Skip to main content

Foundational Trials

BASICS - Fluids

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effect of Intravenous Fluid Treatment With a Balanced Solution vs 0.9% Saline Solution on Mortality in Critically Ill Patients: The BaSICS Randomized Clinical Trial
  • Acronym: BaSICS
  • Year: 2021
  • Journal published in: JAMA
  • Citation: Zampieri FG, Machado FR, Biondi RS, Freitas FGR, Veiga VC, Figueiredo RC, et al. Effect of Intravenous Fluid Treatment With a Balanced Solution vs 0.9% Saline Solution on Mortality in Critically Ill Patients: The BaSICS Randomized Clinical Trial. JAMA. 2021;326(9):818-829.

Context & Rationale

  • Background
    • 0.9% saline is a high-chloride, non-buffered crystalloid; mechanistic and observational work linked chloride-liberal strategies to hyperchloraemic acidosis and potential kidney hypoperfusion/injury.
    • Balanced crystalloids (e.g., Plasma-Lyte, lactated Ringer’s) more closely approximate plasma electrolyte composition and include buffer anions, but their impact on patient-centred outcomes remained uncertain.
    • Pragmatic cluster-crossover trials suggested balanced crystalloids reduce composite kidney endpoints and may reduce mortality in some subgroups, but design features (cluster-level allocation, composite outcomes, single health system) left residual uncertainty regarding true mortality effects in general ICU populations.12
    • Whether balanced crystalloids are beneficial, neutral, or harmful for key ICU subgroups (notably traumatic brain injury) required a large, patient-level randomised trial with robust mortality follow-up.
  • Research Question/Hypothesis
    • In critically ill adult ICU patients requiring IV fluids and at risk of acute kidney injury, does a balanced multielectrolyte solution (Plasma-Lyte 148) compared with 0.9% saline improve 90-day survival?
    • Factorial co-intervention: does a slower infusion rate compared with a usual/control rate affect outcomes, and is there interaction between fluid type and infusion speed?
  • Why This Matters
    • Crystalloids are among the most frequently administered ICU therapies; small relative benefits or harms could have substantial population impact.
    • Practice variation is wide and guidelines historically relied on mixed evidence; a definitive, large ICU RCT could rationalise global fluid selection.
    • Potential subgroup-specific effects (e.g., sepsis vs traumatic brain injury) have direct bedside implications for resuscitation choices.

Design & Methods

  • Research Question: Among critically ill adults in the ICU requiring IV fluids and at risk of AKI, does treatment with a balanced solution (Plasma-Lyte 148) vs 0.9% saline reduce 90-day mortality?
  • Study Type: Investigator-initiated, pragmatic, multicentre, double-blind (fluid type), 2×2 factorial, randomised clinical trial conducted in 75 ICUs in Brazil.
  • Population:
    • Adults admitted to the ICU with expected ICU stay >24 hours, for whom the treating clinician considered both trial fluids appropriate.
    • Required IV fluid therapy in ICU (fluid challenges, maintenance fluids, and/or drug infusions >100 mL), plus ≥1 AKI risk factor (age >65 years; hypotension/MAP <65 mmHg or SBP <90 mmHg or vasopressor use; sepsis; mechanical ventilation including high-flow nasal cannula expected ≥12 hours; oliguria <0.5 mL/kg/h for >6 hours or serum creatinine >1.2 mg/dL; liver cirrhosis or acute liver failure).
    • Key exclusions: need for kidney replacement therapy (or expected within 6 hours); serum sodium ≤120 or ≥160 mmol/L; death considered imminent within 24 hours, suspected/confirmed brain death, or palliative care only; previously enrolled.
  • Intervention:
    • Balanced solution: Plasma-Lyte 148 supplied in indistinguishable 500 mL bags (coded), used as the preferred crystalloid for fluid challenges, maintenance fluids, and drug infusions >100 mL throughout ICU stay (up to 90 days after enrolment).
    • Factorial co-intervention: participants were also randomised to a slower vs usual/control infusion rate strategy; analyses tested interaction between fluid type and infusion speed.
  • Comparison:
    • 0.9% saline supplied in indistinguishable coded 500 mL bags and used with the same indications/duration as the intervention group.
    • Non-trial fluids (including non-trial crystalloids) could be administered when clinically required (e.g., small-volume drug dilutions <100 mL and other non-protocol fluids), and pre-enrolment fluids were not controlled.
  • Blinding: Fluid type was blinded to clinicians, patients, and outcome assessors via coded, indistinguishable bags; the infusion-rate co-intervention was not feasibly blinded, but was randomised in a factorial design.
  • Statistics: A total sample size of 11,000 was planned to detect a hazard ratio of 0.90 for 90-day mortality (assuming 35% baseline mortality) with 89% power at a 5% two-sided significance level; primary analysis was modified intention-to-treat (randomised patients receiving ≥1 dose of trial fluid) using Cox proportional hazards modelling with random effect for site, reporting adjusted hazard ratios; secondary outcomes were not adjusted for multiplicity.
  • Follow-Up Period: 90 days after enrolment (including in-hospital follow-up and post-discharge contact where needed).

