume index the end of the study’s drug infusion). At the 210 mins after start of the infusion. There and extravascular lung water index were conclusion of the observation period, the were no significant differences between not significantly different between groups level was 3.3 (range, 1 to 5) mmol/L greater the groups at baseline (Table 2). Cardiac at baseline, nor were there significant dif- than baseline. Both the peak value and the index increased in the HSS group. Mean ferences in the area under the curve values end value increased significantly compared change in cardiac index compared with between the groups (Table 2). with baseline (p Ͻ .0001). The highest os- baseline, measured as area under the Changes in Serum Sodium. At base- molality measured was 332 mOsm/kg. curve divided by 210 mins, was 0.2 line, mean serum sodium in the HSS group Changes in EVD. As shown in Table 1, L·minϪ1·mϪ2 higher than in the NS was 148 (range, 139 to 157) mmol/L vs. 145 all but three patients had ongoing EVD. group (Table 2). The difference is statisti- (range, 141 to 151) mmol/L in the NS The amounts drained during the hour cally significant. Given the small number of group. The mean difference was 3.0 before the study showed no difference measurements taken, we can draw no con- mmol/L (95% confidence interval, Ϫ1.2 to between the groups: mean of 12.5 (range, clusion about time of maximum effect on 7.2 mmol/L; p ϭ .15). There were no sig- 5 to 28) mL in the HSS group vs. 12.2 cardiac index, but we measured peak effect nificant changes in the NS group during (range, 2 to 25) mL in the NS group, with

2914 Crit Care Med 2006 Vol. 34, No. 12 strates that the patients included were stable and the study design was valid, allowing us to conclude that the statisti- cally significant effects measured for ICP, CPP, and cardiac index in the HSS group can be attributed to the infusion of the hypertonic saline solution. The study design implies that the pa- tient population has normal to only mod- erately increased ICP. This was necessary to conduct a placebo-controlled trial. In a clinical setting, these patients would not be candidates for osmotherapy. The ques- tion is whether our results can be extrap- olated to SAH patients with severe intra- cranial hypertension. We think they can. Figure 1. Changes in intracranial pressure (ICP). Each time point represents the mean of the The results are in accordance with the registrations of the preceding 5 mins; error bars display the 95% confidence interval. NS, normal results from our previously published, saline group; HSS, 7.2% saline in 6% hydroxyethyl starch group. uncontrolled study that was conducted in very much the same manner (23). That study had an equally long observation period of 210 mins and used the same inclusion criteria, except that the pa- tients were to have ICP of Ͼ20 mm Hg. The change in ICP with time is very much alike in the two studies. The abso- lute effect on ICP after HSS was less in the current study, with a mean maximum ICP reduction of 5.6 mm Hg, compared with our previous study, which had a mean reduction of 14.3 mm Hg. This is most likely related to the fact that base- line ICP was lower in the current study. With ICP in the range of 10 to 20 mm Hg we are not on the steep part of the intra- Figure 2. Changes in cerebral perfusion pressure (CPP). Each time point represents the mean of the cranial pressure-volume curve. Equal registrations of the preceding 5 mins; error bars display the 95% confidence interval. NS, normal changes in volume will have less effect on saline group; HSS, 7.2% saline in 6% hydroxyethyl starch group. pressure compared with patients having ICP of Ͼ20 mm Hg. The current placebo- controlled study clearly describes the ef- a mean difference 0.3 mL (95% confi- significantly less than 10 mm Hg, per- fects of HSS. It greatly strengthens our dence interval, Ϫ6.6 to 7.2 mL; p ϭ .93). haps close to zero. If so, this patient previous findings obtained with a rele- The volume was significantly less during should not have been included. Still, the vant patient population. The magnitude the first hour of the study in the HSS results are included according to the in- of the expected effects in clinical practice group, with a mean difference compared tention-to-treat principle. If excluded for would be in the range documented in with the previous hour of Ϫ8.2 mL (95% a “per protocol” analyses, this would en- that study (23). confidence interval, Ϫ13.9 to Ϫ2.5 mL; p ϭ hance statistical significance because this Maximum effect was reached at a .008). This reduction was also signifi- patient represents a nonresponder in the mean of 64 mins in the current study cantly greater than in the NS group, with HSS group. when 2 mL/kg hypertonic solution was a mean difference (HSS Ϫ NS) of Ϫ6.3 infused in 30 mins. In the previous study, Ϫ mL (95% confidence interval, 11.5 to DISCUSSION the infusion was given in 20 mins. Peak Ϫ1.1 mL; p ϭ .02). effect was reached at a mean of 40 mins. Protocol Violation. The fifth patient in Comparing infusion of 2 mL/kg 7.2% In a recent study involving patients with the HSS group had a baseline ICP of 16 saline in 6% hydroxyethyl starch (HSS) traumatic brain injury by Battison et al. mm Hg when included and ICP remained with NS in a laboratory-like model, this (21), 100 mL of 7.5% saline/6% dextran at Ϯ1 mm Hg throughout the observa- study has confirmed the ability of HSS to was infused in 5 mins in comparison with tion period. Two days later, it was discov- reduce ICP and increase CPP in SAH pa- mannitol. The exact time of maximum ered that the zero level of the measuring tients. All the physiologic variables mea- ICP reduction was not stated, but from device had drifted. The correct ICP was sured in the NS group fluctuated around the enclosed time-series plot, it seems 15 mm Hg lower than displayed. We baseline values throughout the 210-min that maximum effect was typically therefore have reason to suspect that this study period, except for the one patient reached between 20 and 30 mins. The patient’s ICP at the time of inclusion was who needed rescue therapy. This demon- majority of the effect is thus not instan-

Crit Care Med 2006 Vol. 34, No. 12 2915 taneous, like the effect of manipulating (33) compared a bolus of 3% saline with a patients with limited ischemic damage

PaCO2. This corresponds to the idea that placebo, demonstrating lower ICP in the have relatively limited disruption of the the osmotic effect is a major contributor hypertonic saline group. Simma et al. blood–brain barrier and might therefore to the overall effect (27, 28). (34) compared 1.7% saline with Ringer be well suited for osmotherapy compared The fact that 10 of 11 patients receiv- lactate infusion during the course of 72 with patients having widespread disrup- ing HSS had ongoing EVD, most likely hrs, targeting a serum sodium level of tion of the blood–brain barrier. The prob- reduced the magnitude of ICP change. 145 to 150 mmol/L in the saline group. lems with vasospasm after SAH can The HSS infusion was accompanied by a They found lower ICP, fewer complica- strengthen the indication for hypertonic reduction of the EVD volume, which was tions, and need for fewer interventions in saline. Suarez et al. (36) have published a probably primarily caused by the ICP re- the hypertonic saline group. The study by retrospective study on hyponatremic pa- duction per se. There may also be other Simma et al. (34) utilizes a model very tients with vasospasm after SAH. Their explanations. There is documentation different from our study, whereas the focus was not on intracranial hyperten- that mannitol decreases the production model used by Fisher et al. (33) is not so sion, and the authors only concluded on of cerebrospinal fluid (29, 30), but we different. A bolus was given, and the ef- the safety of hypertonic saline infusion. have found no publications directly ad- fect was observed for 2 hrs. They reported The question as to whether bolus doses or dressing the effect of hypertonic saline on a significant difference between average continuous infusion represent the better cerebrospinal fluid formation. Indirectly, ICP during the observation period of 15.8 strategy remains unanswered. however, HSS might reduce cerebrospi- mm Hg and an ICP baseline of 19.9 mm Hypovolemia is among the possible nal fluid formation via the increase in Hg in the hypertonic saline group. This adverse effects related to mannitol. It is serum osmolality. This has been demon- corresponds to our finding of ⌬ICP of therefore reassuring to see that hemody- strated with glycine (31). Ϫ3.3 mm Hg in the HSS group; the dif- namic stability is well maintained with Was the choice and volume of study ferences being, however, lower baseline HSS. Intrathoracic blood volume is solutions adequate? As the placebo, we ICP, lower osmotic load, and longer infu- maintained and cardiac index is improved chose normal saline. This, of course, is a sion time in our study. In addition, the in the HSS group. Based on documenta- slightly hypertonic solution with a so- differences between children and adults tion from small-volume hypertonic re- dium content of 154 mmol/L. Given nor- and traumatic brain injury vs. SAH are suscitation research, one can expect this mal serum sodium levels of 140 mmol/L, important. Our study offers a description hemodynamic effect to be longer lasting this would weaken its perceived “placebo” of how ICP and CPP change with time, after an infusion of hypertonic saline in effect. We reasoned that this was more which is lacking in previous studies, and hetastarch or dextran compared with hy- acceptable than strengthening the results our study contains more detailed infor- pertonic saline alone (32). This is benefi- by using a slightly hypotonic solution mation on hemodynamics. cial in patients needing vasopressor sup- such as lactated Ringer. As it turned out, Baseline serum sodium was increased port to maintain a sufficient CPP. From both groups had increased sodium levels in both groups. This was most probably the fact that there were no changes in of 148 and 145 mmol/L. This made nor- caused by our use of normal saline as a intrathoracic blood volume index or ex- mal saline more suited as the placebo standard solution in patients with intra- travascular lung water index, we can also than we had foreseen. Saline solutions of cranial diseases in the days before inclu- conclude that there were no signs of ove- 7.2% to 7.5% in hetastarch or dextran sion in this study. The patients were not rhydration or extravasation of water in have been extensively studied in trials on subject to volume restriction. This is sup- the lungs. small-volume resuscitation (32). The ported by baseline intrathoracic blood We have not recorded any adverse ef- usual dose has been 4 mL/kg. This dose volume index in the normal range (Table fects in the current study. In particular, can be regarded as safe. From previous 2). The changes in serum sodium show we saw no indications of an increase in experience, we expected 2 mL/kg to be an inverse pattern to the changes in ICP. ICP needing intervention at the conclu- sufficient for a reliable effect on ICP with- We found this both in the current and in sion of the observation period. However, out increasing serum sodium more than the previous trial. After an initial increase it is important to recognize that issues 5 to 6 mmol/L. We would also argue that caused by HSS infusion, the sodium level concerning safety can only be studied to a inter-individual differences were reduced decreases with time toward baseline. The very modest degree in a small, short- by using a volume based on patient clinical implication of this is that it period study like the present. weight rather than a fixed volume. For makes sense to continue with a slow in- comparison, it was important to use the fusion of HSS after the initial bolus to CONCLUSIONS same dose as used in our previous study maintain the increased serum sodium (23). Based on the results from that level. Doing this, one could expect to A volume of 2 mL/kg 7.2% saline in study, we would now recommend that maintain the desired effect on ICP for a 6% hydroxyethyl starch 200/0.5 (HSS) re- the initial bolus be Ͻ2 mL/kg 7.2% HSS longer period. Indeed, promising results duced ICP and increased CPP signifi- to not create larger fluctuations in ICP have been published concerning the use cantly in SAH patients with normal to than needed or to exhaust too rapidly the of continuous infusion of hypertonic sa- moderately increased ICP, whereas there therapeutic possibilities of hypertonic sa- line in pediatric patients with traumatic were no changes in the placebo group. line. brain injury (16, 34). On the other hand, The effects of HSS reached their maxi- We have found only two other studies others have found unfavorable results in mum at twice the infusion time of 30 comparing hypertonic saline with a nor- adult head trauma patients given prophy- mins. There were also beneficial hemody- mal saline placebo in patients with in- lactic infusions of hypertonic saline (35). namic effects with increased cardiac in- creased ICP. Both were done in children The results are conflicting, and none of dex in the HSS group. All changes can be with traumatic brain injury. Fisher et al. these trials included SAH patients. SAH attributed to HSS because the patients

2916 Crit Care Med 2006 Vol. 34, No. 12 were otherwise stable and there were no 14. Horn P, Munch E, Vajkoczy P, et al: Hyper- dicator of cardiac preload: Single transpulmo- changes in the placebo group. tonic saline solution for control of elevated nary thermodilution technique versus intracranial pressure in patients with ex- assessment of pressure preload parameters de- hausted response to mannitol and barbitu- rived from a pulmonary artery catheter. J Car- REFERENCES rates. Neurol Res 1999; 21:758–764 diothorac Vasc Anesth 2001; 15:584–588 15. Munar F, Ferrer AM, de Nadal M, et al: Ce- 25. Katzenelson R, Perel A, Berkenstadt H, et al: 1. Qureshi AI, Suarez JI: Use of hypertonic sa- rebral hemodynamic effects of 7.2% hyper- Accuracy of transpulmonary thermodilution line solutions in treatment of cerebral edema tonic saline in patients with head injury and versus gravimetric measurement of extravas- and intracranial hypertension. Crit Care Med raised intracranial pressure. J Neurotrauma cular lung water. Crit Care Med 2004; 32: 2000; 28:3301–3313 2000; 17:41–51 1550–1554 2. Doyle JA, Davis DP, Hoyt DB: The use of 16. Khanna S, Davis D, Peterson B, et al: Use of 26. Matthews JN, Altman DG, Campbell MJ, et al: hypertonic saline in the treatment of trau- hypertonic saline in the treatment of severe Analysis of serial measurements in medical matic brain injury. J Trauma 2001; 50: refractory posttraumatic intracranial hyper- research. BMJ 1990; 300:230–235 367–383 tension in pediatric traumatic brain injury. 27. Jafar JJ, Johns LM, Mullan SF: The effect of 3. Worthley LI, Cooper DJ, Jones N: Treatment Crit Care Med 2000; 28:1144–1151 mannitol on cerebral blood flow. J Neuro- of resistant intracranial hypertension with 17. Peterson B, Khanna S, Fisher B, et al: Pro- surg 1986; 64:754–759 hypertonic saline: Report of two cases. J Neu- longed hypernatremia controls elevated in- 28. Ogden AT, Mayer SA, Connolly ES Jr: Hyper- rosurg 1988; 68:478–481 tracranial pressure in head-injured pediatric osmolar agents in neurosurgical practice: 4. Oken DE: Renal and extrarenal consider- patients. Crit Care Med 2000; 28:1136–1143 The evolving role of hypertonic saline. Neu- ations in high-dose mannitol therapy. Ren 18. Schwarz S, Schwab S, Bertram M, et al: Ef- rosurgery 2005; 57:207–215 Fail 1994; 16:147–159 fects of hypertonic saline hydroxyethyl starch 29. Sahar A, Tsipstein E: Effects of mannitol and 5. Bullock R: Mannitol and other diuretics in solution and mannitol in patients with in- furosemide on the rate of formation of cerebro- severe neurotrauma. New Horiz 1995; 3: creased intracranial pressure after stroke. spinal fluid. Exp Neurol 1978; 60:584–591 448–452 Stroke 1998; 29:1550–1555 30. Donato T, Shapira Y, Artru A, et al: Effect of 6. Arai T, Tsukahara I, Nitta K, et al: Effects of 19. De Vivo P, Del Gaudio A, Ciritella P, et al: mannitol on cerebrospinal fluid dynamics mannitol on cerebral circulation after tran- Hypertonic saline solution: A safe alternative and brain tissue edema. Anesth Analg 1994; sient complete cerebral ischemia in dogs. to mannitol 18% in neurosurgery. Minerva 78:58–66 Crit Care Med 1986; 14:634–637 Anestesiol 2001; 67:603–611 31. Zoghbi HY, Okumura S, Laurent JP, et al: 7. Kaufmann AM, Cardoso ER: Aggravation of 20. Vialet R, Albanese J, Thomachot L, et al: Acute effect of glycerol on net cerebrospinal vasogenic cerebral edema by multiple-dose Isovolume hypertonic solutes (sodium chlo- fluid production in dogs. J Neurosurg 1985; mannitol. J Neurosurg 1992; 77:584–589 ride or mannitol) in the treatment of refrac- 63:759–762 8. Garcia-Sola R, Pulido P, Capilla P: The im- tory posttraumatic intracranial hyperten- 32. Rocha-E-Silva, Poli de Figueiredo LF: Small mediate and long-term effects of mannitol sion: 2 mL/kg 7.5% saline is more effective volume hypertonic resuscitation of circula- and glycerol: A comparative experimental than 2 mL/kg 20% mannitol. Crit Care Med study. Acta Neurochir (Wien) 1991; 109: 2003; 31:1683–1687 tory shock. Clinics 2005; 60:159–172 114–121 21. Battison C, Andrews PJ, Graham C, et al: 33. Fisher B, Thomas D, Peterson B: Hypertonic 9. Kofke WA: Mannitol: Potential for rebound Randomized, controlled trial on the effect of saline lowers raised intracranial pressure in intracranial hypertension? J Neurosurg An- a 20% mannitol solution and a 7.5% sa- children after head trauma. J Neurosurg An- esthesiol 1993; 5:1–3 line/6% dextran solution on increased intra- esthesiol 1992; 4:4–10 10. Polderman KH, van de Kraats G, Dixon JM, cranial pressure after brain injury. Crit Care 34. Simma B, Burger R, Falk M, et al: A prospec- et al: Increases in spinal fluid osmolarity Med 2005; 33:196–202 tive, randomized, and controlled study of induced by mannitol. Crit Care Med 2003; 22. Harutjunyan L, Holz C, Rieger A, et al: Effi- fluid management in children with severe 31:584–590 ciency of 7.2% hypertonic saline hydroxy- head injury: Lactated Ringer’s solution ver- 11. Wakai A, Roberts I, Schierhout G: Mannitol ethyl starch 200/0.5 versus mannitol 15% in sus hypertonic saline. Crit Care Med 1998; for acute traumatic brain injury. Cochrane the treatment of increased intracranial pres- 26:1265–1270 Database Syst Rev 2005, 4:CD001049 sure in neurosurgical patients: A randomized 35. Qureshi AI, Suarez JI, Castro A, et al: Use of 12. Hartl R, Ghajar J, Hochleuthner H, et al: clinical trial. Crit Care 2005; 9:R530–R540 hypertonic saline/acetate infusion in treatment Hypertonic/hyperoncotic saline reliably re- 23. Bentsen G, Breivik H, Lundar T, et al: Pre- of cerebral edema in patients with head trau- duces ICP in severely head-injured patients dictable reduction of intracranial hyperten- ma: Experience at a single center. J Trauma with intracranial hypertension. Acta Neuro- sion with hypertonic saline hydroxyethyl 1999; 47:659–665 chir Suppl (Wien) 1997; 70:126–129 starch: A prospective clinical trial in criti- 36. Suarez JI, Qureshi AI, Parekh PD, et al: Ad- 13. Schatzmann C, Heissler HE, Konig K, et al: cally ill patients with subarachnoid haemor- ministration of hypertonic (3%) sodium Treatment of elevated intracranial pressure rhage. Acta Anaesthesiol Scand 2004; 48: chloride/acetate in hyponatremic patients by infusions of 10% saline in severely head 1089–1095 with symptomatic vasospasm following sub- injured patients. Acta Neurochir Suppl 24. Wiesenack C, Prasser C, Keyl C, et al: Assess- arachnoid hemorrhage. J Neurosurg Anes- (Wien) 1998; 71:31–33 ment of intrathoracic blood volume as an in- thesiol 1999; 11:178–184

Crit Care Med 2006 Vol. 34, No. 12 2917