There remains significant uncertainty and practice variation about the use of crystalloids in the ICU. Normal saline (NS), being both cheap and widely available, is the most commonly administered crystalloid worldwide. Despite its abundant use, however, there has been growing concern about the detrimental effects of NS, particularly in regards to acute kidney injury (AKI) – review covering wider perspective of AKI. This widespread concern has not translated easily into unambiguous evidence for widespread harm, despite very strongly held opinions on both sides.

The concern arises from (arguable) physiological plausibility and a decent mix of observational studies. For example this study from 2012 found a significant reduction in renal artery blood flow and renal cortical tissue perfusion in healthy volunteers given 2 litres of normal saline compared to those given a buffered crystalloid solution and a now classic observational study by Yunos et al. in 2010 (editorial) found that removing chloride rich fluids from their ICU was associated with a significant decrease in the incidence of AKI and use of RRT.

Against a background of these and similar studies The SPLIT Randomized Clinical Trial (editorial) is a cluster randomized, double-crossover study involving four tertiary ICUs in New Zealand. Patients were randomized to NS or Plasma-Lyte 148, with clinicians blinded to which fluid they were administering, but otherwise allowed to use crystalloid as they felt appropriate. See below for comparison of different crystalloid solutions.

Primary outcome was the proportion of patients who developed AKI, based on their relative increase in serum creatinine. Secondary outcomes included; delta creatinine (peak Cr – admission Cr), cumulative incidence of AKI, incidence of RRT, indications for RRT, length of stay, readmission, and all cause mortality. After randomization, 2278 patients were enrolled. The average volume of crystalloid given was 2000ml in first 24 hours (range 1000-3500ml). The majority of admissions (56 and 58%) came to the ICU after elective surgery. A minority of admissions were for trauma (3 and 5%) or sepsis (4%).

This study is almost flawless, a model for critical care research, and despite that is another negative critical care RCT. There was no difference in the important outcomes between the groups, 9.6% of patients receiving buffered crystalloid vs 9.2% of patients receiving saline developed AKI. (RR 1.04 with 95% CI of 0.80-1.36, p=0.77). RRT was used in only 3.3% of the buffered crystalloid group and 3.4% of the saline group. There was no significant difference in ICU length of stay, readmission, or all cause mortality.

We’ve been here before – can we glean anything useful?

We could think about our wider approach to fluids – when we start them, how much we use, what we should use – one question which we talk about every day is when to start vasopressors, particularly in trauma, it is easy to lose focus and end up with a trauma patient who is 5L positive in the first 24h after admission but with no focus of haemorrhage and ongoing hypotension – is fluid really the right thing to do?


For discussion:

This study appears to have excellent methodology and internal validity. Can these results be generalized to our NCCU environment?

Was the volume of crystalloid given large enough to observe an effect, and is this comparable to our clinical practice?

How were two thirds of physicians able to correctly identified the crystalloid solution, and did this influence the study?

This study used Plasma-Lyte 148. Should we extrapolate these findings to other buffered crystalloids such as Lactated Ringers?

Is there a cost difference between these solutions, and should that weigh into our decision?


Further reading:

LITFL

CCForum – Bellomo – essential reading

NEJM – Myburgh and Mythen – essential reading

Nature Reviews Nephrology – Bellomo

PulmCrit (EMCrit) – 1

St Emlyn’s, Virchester – essential reading

PulmCrit (EMCrit) – 2

RebelEM

The Bottom Line – essential reading

Finally, Scott Weingart’s perspective on acid-base and crystalloids: EMCrit