Skip to main content

Foundational Trials

BASICS - Infusion Rate

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effect of Slower vs Faster Intravenous Fluid Bolus Rates 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 slower vs faster intravenous fluid bolus rates on mortality in critically ill patients: the BaSICS randomized clinical trial. JAMA. 2021;326(9):830-838.

Context & Rationale

  • Background
    • Intravenous fluid boluses (“fluid challenges”) are a foundational ICU intervention, yet most high-quality evidence addresses fluid type or fluid volume, not the rate of administration.
    • Theoretic mechanisms cut both ways: faster boluses may improve macrocirculatory variables rapidly, but may also increase venous pressures, endothelial injury, haemodilution, or precipitate pulmonary oedema; slower boluses may enable earlier reassessment and avert “overshoot”.
    • International practice variation is substantial: in the FENICE inception cohort, the 25th percentile infusion rate for a fluid challenge was ~500 mL/h, highlighting uncertainty about clinical consequences of slower rates.1
    • Before BaSICS-rate, evidence consisted largely of small physiologic studies and perioperative extrapolation, with no large pragmatic ICU RCT powered for patient-centred outcomes.
  • Research Question/Hypothesis
    • In critically ill ICU patients receiving crystalloid fluid challenges, does a slower bolus infusion rate (333 mL/h) compared with a faster rate (999 mL/h) improve 90-day survival and reduce organ dysfunction/AKI?
    • Prespecified effect target in planning: hazard ratio ~0.90 for 90-day mortality (two-sided α 0.05).2
  • Why This Matters
    • Infusion rate is a low-cost, highly scalable “behavioural” intervention (pump setting), making any demonstrable benefit immediately implementable.
    • Clarifies whether routine “rapid bolus” practice is evidence-based, or whether the resuscitation emphasis should remain on indication, reassessment, and avoidance of unnecessary fluid.
    • Provides an empirical anchor for guideline statements that recommend early fluids but rarely specify bolus rate.

Design & Methods

  • Research Question: Among ICU patients receiving fluid challenges, does a slower bolus infusion rate (333 mL/h) vs a faster rate (999 mL/h) reduce 90-day mortality and improve kidney/organ outcomes?
  • Study Type: Randomised, multicentre, investigator-initiated, pragmatic 2×2 factorial trial in 75 ICUs in Brazil; infusion-rate allocation open-label; concurrent fluid-type allocation double-blind.
  • Population:
    • Setting: critically ill patients in participating ICUs in Brazil (enrolment May 29, 2017 to March 2, 2020; follow-up completed October 29, 2020).
    • Key inclusion: clinician decision to administer a crystalloid fluid challenge for volume expansion; ≥1 risk factor for worse outcomes (risk-factor list not fully enumerated in the JAMA manuscript).
    • Key exclusions (reported): serum sodium <120 or >160 mmol/L; chloride <85 or >115 mmol/L; potassium <2.0 or >6.0 mmol/L; anticipated death within 24 hours; concurrent enrolment in another clinical trial.
    • Randomisation stratified by centre and by the other factorial assignment (fluid type).
  • Intervention:
    • Slower bolus-rate strategy: each fluid challenge delivered at 333 mL/h (typically 500 mL over ~90 minutes) using an infusion pump.
    • Fluid challenge volume: 500 mL of the assigned crystalloid per episode (fluid type determined by the other factorial allocation).
    • Safety allowance: clinicians could administer faster than assigned rate in emergencies (frequency not reported).
  • Comparison:
    • Faster bolus-rate strategy (control for the rate comparison): each fluid challenge delivered at 999 mL/h (typically 500 mL over ~30 minutes) using an infusion pump.
    • Same fluid challenge volume (500 mL) and same emergency discretion to deviate if clinically required.
  • Blinding: Infusion rate was unblinded (pragmatic pump-setting intervention); fluid type was double-blind (identical masked study fluids). Primary outcome (mortality) is objective; clinician-driven secondary outcomes (e.g., KRT initiation) are potentially performance-biased.
  • Statistics: Planning assumptions: 11,000 patients to detect HR 0.90 for 90-day mortality (assumed 35% mortality) with 89% power at two-sided α 0.05; interim analyses included an early 1,000-patient safety review and efficacy looks at 25%, 50%, and 75% recruitment (per published SAP).2 Primary analysis used mixed-effects Cox models with site as random effect, adjusted for age, baseline SOFA, and admission type; modified intention-to-treat excluding post-randomisation lack of consent and duplicate enrolments; multiple imputation for missing primary outcome data (15 patients).
  • Follow-Up Period: 90 days for primary outcome; organ dysfunction and ventilator outcomes assessed at prespecified early timepoints (days 3 and 7) and up to 28 days (ventilation-free days).

Key Results

This trial was not stopped early. Recruitment and follow-up were completed as planned.

