In the June edition of the articles of the month, I included a paper on hypertonic saline for the treatment of traumatic brain injury. My conclusion (and that of the paper’s authors) was that hypertonic saline did not seem to provide any benefit, either in terms of mortality, or even in terms of lowering intracranial pressure. My friend Scott Weingart pointed out that the paper might not actually support that conclusion. The problem was with the studies they included in the review (which I hadn’t read myself). This is probably an excellent lesson: reviews are nice as an introduction to a topic, but expert clinical practice really requires a familiarity with the original literature. For example, there are many reviews that conclude that tPa is excellent for ischemic stroke, but… well I guess I won’t get into that here. Anyhow, I promised to read the studies on hypertonic saline in a little more depth and post an update, so that is what follows.
Before we get into the update, these are my original comments, cut and paste from the June edition:
Berger-Pelleiter E, Émond M, Lauzier F, Shields JF, Turgeon AF. Hypertonic saline in severe traumatic brain injury: a systematic review and meta-analysis of randomized controlled trials. CJEM. 18(2):112-20. 2016. PMID:26988719
I have heard hypertonic saline mentioned as a replacement for mannitol for the treatment of intracranial hypertension at numerous conferences since finishing residency. I was under the impression it was becoming the treatment of choice, but there is a reason we practice evidence based medicine. This is a systematic review and meta-analysis that identified 11 RCTs covering 1820 adult patients with traumatic brain injury comparing hypertonic saline to either mannitol (½ the studies) or another solution (often normal saline, or even hypotonic saline.) Hypertonic saline did not decrease mortality (RR 0.96, 95%CI 0.83-1.11). It didn’t lower intracranial pressure (weighted mean difference -0.39, 95%CI -3.78 – 2.99). And it didn’t improve functional outcomes (RR 1.12, 95% CI 0.92-1.36). Having the same outcomes as mannitol may not be bad, but in ½ these studies hypertonic saline was compared to iso or even hypotonic crystalloids (placebo?) and didn’t perform any better. On the other hand, it doesn’t look any worse than mannitol, so there still may be a role somewhere for it in trauma.
Bottom line: We probably shouldn’t be rushing to change to hypertonic saline in the management of traumatic brain injury.
So this review concludes that hypertonic saline is ineffective, but what does the base literature show? I looked through the references of that review, as well as performing my own literature search (but without the help of a medical librarian) and these were the relevant papers I could find (sorted by date, for lack of a better system):
Shackford SR, Bourguignon PR, Wald SL, Rogers FB, Osler TM, Clark DE. Hypertonic saline resuscitation of patients with head injury: a prospective, randomized clinical trial. The Journal of trauma. 44(1):50-8. 1998. PMID: 9464749
This is a prospective RCT of 34 adult blunt trauma patients requiring ICP monitoring. They compared usual care (½ normal saline as a maintenance fluid and resuscitation of hypotension of low urine output with boluses of Ringer’s lactate) to a hypertonic strategy (normal saline as a maintenance fluid and 1.6% saline for resuscitation). The study was initially designed to have 4 groups, but struggle with recruitment, and so combined injury types together to just compare the different fluid therapies. Unfortunately, less than half of eligible patients were enrolled, for various reasons, potentially leading to selection bias. Also, for some reason the excluded all patient who died from their analysis without any explanation, but I think those patients are very important in trauma trials. Bottom line is that there was no difference in either the ICP or the CPP (cerebral perfusion pressure) between the two groups. The hypertonic saline group had more episodes of elevated ICP that required interventions. The hypertonic saline group started sicked, so their return to baseline could have been a result of the therapy, but more likely represents regression to the mean. This is the only study using 1.6% saline. Most use more hypertonic solutions.
Bottom line: No evidence of difference here between resuscitation with 1.6% saline and Ringer’s lactate.
