Originally published: Common Sense
November/December 2018
Intravenous fluid (IVF) therapy is one of the most common therapies employed during the care and resuscitation of patients in the emergency, acute, and critical care settings. With several available IVF options such as balanced crystalloids, normal saline (NS), albumin, and other colloid solutions, it is clear that choosing the best solution, the optimal volume, and the appropriate rate of administration is an ever-present clinical challenge. In this review we attempt to address the following questions using recently published literature on the topic:
- Does the use of balanced crystalloids (BC) in the resuscitation of patients confer a mortality/morbidity benefit when compared with NS?
- Does the use of BC decrease the risk of acute kidney injury (AKI) in the acute setting?
D Annane, et al. Effects of Fluid Resuscitation With Colloids versus Crystalloids on Mortality in Critically Ill Patients Presenting With Hypovolemic Shock, The CRISTAL Randomized Trial. JAMA. 2013;310(17):1809-1817.
Crystalloids, including Lactated Ringers (LR), NS, and hypertonic saline, are thought to work based on the osmotic pressure of their contents. Colloids such as albumin, hydroxyethyl starch (HES), and gelatins, on the other hand, act on oncotic pressures to keep fluid in the intravascular space. The authors of the Colloids Versus Crystalloids for the Resuscitation of the Critically Ill (CRISTAL) trial sought to compare the efficacy of these two large classes of fluids in acute hypovolemic shock without making specific comments on the above subtypes.
The study was a multicenter, randomized, open label trial in five countries across three continents. Due to concerns of the inability to blind simultaneously at multiple centers, the authors instead conducted a blinded outcome assessment. Patients needed to have evidence of hypovolemia, defined as hypotension, evidence of low filling pressures and low cardiac index, and signs of tissue hypoperfusion or hypoxia. Exclusion criteria included pregnancy, DNR, brain death, burns >20% of body surface area, ESRD on dialysis, advanced chronic liver disease, known allergy, acute anaphylaxis, anesthesia related hypotension, and dehydration. Patients were categorized as having sepsis, trauma, or hypovolemic shock without sepsis or trauma. Exceptions to assigned fluid type included isotonic crystalloids for maintenance fluids in all patients and the administration of albumin to patients in the crystalloid group to treat hypoalbuminemia. The primary outcome was death at 28 days, with secondary outcomes of 90 day mortality, and days alive without mechanical ventilation or vasopressor therapy.
Across all the centers, 1,414 patients were enrolled in the colloid arm and 1,443 in the crystalloid arm over a nine year period to reach 90% power. There were no significant difference in baseline characteristics or the fluid type administered before enrollment. As expected, patients in the colloid arm received significantly less fluids than those in the crystalloid arm (2 L vs 3 L, p<0.001) over the same time period. A total of 16.5% of patients in the crystalloid arm received albumin, which was permitted for hypoalbuminemia. There was no significant difference in 28 day mortality (359 vs 390 deaths, RR 0.96, p=0.26). They completed subgroup analysis by comparing fluid types one-on-one for all types of hypovolemia and for sepsis, with no significant difference in mortality. They did find significantly less deaths at 90 days in the colloid arm as compared to the crystalloid arm (434 vs. 493 deaths, RR 0.92, p=0.03), as well as more days alive without mechanical ventilation at 28 days (14.6 vs. 13.5 days, p=0.01) and without vasopressor therapy at 28 days (16.2 vs 15.2 days, p=0.03) in the colloid arm. These results were unexpected by the authors, and the study therefore was not adequately powered to detect significant differences between them.
The authors were able to conclude that there was no difference in 28 day mortality, regardless of shock type. They did report that each center had a significantly different mortality rate, but that this difference was not associated with any treatment effect. Although they compared fluid types one-on-one, they did not make any conclusions on hyperoncotic vs. hypooncotic colloids, isotonic vs. hypertonic crystalloids, or buffered vs. non-buffered solutions. This was an expected consequence of their goal of involving a large number of patients across multiple centers, and not instituting strict requirements for the type of fluid administered within the broader categories of colloids and crystalloids. The main flaw in their analysis was the failure to address the 25.3% of patients in the colloid arm who received crystalloids incorrectly and the 6.5% of patients in the crystalloid arm who received colloids incorrectly. These patients are included in the intention to treat analysis without any separate subgroup analysis to determine if their results altered the primary outcome. So while their conclusion that colloids do not increase mortality as previously stated in other studies is an important one, it is difficult to evaluate in the context of this large crossover rate. The authors admit that the reduced 90 day mortality requires additional study, but their conclusion regarding no significant difference in 28 day mortality also requires re-evaluation.