Outcome Slower bolus rate (333 mL/h) Faster bolus rate (999 mL/h) Effect p value / 95% CI Notes
90-day mortality 1406/5276 (26.6%) 1414/5244 (27.0%) HR 1.03 95% CI 0.96 to 1.11; P=0.46 Primary end point; mixed-effects Cox model
Acute kidney failure requiring kidney replacement within 90 days (rate) 414/474.84 (0.87) 445/471.96 (0.94) RR 0.95 95% CI 0.83 to 1.09 Secondary end point; denominator is follow-up time (thousands of patient-days); RR is rate ratio
In-hospital kidney replacement therapy 397/5267 (7.5%) 423/5238 (8.1%) OR 0.92 95% CI 0.80 to 1.06 Secondary end point (in-hospital)
Day 3 SOFA total score Median 4 (IQR 2 to 6) Median 4 (IQR 2 to 7) −0.10 95% CI −0.21 to −0.01 Secondary end point; model-based effect reported as absolute effect
Day 3 cardiovascular SOFA component >2 1252/3847 (32.5%) 1338/3788 (35.3%) OR 0.89 95% CI 0.80 to 0.98 Component analysis; direction favours slower rate
Day 3 coagulation SOFA component >2 180/3847 (4.7%) 146/3788 (3.9%) OR 1.31 95% CI 1.05 to 1.64 Component analysis; direction favours faster rate
Mechanical ventilation–free days within 28 days Median 27 (IQR 18 to 28) Median 27 (IQR 17 to 28) 0.17 95% CI −0.07 to 0.32 Tertiary end point
ICU length of stay Median 3 (IQR 2 to 7) Median 3 (IQR 2 to 7) MR 0.98 95% CI 0.93 to 1.03 Tertiary end point; MR is mean ratio
  • Mortality was essentially identical: 26.6% vs 27.0% at 90 days (HR 1.03; 95% CI 0.96 to 1.11; P=0.46), excluding a clinically large benefit or harm from slowing boluses within this pump-limited range.
  • Kidney outcomes were similar: KRT within 90 days showed RR 0.95 (95% CI 0.83 to 1.09) with consistent in-hospital estimates (OR 0.92; 95% CI 0.80 to 1.06).
  • SOFA signals were small and directionally mixed across components (cardiovascular OR 0.89; respiratory OR 0.81; coagulation OR 1.31 at day 3), limiting mechanistic certainty.

Internal Validity

  • Randomisation and allocation concealment: Central web-based randomisation; random permuted blocks of 12; stratified by centre and the other factorial assignment (fluid type), supporting robust allocation concealment prior to assignment.
  • Post-randomisation exclusions: Of 11,052 randomised, 10,520 were included; exclusions were for lack of consent (n=486 across factorial strata) and duplicate enrolment (n=46), producing a modified intention-to-treat population and potential (though likely small) selection bias.
  • Loss to follow-up / missingness: Vital status at 90 days missing in 25 patients (14 slower, 11 faster); primary outcome imputed for 15 patients; discharge information missing for 5 per group.
  • Performance/detection bias: Infusion rate was open-label; primary outcome (mortality) objective, but kidney replacement initiation and ventilator practices are clinician-mediated and could be influenced by group awareness.
  • Protocol adherence and separation: Exposure contrast defined by pump rates (333 vs 999 mL/h) and typical 500 mL bolus duration (~90 vs ~30 minutes); emergency deviations permitted (frequency not reported). Mean fluid volume on day 1 differed modestly (1162 mL slower vs 1252 mL faster; absolute difference 90 mL), suggesting only limited downstream separation in cumulative early fluid exposure.
  • Baseline comparability: Groups were closely matched (median age 61; APACHE II median 17; baseline SOFA median 4; sepsis ~26%; invasive ventilation ~38%; vasopressors ~13%).
  • Heterogeneity and factorial structure: 75 ICUs and mixed admission types increase pragmatic relevance but introduce practice heterogeneity; mixed-effects modelling with site random effects partially addresses this; no evidence of interaction between infusion rate and fluid type (interaction P=0.98).
  • Timing and dose adequacy: Intervention applied at the time fluid challenges were ordered in ICU (not ED/prehospital); “faster” rate (999 mL/h) is pump-limited and may not represent truly rapid bolus delivery by pressure bag, potentially attenuating biologic separation.
  • Outcome assessment and analysis fidelity: Prespecified outcomes and modelling framework were published in advance (SAP); adjustment set and mixed-effects strategy are appropriate for clustered multicentre data and time-to-event mortality.

Conclusion on Internal Validity: Overall internal validity is moderate-to-strong for the mortality comparison (robust randomisation; low missingness; objective primary outcome), but is weaker for clinician-mediated secondary outcomes due to open-label rate allocation, post-randomisation consent exclusions, and likely modest separation in actual delivered fluid exposure.

External Validity

  • Population representativeness: Large, pragmatic ICU cohort (Brazil) with broad case-mix (planned surgery ~44%, unplanned non-sepsis ~29%, sepsis ~26%), moderate severity (APACHE II 17), making it relevant to many general ICUs.
  • Applicability to other systems: Likely applicable where fluid challenges are commonly delivered by infusion pump within similar timeframes; may be less applicable in settings where boluses are delivered by pressure bag/manual push (much faster than 999 mL/h), or where early shock resuscitation occurs primarily in ED/prehospital environments.
  • Intervention generalisability: The tested contrast (333 vs 999 mL/h) reflects a pragmatic pump-limited range; does not directly answer whether “very rapid” boluses or “very slow” infusions change outcomes, nor whether rate matters during the earliest hours of septic shock outside ICU.