Vialet R, Albanèse J, Thomachot L. Isovolume hypertonic solutes (sodium chloride or mannitol) in the treatment of refractory posttraumatic intracranial hypertension: 2 mL/kg 7.5% saline is more effective than 2 mL/kg 20% mannitol. Critical care medicine. 31(6):1683-7. 2003. PMID: 12794404
This is a prospective, randomized trial of 20 patients with ICP monitors in the ICU after traumatic brain injury, whose elevated ICP was refractory to standard therapy (which for some patients included drainage of CSF.) They compared 2ml/kg of 7.5% saline or 20% mannitol. 1 of 10 patients in the hypertonic saline group had persistently elevated ICP as compared to 7/10 in the mannitol group (p=0.014). In terms of patient important outcomes (death and 90 day neurologic outcomes) there were no differences between the groups. The biggest weakness from an emergency department standpoint, aside from being a tiny non-blinded study, is that these patients were treated with aggressive therapy including CSF drainage, so the results may not be applicable to our population.
Bottom line: Hypertonic saline looks better in terms of the surrogate outcome (ICP) but important outcomes are the same. However, this study is severely underpowered to look at any important outcomes.
Cooper DJ, Myles PS, McDermott FT. Prehospital hypertonic saline resuscitation of patients with hypotension and severe traumatic brain injury: a randomized controlled trial. JAMA. 291(11):1350-7. 2004. PMID: 15026402 [free full text]
This is a double blind RCT comparing a 250 ml bolus of either 7.5% saline or Ringer’s lactate in 229 prehospital patients with traumatic brain injury. There was no difference in the primary outcome of neurologically intact survival at 6 months. Much like the Bulger study below, this is a well done study and I like that they chose a patient-oriented rather than a surrogate outcome. The biggest problem might be treating all comers with no sense of their ICP, rather than just those that I would consider treating in the ED – the patients with signs or symptoms of impending herniation.
Bottom line: Prehospital hypertonic saline does not result in improved clinical outcomes of all comers with traumatic brain injury.
Battison C, Andrews PJ, Graham C, Petty T. Randomized, controlled trial on the effect of a 20% mannitol solution and a 7.5% saline/6% dextran solution on increased intracranial pressure after brain injury. Critical care medicine. 33(1):196-202. 2005. PMID: 15644669
This is a prospective, randomized, crossover trial of 9 patients with ICPs greater than 20mmHg, comparing 200 ml of 20% mannitol to 100 ml of 7.5% saline with 6% dextran. The hypertonic saline/ dextran solution lowered ICP more (a median of 13 mmHg versus 7.5mmHg, p=0.044). The clinical importance of that difference is unclear.
Bottom line: A tiny study, but hypertonic saline outperformed mannitol in terms of lowering ICP.
Bentsen G, Breivik H, Lundar T, Stubhaug A. 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. Critical care medicine. 34(12):2912-7. 2006. PMID: 17075367
This is a single-center randomized controlled trial of 22 mechanically ventilated ICU patients with subarachnoid hemorrhage. They were concerned that it would be unethical to perform a placebo controlled trial in patients with elevated ICP, so instead chose patients with a stable ICP in the range of 10-20 mmHg (which could limit the effect of any intervention). The compared 2ml/kg of either 7.2% saline in 6% starch to normal saline, both given over 30 minutes. Unfortunately, they had more than 1 primary outcome (not allowed) and all were surrogates. The change in ICP over 90 minutes was greater in the hypertonic saline group (-3.3 versus -0.3mmHg, p=0.004). The change in CPP was also greater in the hypertonic group (5.4 versus 0.2 mmHg, p=0.002). The clinical significance of these changes is unclear, and no patient oriented outcomes were measured.
Bottom line: In SAH patients without elevated ICP, hypertonic saline seems to lower ICP and increase CPP as compare to NS.