P Young, et al. Effect of a Buffered Crystalloid Solution vs Saline on Acute Kidney Injury Among Patients in the Intensive Care Unit, The SPLIT Randomized Clinical Trial. JAMA. 2015;314(16):1701-1710.
The non-physiologic strong ion difference (SID) of NS is hypothesized to be behind its association with AKI when compared with balanced salt solutions. However, much of the evidence for the association between NS and AKI was derived from observational studies with perioperative subjects. The Saline vs Plasma-Lyte 148 for ICU fluid Therapy (SPLIT) trial was designed to assess whether there was such an association among patients admitted to the ICU. This was a double blinded, cluster, randomized, double crossover study comparing the effects of NS and plasma-lyte (PL) on renal function. It was performed in four tertiary care ICUs in New Zealand, 3 of which were mixed (adult and pediatric) medical and surgical ICUs, and the other with predominantly cardiothoracic and vascular surgical patients. The primary outcome was the proportion of patients with AKI (based on the RIFLE definition); secondary outcomes were the difference between baseline and peak serum creatinine, cumulative incidence of AKI based on RIFLE and KDIGO definitions, the use and indications for renal replacement therapy (RRT) in the ICU, and the requirement for RRT post discharge. Each participating ICU was assigned to use a blinded study fluid (either NS or PL) for alternating treatment blocks lasting 7 weeks. After 7 weeks passed, the blinded fluid was crossed over. As the study period lasted 28 weeks, each participating ICU used each study fluid twice. Fluids were contained in indistinguishable 1 L bags, either labeled “fluid A” or “fluid B” depending on the treatment block. Two ICUs initially used 1 fluid, and the other two initially used the alternative. Both investigators and clinicians were blinded to the fluid allocation for the study duration. The study was funded via a grant from the Baxter Healthcare Corporation.
There were 1162 patients (10 opted out; 85 had missing primary outcome [no baseline and/or no ICU creatinine measured] data) in the PL group and 1116 patients (6 opted out) in the NS group. Both groups had similar baseline characteristics and received similar volumes of IVF (median 2 L for NS [range 1–3.5 L] vs median 2 L for PL [range 1-3.25 L]). Patients who remained in the ICU for more than one treatment block were continued on their originally assigned fluid. Of note, both groups had large proportions of surgical (particularly elective) patients.
The primary outcome of AKI was seen in 9.2% (94/1025) of the NS group and in 9.6% (102/1067) of the PL group, (absolute risk 0.4% [95% CI -2.1% to 2.9%], relative risk 1.04 [95% CI 0.8 to 1.36, p=0.77), with a number needed to treat of 250. Similarly, there were no significant differences noted between the two study groups with respect to the secondary outcomes, duration of mechanical ventilation, duration of ICU stay, hospitalization, ICU mortality, or hospital mortality. Subgroup analysis suggested an association between ICU site, fluid choice, and the risk of developing AKI (composite p=0.05).
The trial had several key limitations. First, though patients were allocated to either Fluid A or Fluid B depending on their location and the treatment block, open label NS and balanced salt solution were available in situations with a specific indication for their use. Additionally, there were no restrictions on the use of other fluids or therapies beyond the use of fluid A or fluid B during the study period, and as study enrollment was defined as the time from first administration of study fluid, there was no control in place for fluid choice and administration prior to ICU admission/transfer (93% of the subjects from each group received IVF, mostly balanced salt solutions, prior to study enrollment). Finally, the relatively high p values and the authors’ admission that they did not perform sample size calculation indicate that the trial was likely insufficiently powered to detect a significant difference between groups. As such, it is difficult to make any definitive conclusions regarding choice of IVF based on the SPLIT trial.
M Semler, et al. Balanced Crystalloids versus Saline in Critically Ill Adults. N Engl J Med. 2018; 378:829-839.