Conclusion on External Validity: Findings are highly generalisable to ICU fluid challenges delivered by pump over ~30–90 minutes, but are less generalisable to early resuscitation contexts and to bolus delivery speeds outside the tested range.

Strengths & Limitations

  • Strengths: Very large sample; multicentre pragmatic ICU design; clinically meaningful objective primary outcome; rigorous mixed-effects modelling accounting for centre clustering; factorial design allowed assessment of interaction with fluid type; minimal missing primary outcome data.
  • Limitations: Open-label infusion-rate allocation; post-randomisation exclusions for lack of consent and duplicate enrolments (532/11,052); modest separation in early cumulative fluid exposure (day 1 volume difference 90 mL); emergency deviations permitted but frequency not reported; “faster” arm limited to pump rate (999 mL/h) and may not represent common pressure-bag bolus practice; intervention begins in ICU (not early ED resuscitation).

Interpretation & Why It Matters

  • Clinical implication
    • Within a pragmatic pump-based range (~30 vs ~90 minutes per 500 mL bolus), slowing fluid challenges did not reduce 90-day mortality and produced no convincing kidney benefit.
    • The results support prioritising indication, reassessment, and avoidance of unnecessary fluids over routine manipulation of pump rate within this range.
  • Mechanistic interpretation
    • Small, mixed SOFA component differences at day 3 and only modest divergence in day-1 fluid volume suggest any rate effect (if present) is subtle and potentially mediated by clinical reassessment rather than purely haemodynamic kinetics.
  • Methodological significance
    • BaSICS-rate is one of the first very large ICU RCTs to randomise a process-of-care parameter (infusion speed) rather than a drug/device, illustrating both feasibility and the challenges of exposure separation in pragmatic settings.

Controversies & Subsequent Evidence

  • The accompanying editorial concluded that neither crystalloid composition nor infusion rate meaningfully affected 90-day mortality in BaSICS, and that the small day-3 SOFA signal has uncertain clinical relevance given mixed component directions.3
  • The tested contrast (333 vs 999 mL/h) reflects infusion-pump constraints and may not address the common real-world scenario of pressure-bag/manual “rapid bolus”, raising the possibility of biologic under-separation for questions about truly rapid infusion.
  • Observed practice variation in FENICE underscores the rationale for the trial but also highlights that “rate” is entangled with clinician reassessment and subsequent fluid decisions, complicating interpretation of causal pathways.1
  • The companion BaSICS fluid-type report found no overall mortality difference between balanced solution and saline, with a signal of harm in traumatic brain injury, reinforcing that “fluid strategy” effects may be context-dependent and that rate findings should not be extrapolated to all phenotypes without caution.4
  • Subsequent major resuscitation trials have continued to emphasise targets and fluid vs vasopressor balance rather than bolus rate; ANDROMEDA-SHOCK, for example, showed no mortality benefit from peripheral perfusion–targeted resuscitation compared with lactate targeting, illustrating ongoing uncertainty about which modifiable fluid-related behaviours translate into outcome benefit.5
  • Guidelines (e.g., Surviving Sepsis Campaign 2016) emphasise early fluids and reassessment but provide limited specific direction on infusion rate, consistent with BaSICS-rate demonstrating equipoise within common pump-based rates.6

Summary

  • In 10,520 ICU patients, slowing fluid boluses to 333 mL/h did not improve 90-day survival compared with 999 mL/h (26.6% vs 27.0%; HR 1.03; 95% CI 0.96 to 1.11).
  • Kidney outcomes were similar, including KRT within 90 days (RR 0.95; 95% CI 0.83 to 1.09) and in-hospital KRT (OR 0.92; 95% CI 0.80 to 1.06).
  • Day 3 SOFA total score differed minimally (−0.10; 95% CI −0.21 to −0.01) with mixed component directions, limiting mechanistic interpretation.
  • Open-label rate allocation and modest separation in early cumulative fluid exposure are key constraints, but objective mortality ascertainment and large scale support robust inference for the primary outcome.
  • The findings suggest bolus infusion rate (within 30–90 min per 500 mL) is a lower-yield target than decisions about whether to give a bolus and how to reassess response.

Overall Takeaway

BaSICS-rate is a landmark pragmatic ICU trial showing that, within a pump-based range (~30 vs ~90 minutes for a 500 mL crystalloid bolus), slowing fluid challenges does not improve 90-day survival and does not meaningfully change kidney outcomes. Its principal contribution is to de-emphasise infusion rate as a major driver of outcome in routine ICU fluid challenges and to refocus clinicians on judicious bolus indication, reassessment, and minimising avoidable fluid exposure.

Overall Summary

  • In a large ICU cohort, bolus infusion rate (333 vs 999 mL/h) did not alter 90-day mortality, supporting equipoise for pump-delivered boluses within this range.

Bibliography