Francony G, Fauvage B, Falcon D. Equimolar doses of mannitol and hypertonic saline in the treatment of increased intracranial pressure. Critical care medicine. 36(3):795-800. 2008. PMID: 18209674
This is a randomized, non-blinded trial comparing 20% mannitol to 7.45% hypertonic saline in 20 stable patients with sustained elevated ICP after traumatic brain injury or stroke. At 30 minutes, both treatments resulted in a similar reduction in ICP (about 10 mmHg), and the ICPs were also the same out to 120 minutes. They measure about 50,000 things in this study and I can’t see a primary outcome listed anywhere. I would therefore be very hesitant to get too granular with the data (especially because the trial only includes 20 patients).
Bottom line: Both mannitol and 7.5% saline resulted in similar reductions in ICP.
Koenig MA, Bryan M, Lewin JL, Mirski MA, Geocadin RG, Stevens RD. Reversal of transtentorial herniation with hypertonic saline. Neurology. 70(13):1023-9. 2008. PMID: 18272864
This is a single center retrospective chart review. They looked for all patients whose chart had an ICD code for cerebral herniation (already a problem, as this coding is unlikely to be accurate), and found those patients who had received 23.4% saline from the pharmacy. The primary outcome they were looking for was clinical reversal of transtentorial herniation (defined as reduction in pupillary diameter, return of light reflex, and an increased in the GCS of 2 or more within one hour of getting hypertonic saline.) This definition is reasonable, but you can imagine how difficult it would be to get accurate information about this neuro exam from the chart. Over a 4 year period, they identified 68 patients who had 76 herniation events treated with 23.4% saline. According to the chart, clinical reversal occurred in 75% of those events. This trial was uncontrolled, and the patients were also simultaneously receiving other therapies, such as mannitol, 2 or 3% saline, propofol, phenobarbital, and even surgical interventions. In fact, 33% of patients were already on a hypertonic saline drip at the time of herniation, and another 33% were given a bolus of 2 or 3% saline in addition to the 23.4%. Clearly, this data is a mess. The retrospective nature of data collection raises big questions about accuracy, but more importantly, the trial is uncontrolled and patients were receiving multiple other medications simultaneously, so who knows what actually reversed the herniation. I will note that the mean serum sodium rose from 140 to 155 mmol/L at 24 hours. Despite all the problems, the authors conclude, “treatment with 23.4% saline was associated with rapid reversal of transtentorial herniation (TTH) and reduced intracranial pressure, and had few adverse effects.”
Bottom line: I don’t think this paper provides any clinically useful information.
Ichai C, Armando G, Orban JC. Sodium lactate versus mannitol in the treatment of intracranial hypertensive episodes in severe traumatic brain-injured patients. Intensive care medicine. 35(3):471-9. 2009. PMID: 18807008
This is a prospective, open-label, single center randomized trial comparing sodium lactate (in a concentration that pretty closely matches 3% saline) to 20% mannitol in 34 patients with isolated severe traumatic brain injuries who were already in the ICU with ICP monitors in place. Patients received a 1.5ml/kg infusion of their assigned medication over 15 minutes. If there was not an adequate response, they were also given the other medication (which occurred in 13 of the 34 patients). The sodium lactate solution did result in a statistically significant lower ICP that then mannitol. I am never involved in ICP monitoring, so I am not sure how clinically important the absolute difference of 4 mmHg is. The group of patients that received both medications had less decreased in their ICP than either of the medications alone, presumably because they were sicker patients. They state in the manuscript that long term neurologic outcomes were better with the sodium lactate, but they only present the data is an electronic supplement that I can’t access. (If you publish a paper that only has one patient oriented outcome, please put the results in the actual manuscript).
Bottom line: Sodium lactate may result in lower ICPs than mannitol, but this small, open-label trial doesn’t definitely prove it.
Bourdeaux C, Brown J. Sodium bicarbonate lowers intracranial pressure after traumatic brain injury. Neurocritical care. 13(1):24-8. 2010. PMID: 20422466
This study is prospective and uncontrolled. They gave 85 ml of 8.4% sodium bicarbonate to 7 patients during 10 different episodes of raised ICP. The ICP feel from a mean of 29 mmHG to a mean of 10 mmHG and remaied below 20 mmHg for 6 hours.
