Another 0.9% Saline versus Balanced Solution Trial: BaSICS released
“May I be the tiniest nail in the house of the universe, tiny but useful."
The use of ‘balanced’ intravenous fluids as compared to 0.9% saline – so-called ‘normal’ saline – has garnered much scientific interest in the last half-decade. Propelling some of this curiosity is a newfound appreciation for Peter Stewart’s ‘strong-ion’ explanation of plasma acid-base balance. Central to Stewart’s mathematical epistemology is the chloride anion which, like other strong ions, is an independent variable controlling proton concentration. Accordingly, fluids rich in chloride shrink the strong ion difference, diminish pH and, ostensibly, harm patients. Further, chloride is known to activate juxta-glomerular feedback and, therefore, disturb renal perfusion. Given these aforementioned concerns and the hundreds of millions of litres of crystalloid administered every year, employing chloride deplete solutions for intravenous resuscitation might prompt favourable patient-centred outcomes, especially at the population level.
With this, Zampieri and colleagues have recently published the Balanced Solutions in Intensive Care Study [BaSICS], comparing both 0.9% saline to balanced IV solutions as well as two infusions rates; the former aspect of their laudable clinical investigation is touched upon below.
What they did
Across 75 intensive care units in Brazil, patients were enrolled who:
1.] needed at least 1 fluid expansion,
2.] were anticipated to stay in the ICU beyond 24 hours,
3.] had at least one of the following risk factors: a.] older than 65 years; b.] hypotension; c.] sepsis – defined as suspected or confirmed infection plus acute organ dysfunction; d.] required invasive or non-invasive mechanical ventilation [including high-flow nasal cannula] for at least 12 hours; e.] early signs of kidney dysfunction [either oliguria or serum creatinine >1.2 or > 1.4 mg/d for women and men, respectively; or f.] had cirrhosis or acute hepatic failure.
These patients were randomized in a two-by-two factorial design to either a balanced IV solution or 0.9% saline as well as to two different infusion rates of said fluids. The bags of fluid were identical; the study was therefore double-blinded. Importantly, in this study, the balanced solution was Plasma-lyte 148 [i.e., Plasma-lyte A].
The primary outcome was 90-day survival. Secondary outcomes included: the need for kidney replacement therapy up to 90 days; acute kidney injury defined as the progression KDIGO stage 0 or 1 at enrollment to at least 2 at days 3 and 7; SOFA score; and the number of days not requiring mechanical ventilation within 28 days. The tertiary outcomes were ICU and hospital mortality and ICU and hospital length of stay.
What they found
11,052 patients were randomized and 10,520 patients were available for final analysis; 5,230 patients were randomized to balanced fluid resuscitation and 5,290 to 0.9% saline. The median cumulative study fluid infused for both groups through the first 3 days was approximately 2.9 litres with an additional 1.2 litres of non-study fluid during the same time frame.
At 90 days there was no statistically significant difference in mortality between the two groups [i.e., 26.4% in the balanced group and 27.2% in the saline group].
In total 19 secondary outcomes were evaluated. Notably, at day 7, neurological SOFA score was more likely to be above 2 in those randomized to balanced fluids. As well, 90-day survival was significantly lower in the pre-specified subgroup of patients with traumatic brain injury who received balanced solution.
Upon reading BaSICS, I was struck by how similar the results were to both SMART and SPLIT. Recall, that while SMART was a larger [n = 15,802], ‘positive’ study because there was a statistically-significant difference between randomized treatment groups, SMART evaluated a composite primary end-point. Death from any cause, new renal-replacement therapy, or persistent renal dysfunction at 30 days was the composite, primary end-point in SMART. Individually, none of these 3 outcomes were statistically significant, but in totality they were reduced by 1.1% within the patients randomized to balanced solution as compared to ‘normal’ saline. Interestingly, death at 30 days was the largest contributor to the composite end-point in SMART; death from any cause was diminished by 0.8% in those randomized to receive balanced fluids. Unfortunately, the mortality outcomes cannot be directly compared between these trials as BaSICS reports death at 90 days. SMART does report in-hospital death at 60 days – a 0.7% absolute risk reduction in those who received balanced solutions – but the mortality curves in SMART diverge around day 7 while this is clearly not realized in the cumulative mortality curves presented in BaSICS.
