Do colloids save lives in hypovolemic shock?

So, your patient's in shock: quick, give some fluids. But colloids or crystalloids? How to choose? They both raise blood pressure, they both improve organ perfusion -- but one's less filling, the other tastes great (what, you haven't tried them?).

It's no wonder the question makes your head hurt; the evidence base is a jumble. The SAFE study found no difference between normal saline or 4% albumin in outcomes of 7,000 ICU patients who required intravascular resuscitation for a variety of diagnoses. A meta-analysis of 1,977 patients with sepsis detected an 18% relative mortality reduction among patients receiving albumin solutions for resuscitation compared to crystalloid. The Surviving Sepsis Guidelines authors advise use of crystalloids as first-line, but endorse albumin as a helpful adjunctive therapy for septic shock. Hydroxyethyl starches, meanwhile, fell into regulatory disfavor for use in septic shock after they were shown to produce renal failure and death in sepsis patients in randomized trials.

In hypovolemic shock, on the other hand, colloids (including hydroxyethyl starches) are considered helpful as "volume expanders," and are widely used with less controversy. The U.S. military gives hydroxyethyl starch (hetastarch) to its wounded in combat theaters; its protocol teaches medics that a liter of hetastarch "is intravascularly equivalent to six liters of Ringers lactate" (no reference provided; this may have been written during the government shutdown). Albumin is also used for intravascular repletion in hypovolemic patients.

So ... what about septic shock patients who happen to also be hypovolemic? Are hydroxyethyl starches good for them?

Djillali Annane et al stir that pot by reporting the results of their massive CRISTAL randomized trial testing colloids vs. crystalloids for hypovolemic shock online in the October 9, 2013 JAMA. Their work sows confusion over the previous trials demonizing hydroxyethyl starches, and begs the question of whether can colloids not just substitute for crystalloid, but actually save lives in people with hypovolemic shock -- including those with concomitant septic shock.

What They Did

Investigators randomized 2,857 consecutive patients admitted to 57 ICUs with hypovolemic shock in France, Belgium, North Africa and Canada to receive either colloids or crystalloids for all resuscitation fluids throughout their ICU stays.

To be eligible, patients had to be in objective shock, and also to have evidence for hypovolemia with a low cardiac index, measured invasively or noninvasively (orthostatic hypotension, increased pulse pressure variation, or clinical signs). Importantly, patients could be enrolled with severe sepsis, if they also had evidence for hypovolemia. Such patients comprised about 55% of both cohorts. About 40% were pure hypovolemic shock without trauma, while 6% had multiple trauma.

Crystalloids could be isotonic saline (86% of patients in the crystalloid group) or hypertonic saline or Ringer lactate; colloids could include 4% or 20% albumin, gelatins, dextran, or hydroxyethyl starches (70% of patients in the colloid group got HES). Patients all received isotonic crystalloids as maintenance fluids.

The primary outcome was 28-day mortality; multiple prespecified secondary outcomes included 90-day mortality, ventilator-free days, vasopressor days, renal failure, etc. Enrollment took 9 years to complete. The study was funded by the French Ministry of Health.

Although the treatment teams knew who was receiving which treatment (open label; unblinded), those assessing patients' outcomes were blinded to their treatment assignment.

What They Found

Patients in the colloids group received an average of 2,000 mL in the first 7 ICU days, compared to 3,000 mL in the crystalloid group.

Mortality was statistically similar between groups at 28 days (25% with colloids, 27% with crystalloids), with a trend favoring colloids. At 90 days, mortality was significantly lower among the colloid-treated patients (31% vs 34%, p=0.03), with 59 more patients alive in the colloid-treated group. There were apparent small improvements with colloid therapy in other secondary endpoints, as well, with an average of a one-day reduction in the need for mechanical ventilation or vasopressors over 28 days, per patient.

Notably, there was no increase need for renal replacement therapy in the colloid treated group. Hydroxyethyl starches had previously been shown to cause real failure in patients with severe sepsis and septic shock. Patients with severe chronic renal failure were excluded from the trial, though.

About 20% of patients in the colloid group received protocol-violation crystalloids; few in the crystalloid group received colloids.

There was no signal of harm for hydroxyethyl starch in patients with severe sepsis / septic shock, in non-prespecified subset analyses.

What It Means

The makers of hydroxyethyl starches are likely ebullient over this trial, which showed no evidence of harm in the use of hydroxyethyl starch solutions in patients with septic shock who are also hypovolemic. This, after earlier randomized trials showing harm resulted in an effective ban of these products in patients with septic shock in Europe and the U.S.

However, its impact on clinical practice is likely to be minimal, at least in the U.S. Although hydroxyethyl starches were given to 70% of the patients in the colloid group here, in multiple prior trials specifically testing HES or similar preparations in a more controlled (and blinded) fashion among septic shock patients, there was either no benefit or possible harm from HES or similar starch solutions. Authors' explanation for the absence of harm signal here was their restriction of HES dose below prespecified regulatory limits (and, by implication, excessive dosing in the prior trials showing harm). However, the lack of blinding to treatment allocation resulting in unmeasured treatment differences between groups is just as plausible a possibility.

For most of us, the bottom line is that regulatory agencies on two continents and clinical practice guidelines advise against the use of hydroxyethyl starch in patients with severe sepsis. This real-world/pragmatic trial of heterogeneous patients, however interesting, won't change that. Given the preponderance of negative or harm-trending trials in this area (now totaling more than 18,000 patients), the "definitive" study (i.e., large enough to settle the question) probably can't and won't ever be done.

Even after reading the supplementary appendix, I can't understand how they diagnosed hypovolemia in their subjects. It's notoriously difficult to do reliably, without a pulmonary artery catheter (and some would say, with one) -- which only 31 of their patients received. It looks to me like hypovolemia was identified by clinical criteria like labs and patient symptoms like thirst and dizziness in the large majority of cases. If this is in fact how they did it, I wonder how much we really know about the patients' initial volume status, which was key to the whole trial. Nor do I understand the diagnoses for the ~40% of patients with nontraumatic hypovolemic shock (were they GI bleeds? Why not just say so?).

Clinical Takeaway: Methodologic quibbles notwithstanding, CRISTAL is another essentially negative trial that raises the bar even higher for any future trial to establish a convincing benefit from colloids as a standard resuscitation fluid for patients with shock, hypovolemic or otherwise. Most experts will probably agree it remains entirely appropriate to use colloids in selected patients in shock, especially when volume overload is a concern (e.g., concomitant decompensated congestive heart failure or cirrhosis). I'll be using albumin.

Djillali Annane et al. Effects of Fluid Resuscitation With Colloids vs Crystalloids on Mortality in Critically Ill Patients Presenting With Hypovolemic Shock: The CRISTAL Randomized Trial. JAMA. 2013;310(17):1809-1817. doi:10.1001/jama.2013.280502.