Key Results

This trial was not stopped early. Interim monitoring used a conservative stopping boundary; the trial completed planned enrolment with 11,052 randomised.

Outcome Balanced solution (Plasma-Lyte 148) 0.9% saline Effect p value / 95% CI Notes
90-day mortality (primary) 1381/5230 (26.4%) 1439/5290 (27.2%) Adjusted HR 0.97 95% CI 0.90 to 1.05; P=0.47 Modified intention-to-treat population (received ≥1 dose of trial fluid).
Kidney replacement therapy during hospital stay 393/5218 (7.5%) 427/5287 (8.1%) OR 0.93 95% CI 0.81 to 1.06; P=Not reported Prespecified secondary outcome; no multiplicity adjustment.
AKI (KDIGO stage ≥2) at day 3 850/3128 (27.2%) 859/3094 (27.8%) OR 0.99 95% CI 0.88 to 1.11; P=Not reported Assessed among those with creatinine/urine output data at day 3.
Total SOFA score at day 7 Median 4 (IQR 2 to 7); n=1531 Median 4 (IQR 2 to 7); n=1594 Absolute difference 0.27 95% CI 0.08 to 0.45; P=Not reported Day-7 SOFA available in a minority due to discharge/death before day 7.
Neurological SOFA >2 at day 7 492/1531 (32.1%) 415/1594 (26.0%) OR 1.40 95% CI 1.18 to 1.66; P=Not reported One of few secondary outcomes differing between groups; interpret cautiously given multiplicity.
Days not requiring mechanical ventilation within 28 days Median 27 (IQR 13 to 28) Median 27 (IQR 10 to 28) Absolute difference 0.14 days 95% CI −0.35 to 0.64; P=Not reported Prespecified secondary outcome; continuous outcome reported with CI for difference.
ICU mortality 906/5218 (17.4%) 922/5287 (17.4%) OR 1.01 95% CI 0.91 to 1.12; P=Not reported Tertiary outcome.
Hospital length of stay Median 8 days (IQR 5 to 18) Median 9 days (IQR 5 to 18) MR 0.98 95% CI 0.93 to 1.03; P=Not reported Tertiary outcome; MR = mean ratio.
Prespecified subgroup: traumatic brain injury (90-day mortality) 77/246 (31.3%) 50/237 (21.1%) HR 1.48 95% CI 1.03 to 2.12; interaction P=0.02 No difference in non–traumatic brain injury subgroup (interaction driven by TBI subgroup).
Unexpected treatment-related severe adverse events 0 reported 0 reported Not applicable Not reported No unexpected treatment-related severe adverse events were reported in either group.
  • Overall, balanced solution did not improve 90-day survival compared with saline (HR 0.97; 95% CI 0.90 to 1.05), and kidney outcomes (RRT/AKI) were similar.
  • A statistically significant interaction suggested higher mortality with balanced solution in traumatic brain injury (31.3% vs 21.1%; HR 1.48; 95% CI 1.03 to 2.12), warranting neurocritical care caution.
  • Exposure separation was incomplete: 67.6% vs 68.2% received IV fluids in the 24 hours pre-enrolment (median 1000 mL in each group), and 51.5% vs 52.8% received non-trial crystalloids on day 1 despite near-universal study fluid use.