Some evidence in the critical care literature suggests that the use of NS solution is associated with AKI, hyperchloremic acidosis, and an increase in mortality. Therefore, there has been great interest in identifying whether or not the type of IVF composition truly has any effect on any or all of these outcomes. The Isotonic Solutions and Major Adverse Renal Events Trial (SMART) was a pragmatic, cluster-randomized, unblinded, multiple crossover trial that compared BC – either LR or PL to NS as the fluid of choice in critically ill adults in both surgical and medical intensive care units (ICUs). The trial was conducted across 5 ICUs at Vanderbilt University, a single large academic center in the U.S. The primary outcome was major adverse kidney events at 30 days (composite outcome, including all-cause mortality, need for RRT, and persistent renal dysfunction [final creatinine ≥200% baseline]). Each month participating ICUs and the corresponding department that primarily admitted to them, ED or operating room, were randomized to either BC or NS.
There were 7942 patients in the BC group compared to 7860 patients in the NS group. The baseline characteristics between the two groups were similar. The median volume of IVFs administered in the BC group was 1 L with an interquartile range (IQR) of 0-3.21 L, while in the NS group it was 1.020 L (IQR 0-3.5 L). Since the type of fluid given depended on which IVF was randomly assigned that particular month, there was potential for receiving both BC and NS in patients who were in the ICU at the transition from one month to the next. However, only 5.4% and 4.4% of the patients in the BC and NS group, respectively, received a fluid from the other group as a result of this.
The composite primary outcome of major adverse kidney events within 30 days occurred in 14.3% of the BC group and in 15.4% of the NS group (marginal OR 0.91, 95% CI 0.84-0.99, conditional OR 0.90, 95% CI 0.82-0.99, p=0.04). Interestingly, in a prespecified subgroup of septic patients, the outcome of 30 day in-hospital mortality was 25.2% in the BC group compared to 29.4% in the NS group (adjusted OR 0.80, 95% CI 0.67-0.97, p=0.02), with a number needed to treat of 23. Serum chloride and bicarbonate concentrations were significantly higher in the NS group and lower in the BC group. The difference was proportional to the volume of IVF administered. There was no statistically significant difference between the individual components of the composite outcome across the two groups.
This trial adds further evidence to the idea that BC may reduce the likelihood of all cause death and AKI in the ICU, particularly in septic patients. While the trial only showed a small benefit in absolute risk reduction of the composite outcome, the authors argue that when this small benefit is applied to large numbers of ICU patients each year, there is the potential for significant benefit. The trial does have some limitations, including the fact that it is single center, possibly affecting its external validity and applicability to other settings. Importantly, the use of a composite primary outcome may overestimate any benefit. In addition, the trial was unblinded, and although the trial compared BC to NS, both PL and LR were included in the BC group, making direct comparisons between the two difficult.
W Self, et al. Balanced Crystalloids versus Saline in Noncritically Ill Adults. N Engl J Med. 2018; 378:819-828.
In response to prior studies indicating the potential benefit of BC over NS, researchers at Vanderbilt decided to develop a sister study to the already up-and-running SMART trial to determine whether the difference in outcomes between patients receiving BC versus NS was affected by the underlying severity of illness in the patient. In addition to potentially expanding the general applicability of the findings in the SMART trial, the inclusion of a study (Saline Against Lactated Ringer’s or Plasma-Lyte in The Emergency Department [SALT-ED]) that focused on non-critically ill patients piqued the interest of providers in the ED setting, where the majority of patients do not meet the definition of critically ill as defined in the earlier SMART trial.
Like its sister study, SALT-ED was a single-center (academic tertiary care center), prospective, doubly-unblinded, pragmatic, multiple-crossover trial comparing BC to NS. Patients were enrolled over a 16 month period but overlapped with that of the SMART Trial. Inclusion criteria required that subjects be ED patients greater than 18 years old, receive at least 500 mL of crystalloids, and be admitted to the hospital. The exclusion criteria were admission to the ICU or discharge home. A total of 13,347 patients met criteria for inclusion, with 6708 (50.3%) assigned to BC and 6639 (49.7%) assigned to NS. The study-designated crystalloid alternated each month, but the final decision of which fluid to administer was ultimately up to the clinician. During BC months, the clinicians could choose either LR or PL. In addition, the selection of fluids after transfer from the ED was not included in the final assessment.