There are other important differences between BaSICS and SMART worth noting. It might be tempting to describe BaSICS patients as less critically-ill given the large portion of admissions following elective surgery [i.e., 48.4%], however, in totality BaSICS patients appear more ill at baseline. Larger fractions of patients enrolled in BaSICS were hypotensive or receiving vasopressors and mechanical ventilation as compared to SMART patients [i.e., ~60% versus ~26% and ~45% versus ~35%, respectively]. As well, predicted in-hospital mortality was greater at baseline for BaSICS compared to SMART [i.e., 15% versus 9.5%]. Curiously, the observed, in-hospital mortality was much greater for BaSICS patients at roughly 23% while in-hospital death before 60 days in SMART was approximately 11%. Lastly, patients in BaSICS had worse baseline renal function with median and Q3 creatinine levels of 1.0 mg/dL and 1.4 mg/dL, while these values were 0.9 mg/dL and 1.1 mg/dL for SMART.
Arguably, the sicker, less renally-robust patients in BaSICS would derive the most benefit from the smaller chloride load engendered by Plasma-lyte, however, in the first 7 days, the difference in chloride levels between the saline and balanced groups was remarkably similar between BaSICS and SMART [i.e., roughly 1-2 mEq/L difference]. Why might this be?
The volume of infused, trial-fluid in the first week appeared slightly greater in BaSICS. Specifically, in SMART, those randomized to balanced fluids received slightly more than 2 litres of balanced fluids in the first 7 days [approximately 50% of this was lactated ringers and the other half Plasma-lyte]. This group also received about 500 mL of 0.9% saline. SMART patients randomized to 0.9% saline similarly received slightly over 2 litres of saline and roughly 500 mL of balanced fluids in the first week [again split between lactated ringers and Plasma-lyte]. However, in BaSICS, these values were approximately 3 litres of assigned, trial-fluids to each group [though the balanced group received all Plasma-lyte] plus additional 500-1000 mL of non-trial fluids. Importantly, however, a much larger fraction of patients in BaSICS received non-trial fluids in the 24 hours prior to enrollment. That is, roughly 1-in-5 patients randomized to Plasma-lyte in BaSICS received at least 1000 mL of 0.9% saline in the 24 hours prior to enrollment. This was avoided in SMART because the trial was coordinated with the emergency and surgical departments at the single academic centre running the study.
Given the above, the narrow chloride difference observed between the two groups in BaSICS may have been due to contamination from the high chloride load in 0.9% saline. As a reminder, the chloride concentration of 0.9% saline is 154 mEq/L, while it is 109 and 98 mEq/L, respectively, for lactated ringers [LR] and Plasma-lyte. Plasma-lyte is buffered by acetate and gluconate, while lactate buffers LR. Consequently, given the lower chloride concentration in Plasma-lyte and the slightly larger volumes over the first week, the expectation is that the chloride levels would achieve a greater separation in BaSICS as compared to SMART, but this was not observed, ostensibly due to cross-contamination.
Though sub-group analyses hint at superior neurological outcome with 0.9% saline, the authors rightly caution on drawing firm conclusion. Nevertheless, the greater osmolarity of 0.9% saline compared to Plasma-lyte [i.e., 308 versus 294 mOsm/L] does have physiological plausibility. Yet, recall the large benefit observed in the subgroup of septic patients randomized to balanced fluids in SMART; this was not replicated in BaSICS.
In summary, we are left with 3 large trials [i.e., SPLIT, SMART and now BaSICS], having cumulatively randomized over 29,000 critically-ill patients to 0.9% saline or balanced solution, each showing a small, clinically-significant, statistically-insignificant mortality reduction in those randomized to balanced solutions and completely underwhelming effects on renal-specific outcomes. Can we get a meta-analysis for 90-day mortality please? For context, consider that IV beta-blockers during acute myocardial infarction fell from favour after a 45,852-patient trial showed an absolute risk increase of cardiogenic shock by 1.1% in those who received beta-blockers intravenously even though this risk increase was offset by 1% risk reduction in re-infarction and ventricular fibrillation. If 90-day mortality were statistically-significantly reduced by slightly less than 1% in a 29,000 patient meta-analysis of 0.9% saline versus balanced solutions, would this change practice?