Internal Validity

  • Randomisation and Allocation
    • Central web-based randomisation using permuted blocks (size 12), stratified by centre within a 2×2 factorial design.
    • Allocation concealment for fluid type was supported by indistinguishable coded fluid bags supplied for both arms.
  • Dropout / Exclusions (post-randomisation)
    • 11,052 randomised; 10,520 included in the modified intention-to-treat analysis (5230 balanced vs 5290 saline).
    • 532/11,052 (4.8%) were excluded after randomisation because trial fluid was not administered (predominantly lack of consent; some duplicate randomisation).
    • Lost to 90-day follow-up: 15 balanced vs 10 saline; primary outcome data were imputed for a small number with missing death status.
    • Implication: modified ITT and post-randomisation exclusions could bias estimates if exclusions were prognostically imbalanced, although reasons were similar in nature across groups.
  • Performance / Detection Bias
    • Fluid type was blinded to treating clinicians and outcome assessors; primary outcome (mortality) is objective and resistant to ascertainment bias.
    • The infusion-rate co-intervention could not be practically blinded; however, factorial randomisation balances this influence across fluid-type groups.
  • Protocol Adherence and Separation of the Variable of Interest
    • Study fluid exposure (day 1): 5200/5218 (99.7%) balanced vs 5276/5287 (99.8%) saline; median study fluid volume day 1 was 1500 mL (IQR 500 to 2500) vs 1500 mL (IQR 500 to 2000).
    • Non-trial crystalloids (day 1): 2688/5218 (51.5%) balanced vs 2794/5287 (52.8%) saline; median 500 mL (IQR 250 to 1000) in both groups.
    • Pre-enrolment fluids (within 24 hours): 3537/5230 (67.6%) balanced vs 3608/5290 (68.2%) saline; median total 1000 mL (IQR 500 to 2000) in both groups; >1000 mL in 44.6% vs 46.0%.
    • Serum chloride differed between groups over time (P<0.001), indicating some biochemical separation, but absolute values were not tabulated in the main results.
    • Implication: substantial pre-randomisation and concurrent non-trial crystalloid exposure plausibly diluted any chloride-mediated treatment effect.
  • Baseline Characteristics
    • Groups were well balanced at baseline: mean age 60.9 vs 61.2 years; women 44.4% vs 44.1%; sepsis 18.5% vs 19.2%; traumatic brain injury 4.7% vs 4.5%.
    • Severity of illness was moderate (APACHE II median 12 in both; total SOFA median 4 in both), with a large elective surgical cohort (planned admission 47.8% vs 49.0%).
    • Time from ICU admission to randomisation was early (median 0 days [IQR 0 to 1] in both groups), yet many had already received fluids before enrolment.
  • Heterogeneity, Timing, Dose, and Statistical Rigor
    • No evidence of interaction between fluid type and infusion rate for mortality (interaction P=0.98), supporting interpretation of the main fluid-type comparison within the factorial design.
    • Observed 90-day mortality (~27%) was lower than the 35% assumed for the power calculation, and analysed sample size (10,520) was modestly below the planned 11,000, both of which may reduce statistical power for small effects.
    • Secondary outcomes were numerous and not adjusted for multiplicity; two secondary findings (day-7 total SOFA and neurological SOFA >2 at day 7) should be interpreted cautiously.
  • Outcome Assessment
    • Primary outcome: 90-day mortality (objective), with near-complete ascertainment and limited imputation.
    • Kidney outcomes were defined using standard KDIGO staging and kidney replacement therapy events; SOFA components are standardised but require available day-specific data (substantial missingness by day 7 due to discharge/death).

Conclusion on Internal Validity: Overall, internal validity for the primary mortality comparison appears moderate-to-strong given robust randomisation, effective blinding for fluid type, and objective primary outcome; key threats include modified intention-to-treat exclusions after randomisation and incomplete exposure separation due to pre-enrolment and non-trial crystalloid use.

External Validity

  • Population Representativeness
    • Represents a broad mixed ICU population at risk of AKI, but with a high proportion of planned surgical admissions and overall moderate illness severity.
    • Patients needing imminent kidney replacement therapy, those with extreme sodium derangements, and those expected to die within 24 hours were excluded; effects may differ in more extreme phenotypes (e.g., profound septic shock with high-volume resuscitation).
    • Conducted in 75 Brazilian ICUs; applicability depends on similarity of ICU case-mix, fluid practices, and resource availability.
  • Applicability
    • Directly applicable to settings where Plasma-Lyte 148 and 0.9% saline are available and routinely used for ICU maintenance/resuscitation and larger-volume drug infusions.
    • Generalisation to other “balanced” solutions (e.g., lactated Ringer’s) is plausible but not directly tested here.
    • Neurocritical care: the traumatic brain injury interaction suggests findings should not be extrapolated uncritically to intracranial pathology populations.

Conclusion on External Validity: Overall generalisability is good for mixed medical–surgical ICU patients receiving modest-to-moderate crystalloid volumes, but is more limited for neurocritical care (traumatic brain injury) and for scenarios involving very large-volume resuscitation or materially different baseline fluid practices.

Strengths & Limitations

  • Strengths:
    • Large, multicentre ICU trial (10,520 analysed) with patient-level randomisation and pragmatic implementation.
    • Effective blinding for fluid type, reducing performance and detection bias for the key comparison.
    • Patient-centred primary endpoint (90-day mortality) with high completeness of follow-up.
    • Factorial design efficiently tested two clinically important questions and enabled interaction testing.
  • Limitations:
    • Modified intention-to-treat with 4.8% excluded after randomisation (mostly consent not obtained), which may introduce selection bias.
    • Incomplete separation of exposure due to common pre-enrolment fluids and frequent non-trial crystalloids during ICU stay.
    • Overall illness severity and crystalloid exposure were modest; if benefits accrue mainly in high-volume resuscitation phenotypes, the trial may underestimate them.
    • Multiple secondary outcomes without multiplicity adjustment and substantial missingness for day-7 SOFA measures due to discharge/death.
    • Signal of harm in traumatic brain injury subgroup requires careful interpretation and may not be fully mechanistically resolved within the trial dataset.

Interpretation & Why It Matters

  • Clinical practice
    • In a broad ICU population where clinicians deemed either fluid acceptable, Plasma-Lyte 148 did not reduce 90-day mortality or kidney endpoints compared with 0.9% saline.
    • The confidence interval around mortality is narrow enough to make a large survival advantage unlikely in the average patient, shifting the debate towards small effects and subgroup tailoring.
  • Neurocritical care caution
    • The traumatic brain injury interaction (higher mortality with balanced solution) supports avoiding routine use of Plasma-Lyte–type balanced crystalloids in TBI until further corroboration and mechanistic clarity.
  • Evidence integration
    • BaSICS provides high-quality, blinded RCT evidence that complements earlier cluster-crossover trials and later mega-trials, implying that average mortality effects of balanced vs saline are small and context dependent.

Controversies & Subsequent Evidence

Summary

  • In 10,520 ICU patients, balanced solution (Plasma-Lyte 148) did not reduce 90-day mortality compared with 0.9% saline (26.4% vs 27.2%; adjusted HR 0.97; 95% CI 0.90 to 1.05; P=0.47).
  • Key kidney outcomes were similar (RRT in hospital 7.5% vs 8.1%; OR 0.93; 95% CI 0.81 to 1.06; KDIGO stage ≥2 at day 3: 27.2% vs 27.8%; OR 0.99; 95% CI 0.88 to 1.11).
  • A prespecified subgroup interaction suggested higher mortality with balanced solution in traumatic brain injury (31.3% vs 21.1%; HR 1.48; 95% CI 1.03 to 2.12; interaction P=0.02), while non-TBI patients had no difference.
  • Exposure separation was incomplete: ~68% received pre-enrolment fluids (median 1000 mL in the prior 24 hours), and ~52% received non-trial crystalloids on ICU day 1 despite near-universal study fluid use.
  • BaSICS, together with later ICU trials and meta-analyses, supports that average mortality effects of balanced vs saline are small and may be phenotype dependent, particularly for neurocritical care.

Overall Takeaway

BaSICS is a landmark large, blinded ICU RCT showing that, for a broad mixed ICU population in whom clinicians considered either fluid appropriate, Plasma-Lyte 148 did not improve 90-day survival or kidney outcomes compared with 0.9% saline. The precision of its neutral estimate, reinforced by subsequent mega-trials, indicates that any average mortality benefit of balanced crystalloids is likely modest at most; however, the traumatic brain injury interaction highlights that subgroup safety and context should guide fluid choice rather than assuming a uniform class effect.

Overall Summary

  • Balanced crystalloid (Plasma-Lyte 148) was not superior to 0.9% saline for 90-day mortality in a large ICU RCT, with a clinically important signal of potential harm in traumatic brain injury.

Bibliography