Bottom line: Sodium bicarb may lower ICP, although this uncontrolled trial doesn’t prove it, nor does it tell us anything about clinically important outcomes.
Bulger EM, May S, Brasel KJ. Out-of-hospital hypertonic resuscitation following severe traumatic brain injury: a randomized controlled trial. JAMA. 304(13):1455-64. 2010. PMID: 20924011 [free full text]
This is a large prehospital trial run by the Resuscitation Outcomes Consortium. They compared a single bolus of 250 ml of 7.5% saline to 7.5% saline plus 6% dextran to 0.9% (normal) saline in 1282 traumatic brain injury patients with a GCS of 8 or less without signs of hemodynamic instability. The plan was to enroll 2122 patients, but the trial was stopped early because they met the prespecified criteria for futility. There was no difference in the primary outcomes of neurologically intact survival at 6 months. This trial has some major strengths: it is large, blinded, and the primary outcome is a clinically important outcomes (neurologically intact survival) rather than the surrogate (ICP) that is used in a large number of other trials. The biggest problem might be the enrollment strategy. They included all comers with a GCS of 8 or less. In general, I am only considering using mannitol or hypertonic saline in situations where there are signs of impending herniation. Only 28% of this population was thought to require ICP monitoring on arrival to hospital, and the mean initial ICP at that time was 17 mmHg (but with a large distribution). It is hard to demonstrate the benefit of a medication that reduces ICP in a population where most patients don’t have an elevated ICP. The authors comment that there was no difference in ICP between the two groups, but the monitors weren’t put in until 5 hours after the medications, so that it not surprising.
Bottom line: Routine prehospital administration of hypertonic saline to traumatic brain injury patients does not results in improved clinical outcomes.
Bourdeaux CP, Brown JM. Randomized controlled trial comparing the effect of 8.4% sodium bicarbonate and 5% sodium chloride on raised intracranial pressure after traumatic brain injury. Neurocritical care. 15(1):42-5. 2011. PMID: 21298358
This is a prospective, randomized trial of patients with severe traumatic brain injury, all with ICP monitors in place, comparing 100 ml of 5% sodium chloride to 85 ml sodium bicarbonate 8.4%. They looked at a total of 20 episodes of intracranial hypertesion in 11 patients, and both solutions results in a decrease in ICP at 30 minutes, with no difference between the two. Sodium bicarb looked better at 120 minutes.
Bottom line: Sodium bicarbonate and 5% saline result in similar lowering of ICP, with no assessment of clinical outcomes in this trial.
Cottenceau V, Masson F, Mahamid E. Comparison of effects of equiosmolar doses of mannitol and hypertonic saline on cerebral blood flow and metabolism in traumatic brain injury. Journal of neurotrauma. 28(10):2003-12. 2011. PMID: 21787184
This is a prospective, randomized trial that compared 7.5% hypertonic saline (2ml/kg) to 20% mannitol (4ml/kg) in 47 traumatic brain injury patients with elevated ICP. These patients were all sick enough to already be undergoing ICP monitoring and mechanical ventilation. There were no differences between hypertonic saline and mannitol, either in ICP measurement or neurologic outcomes at 6 months. Both treatment lowered ICP by about 6 mmHg at 30 minutes and 3 mmHg at 2 hours.
Bottom line: Mannitol and hypertonic saline appear equivalent here. They do demonstrate a decrease in ICP with both agents, but it is not clear if that translates into any clinically important outcomes.