The primary outcome was a composite of in-hospital death and hospital length of stay (LOS), and was defined as the number of days alive and out of the hospital between the index ED visit and 28 days later. The three secondary outcomes were 1) major AKI within 30 days (composite outcome, including death, new RRT, or persistent renal dysfunction [final creatinine ≥200% baseline]); 2) AKI of stage 2 and higher (maximum creatinine ≥200% baseline, creatinine ≥4 mg/dL with an absolute increase of ≥0.5 mg/dL, or initiation of new RRT before hospital discharge or within 30 days of the initial ED visit); and 3) all-cause in-hospital mortality. Of note, patients with no available baseline creatinine level were still included in the study, but they were assigned a calculated baseline creatinine based on an equation developed by the authors.
The authors concluded that despite differences in the levels of chloride and bicarbonate between the NS and BC groups, there were no significant differences in the number of hospital free days. However, they did argue for a difference in their secondary outcome, citing a lower incidence of 30-day major AKI in the BC group (4.7% vs. 5.6%; adjusted OR, 0.82; 95% CI, 0.70 to 0.95; p=0.01).
The trial was overall well designed but its results are inherently limited by all of the obvious biases associated with single-center, unblinded studies. Its findings are further mitigated by their nature, as composite outcomes are inherently less reliable than individual outcomes. In addition, despite the study’s noble attempt to include non-critical patients, its broader applicability is still limited by its overall demographics, which were largely white (77%), middle-aged (median age 54 years) patients (~1:1 male:female) who were primarily admitted to medical (~75%) services rather than surgical (~25%) services. In addition, the study population had little to no baseline renal dysfunction (baseline creatinine 1.32 [mean], 0.93 [median]; ~20% had baseline creatinine >1.5; ~10% had chronic kidney disease [CKD] ≥ stage 2; ~2% had baseline end-stage renal disease. Thus, these results may not apply to non-white patients, surgical patients, or those with more advanced renal disease.
The study’s two most critical flaws center on its reliance on calculated baseline renal function and on the low volume of exposure to the treatment fluids. More than one third of the patients’ baseline creatinine levels were unavailable. This should give the reader significant pause, since the study’s only statistically significant finding was a calculated, secondary, composite outcome. Perhaps equally concerning is the amount of fluids given; the total amount of fluid administered to each patient was startlingly low, with a median volume of only 1 liter per patient. As a result, it is extremely hard to imagine how such small volumes of fluids could be the driving force for any significant differences in the patients’ overall mortality or renal outcomes.
Despite its good intentions, the SALT-ED trial was unfortunately too strongly limited by its inherent biases, limited fluid volumes, and composite outcomes to be able to glean any real insight into the potential differences between BC and NS in non-critically ill patients.
Conclusion
Studies of IVF therapy have yet to make any strong statement regarding a superior choice with regards to shock type, adverse outcomes, or mortality benefit. The SMART trial was able to conclude that renal outcomes may be improved with BC although their analysis methods have been called into question, while the other studies, including the SPLIT trial, were not powered or designed appropriately to be practice changing. Overall, the studies into IVFs demonstrate that there is not one appropriate IVF choice, but rather with the information available an IVF must be chosen to fit the clinical situation. Further research into fluid use will be important in understanding their respective effects. It is clear that well designed studies will be difficult to produce given the inherent challenges in providing controls and preventing bias.
Now we return to the questions posed in the introduction:
- Does the use of BC in the resuscitation of patients confer a mortality/morbidity benefit when compared with NS?
The available evidence is not strong enough to make any conclusive statement about mortality benefit of BC over NS.
- Does the use of BC decrease the risk of AKI in the acute setting?
There does appear to be a trend towards improved renal outcomes with BC with the best evidence from the SMART trial, however the other studies were not designed or powered to state this with certainty. Further research is required to detect a statistically significant change in the acute care setting.
References
- Bampoe, S. et al. Perioperative administration of buffered versus non-buffered crystalloid intravenous fluid to improve outcomes following adult surgical procedures. Cochrane Database Systematic Reviews, 9
- Finfer, S., Myburg J., and Bellomo, R. Intravenous Fluid Therapy in Critically Ill adults. Nature Reviews Nephrology. 14(9): 541-557
- Myburgh JA, Mythen MG. Resuscitation Fluids. New England Journal of Medicine. 2013;369(13):1243-1251. doi:10.1056/nejmra1208627.