Aside from the most recent review I covered, here are two other reviews that covered this topic:
Kamel H, Navi BB, Nakagawa K, Hemphill JC, Ko NU. Hypertonic saline versus mannitol for the treatment of elevated intracranial pressure: a meta-analysis of randomized clinical trials. Critical care medicine. 39(3):554-9. 2011. PMID: 21242790
The authors conclude: “We found that hypertonic saline is more effective than mannitol for the treatment of elevated intracranial pressure.” However, I am not so sure that was actually what they found. The relative risk for ICP control with hypertonic saline was 1.16 with a 95% confidence interval of 1.00-1.33 (so touching 1). The mean difference in ICP was 2 mmHg lower in the hypertonic group, but the 95% confidence interval crossed 1 (-1.6 to 5.7). These differences, even if statistically significant, are small. At least the authors add to their conclusion: “Our meta-analysis is limited by the small number and size of eligible trials, but our findings suggest that hypertonic saline may be superior to the current standard of care and argue for a large, multicenter, randomized trial to definitively establish the first-line medical therapy for intracranial hypertension.”
Rickard AC, Smith JE, Newell P, Bailey A, Kehoe A, Mann C. Salt or sugar for your injured brain? A meta-analysis of randomised controlled trials of mannitol versus hypertonic sodium solutions to manage raised intracranial pressure in traumatic brain injury. Emergency medicine journal : EMJ. 31(8):679-83. 2014. PMID: 23811861
This time the authors conclude “the evidence shows that both agents [mannitol and hypertonic saline] effectively lower ICP. There is a trend favouring the use of hypertonic sodium solutions in patients with TBI”. I wouldn’t be so certain about the “trend” thing, and the difference between the two agents was only 1.39 mmHg, so even if real, I am not sure how clinically relevant it is.
So why did the Berger-Pelleiter paper come to different conclusions? Honestly, I am not sure. They are citing the same underlying trials. The only difference is that Berger-Pelleiter et al seem to consider mannitol in the same class as isotonic comparators, so that the conclusion that hypertonic saline is no different from mannitol gets morphed into hypertonic saline is no different than “any other solution”, which is easily read as placebo. However, the Berger-Pelleiter paper is entirely correct in their conclusion that the current evidence does not indicate any patient oriented benefit (in terms of mortality or improved neurologic function) from treatment with hypertonic saline.
Overall Bottom Line
So what can we take from all of that? It seem pretty clear that hypertonic saline is at least as effective as mannitol at lowering ICP. In fact, there were only 3 trials that reported a difference between the two solutions, and in all three of those trials, hypertonic saline resulted in better ICP control. (However, this is a subject that could easily suffer from publication bias, especially given the small size of all these trials.)
I think the most important question is: should I care about intracranial pressure? ICP is a surrogate endpoint that doesn’t necessarily mean anything for the patient. Every trial that looked at either death or neurologic function was negative. (Including the largest trials to date, in which hypertonic saline was no better than placebo.) Unfortunately, many of those trials included patients without raised ICP, so any real benefit would be diluted. The clear answer is that we don’t know if hypertonic saline helps patients, and that they best evidence to date indicates no improvement in patient oriented outcomes.
Based on these studies, hypertonic saline is probably as effective as mannitol, but how effective is mannitol? Unfortunately, we just don’t know. Mannitol is one of those medications that has been around since before the EBM era, so we use it without real evidence. According to the Cochrane review (Wakai et al 2013 PMID 23918314), there is only one placebo controlled trial ever looking at mannitol, and in that trial mortality was actually increased in the mannitol group (although not statistically so, RR death 1.75 95% CI 0.48 to 6.38).
My conclusion: For now, either hypertonic saline or mannitol is a reasonable choice if you need to treat an elevated intracranial pressure. When we should be treating elevated ICP and whether such treatment results in any patient oriented benefit is still a mystery. We need some large, high quality trials. If you are running one of those trials, please set the primary outcome as a patient oriented outcome such as neurologically intact survival.
The SGEM covers that original review by Dr. Pelletier and conclude that “hypertonic saline as a first line treatment for patients with severe traumatic brain injury cannot be recommended at this time”.
Morgenstern, J. Hypertonic saline for elevated ICP (Articles of the Month special edition), First10EM, October 17, 2016. Available at: