Skip to main content

Foundational Trials

CHEST 2000

Image: Shutterstock / Jub-Job

Hydroxyethyl starch fluid hanging from a drip stand

Publication

  • Title: Hydroxyethyl starch or saline for fluid resuscitation in intensive care
  • Acronym: CHEST (Crystalloid versus Hydroxyethyl Starch Trial)
  • Year: 2012
  • Journal published in: The New England Journal of Medicine
  • Citation: Myburgh JA, Finfer S, Bellomo R, et al. Hydroxyethyl starch or saline for fluid resuscitation in intensive care. N Engl J Med. 2012;367:1901-11.

Context & Rationale

  • Background
    • Intravenous fluid resuscitation is a core therapy in critical illness, yet the optimal resuscitation fluid (crystalloid vs colloid) remained uncertain.
    • Hydroxyethyl starch (HES) solutions were widely used to achieve haemodynamic targets with less volume than crystalloids, but concerns persisted regarding renal toxicity, coagulopathy, tissue storage, and pruritus.
    • Earlier generations of starches and sepsis-focused trials raised safety signals (acute kidney injury and mortality), while “tetrastarch” formulations (e.g., HES 130/0.4) were promoted as safer despite limited large-scale, blinded ICU outcome data.
    • Practice variation and high population exposure meant that even small differences in patient-centred outcomes would have major clinical and health-system implications.
  • Research Question/Hypothesis
    • In adult ICU patients requiring fluid resuscitation, does using 6% HES 130/0.4 (in saline) compared with 0.9% saline for resuscitation affect 90-day all-cause mortality?
    • Key secondary focus: renal outcomes (need for renal-replacement therapy; acute kidney injury by RIFLE) and new organ failures.
  • Why This Matters
    • Fluid choice is a near-universal exposure in ICU care; safety and effectiveness evidence must be anchored in patient-centred outcomes (mortality and major morbidity), not surrogate haemodynamic endpoints alone.
    • A pragmatic, blinded comparison of a commonly used “modern” starch versus a standard crystalloid was necessary to inform clinical practice, guideline recommendations, and regulatory decisions.
    • The trial also tested the feasibility of large, multicentre, double-blind fluid trials with robust follow-up and clinically relevant endpoints.

Design & Methods

  • Research Question: Among adult ICU patients requiring fluid resuscitation, does resuscitation with 6% HES 130/0.4 (in saline) versus 0.9% saline change 90-day all-cause mortality (and renal/organ failure outcomes)?
  • Study Type: Multicentre, randomised, double-blind, parallel-group, investigator-initiated pragmatic trial in 32 adult ICUs in Australia and New Zealand (recruitment Dec 2009 to Jan 2012); web-based randomisation using minimisation (stratified by site and trauma diagnosis).
  • Population:
    • Setting: Adult mixed medical-surgical ICUs (Australia/New Zealand); enrolment when treating clinicians judged fluid resuscitation was required.
    • Key inclusion criteria: Age ≥18 years; ICU admission; clinician judged intravascular fluid resuscitation was required; clinician considered both study fluids appropriate options; at least one prespecified indicator of hypovolaemia or impaired perfusion (e.g., tachycardia >90 beats/min; systolic blood pressure <100 mm Hg or mean arterial pressure <75 mm Hg; low filling pressures where measured; dynamic indices consistent with fluid responsiveness where used; prolonged capillary refill; low urine output).
    • Key exclusion criteria: Traumatic intracranial haemorrhage; receipt of renal-replacement therapy or imminent requirement; severe renal dysfunction (including creatinine ≥350 μmol/L with oliguria); severe hypernatraemia or hyperchloraemia; pregnancy/breastfeeding or inability to exclude pregnancy in women of childbearing potential; ICU admission after cardiac surgery; burns or liver transplantation; expected death/imminent death or limitations of treatment; substantial HES exposure prior to randomisation (protocol threshold); prior enrolment; transfer from another ICU with resuscitation fluids already administered there.
  • Intervention:
    • 6% hydroxyethyl starch 130/0.4 in saline (HES 130/0.4), supplied in indistinguishable 500-mL study bags.
    • Used for all fluid resuscitation episodes while the patient remained in ICU (until ICU discharge, death, or day 90 after randomisation, per trial procedures).
    • Daily dose cap: maximum 50 mL/kg of study HES per 24 hours; if this limit was reached, subsequent resuscitation fluid defaulted to open-label 0.9% saline.
    • Bolus size and rate of administration were clinician-directed (pragmatic delivery), within dosing restrictions.
  • Comparison:
    • 0.9% sodium chloride (normal saline) in indistinguishable 500-mL study bags.
    • Used for all ICU resuscitation episodes (no protocol-imposed daily volume cap on saline).
    • Other fluids (e.g., maintenance fluids, nutrition-related fluids, blood products) were permitted as clinically indicated; protocol adherence assessed via prespecified violation reporting and non-study fluid tracking.
  • Blinding: Double-blind (patients, treating clinicians, investigators, outcome assessors); study fluids packaged and labelled to be indistinguishable; allocation via central web-based system.
  • Statistics: A total sample size of 7000 patients was planned to detect an absolute difference of 3.5 percentage points in 90-day mortality (baseline mortality estimate 26%) with 90% power at the 5% significance level, allowing for 5% loss to follow-up; primary analyses were intention-to-treat with relative risks for binary outcomes and prespecified subgroup interaction testing; Kaplan–Meier/log-rank for time-to-event mortality.
  • Follow-Up Period: 90 days after randomisation (primary endpoint), with secondary outcomes including ICU/hospital course, organ failure, and renal endpoints.

Key Results

This trial was not stopped early. Recruitment completed at the planned sample size (n=7000); prespecified interim analyses did not trigger stopping.

Outcome 6% HES 130/0.4 0.9% Saline Effect p value / 95% CI Notes
Death at 90 days (primary) 597/3315 (18.0%) 566/3336 (17.0%) RR 1.06 95% CI 0.96 to 1.18; P=0.26 Primary outcome data available in 94.7% vs 95.3% of randomised patients.
Death within 28 days 416/3315 (12.5%) 393/3336 (11.8%) RR 1.06 95% CI 0.92 to 1.22; P=0.44 No early survival advantage.
Renal-replacement therapy (during 90 days) 235/3352 (7.0%) 196/3375 (5.8%) RR 1.21 95% CI 1.00 to 1.45; P=0.04 Denominators reflect patients who received any study fluid after randomisation.
Acute kidney injury (RIFLE-R) 1788/3309 (54.0%) 1912/3335 (57.3%) RR 0.94 95% CI 0.90 to 0.98; P=0.007 Composite definition incorporates serum creatinine and urine output criteria.
Acute kidney injury (RIFLE-I) 1130/3265 (34.6%) 1253/3300 (38.0%) RR 0.91 95% CI 0.85 to 0.97; P=0.005 Composite definition incorporates serum creatinine and urine output criteria.
Acute kidney injury (RIFLE-F) 336/3243 (10.4%) 301/3263 (9.2%) RR 1.12 95% CI 0.97 to 1.30; P=0.12 Failure category did not differ significantly by RIFLE composite criteria.
New cardiovascular failure 663/1815 (36.5%) 722/1808 (39.9%) RR 0.91 95% CI 0.84 to 0.99; P=0.03 New organ failure defined by SOFA criteria in those without baseline failure.
New hepatic failure 55/2830 (1.9%) 36/2887 (1.2%) RR 1.56 95% CI 1.03 to 2.36; P=0.03 Absolute event rates were low.
Duration of ICU stay (days) 7.3 ± 0.2 6.9 ± 0.2 Mean diff 0.4 95% CI -0.0 to 0.9; P=0.07 Values are mean ± SE.
Any treatment-related adverse event 180/3871 (4.6%) 95/2879 (3.3%) Not reported P=0.006 Safety denominators reflect HES exposure grouping (not the ITT randomised set).
Serious treatment-related adverse event 2/3871 (0.1%) 2/2879 (0.1%) Not reported P=0.77 Serious events were rare in both groups.
  • Mortality: No significant difference in 90-day mortality (18.0% vs 17.0%; RR 1.06; 95% CI 0.96 to 1.18; P=0.26).
  • Renal safety signal: More patients required renal-replacement therapy with HES (7.0% vs 5.8%; RR 1.21; 95% CI 1.00 to 1.45; P=0.04), despite lower RIFLE-R and RIFLE-I event rates using composite criteria.
  • Prespecified subgroup consistency: No evidence of heterogeneity for 90-day mortality across prespecified subgroups (e.g., sepsis: 25.4% vs 23.7%; RR 1.07; 95% CI 0.92 to 1.25; traumatic brain injury: 3.7% vs 10.0%; RR 0.37; 95% CI 0.04 to 3.35).

Internal Validity

  • Randomisation and allocation concealment:
    • Centralised, encrypted, web-based randomisation with a minimisation algorithm; stratified by study site and trauma diagnosis.
    • Allocation concealment and double-blinding supported by indistinguishable study fluid packaging and labelling.
  • Drop out / exclusions after randomisation:
    • Randomised: 3500 per group.
    • Withdrawal of consent: 137/3500 (3.9%) in HES vs 113/3500 (3.2%) in saline.
    • Primary outcome (90-day mortality) ascertainment: 3315/3500 (94.7%) in HES vs 3336/3500 (95.3%) in saline.
    • Loss to follow-up reported as rare (0.1% per group overall), supporting low attrition bias for the primary endpoint.
  • Performance and detection bias:
    • Blinding was feasible and implemented; primary endpoint (all-cause mortality) is objective and minimally susceptible to assessor bias.
    • Renal-replacement therapy initiation is clinician-mediated (practice variation possible), but decision-making occurred under blinding conditions across multiple centres.
  • Protocol adherence and treatment separation:
    • Protocol violations: 319/3358 (9.5%) in HES vs 321/3384 (9.5%) in saline; most were “incorrect study fluid administration”.
    • Daily study fluid exposure (first 4 days): 526 ± 425 mL/day in HES vs 616 ± 488 mL/day in saline (P<0.001).
    • Daily non-study fluid exposure (first 4 days): 851 ± 675 mL/day in HES vs 1115 ± 993 mL/day in saline (P<0.001).
    • Net fluid balance (first 4 days): 921 ± 1069 mL in HES vs 982 ± 1161 mL in saline (P=0.03).
    • Central venous pressure (first 4 days): 11.3 ± 4.8 mm Hg in HES vs 10.4 ± 4.4 mm Hg in saline (P<0.001), consistent with measurable haemodynamic separation.
  • Baseline characteristics and illness severity:
    • Groups were closely balanced for demographics and severity (APACHE II median 17 [IQR 12–22] vs 17 [12–23]).
    • Small baseline imbalances were present (e.g., CVP 9.5 ± 5.4 vs 8.9 ± 5.1 mm Hg; lactate 2.1 ± 2.0 vs 2.0 ± 1.5 mmol/L), unlikely to fully explain outcome differences in a trial of this size.
    • Predefined subgroup strata were well represented (e.g., sepsis at randomisation 29.2% vs 28.4%; baseline RIFLE-defined AKI 36.0% vs 36.0%).
  • Timing and dose considerations:
    • Mean time from ICU admission to randomisation was ~11 hours with wide dispersion (10.9 ± 156.5 hours vs 11.4 ± 165.4 hours), reflecting pragmatic enrolment when resuscitation was clinically indicated rather than a tightly standardised “early shock” window.
    • Maximum HES dose was capped at 50 mL/kg/day; observed mean daily study fluid volumes suggest many patients received moderate rather than extreme starch exposure, anchoring inference to “typical ICU resuscitation” use.
  • Outcome assessment and endpoint structure:
    • Primary endpoint was all-cause mortality at 90 days (objective).
    • Renal endpoints used both clinician-driven therapy (RRT) and composite AKI staging (RIFLE) incorporating urine output and creatinine, which can move in different directions when fluids alter haemodynamics and dilution.
  • Statistical rigour:
    • Prespecified power, interim monitoring, and intention-to-treat analysis were used; effect estimates were reported as relative risks with 95% confidence intervals.
    • Adjusted analyses (including prespecified baseline covariates and study site effects) were consistent with primary analyses for key outcomes (mortality and renal endpoints).

Conclusion on Internal Validity: Overall, internal validity appears strong given robust randomisation and blinding, high follow-up completeness for the primary outcome, measurable treatment separation, and consistent results across prespecified and adjusted analyses; renal endpoint interpretation is intrinsically complex because of composite definitions and clinician-mediated RRT initiation.

External Validity

  • Population representativeness:
    • Broad adult ICU cohort with mixed medical and surgical admissions (surgical ~42.5% vs 42.9%); moderate illness severity (APACHE II median 17).
    • Substantial representation of sepsis (~29%) and trauma (~8%), supporting applicability to common ICU phenotypes.
  • Key exclusions limiting generalisability:
    • Patients with traumatic intracranial haemorrhage, burns, post-cardiac surgery, liver transplantation, pregnancy/breastfeeding, and severe renal dysfunction (including ongoing RRT or creatinine ≥350 μmol/L with oliguria).
    • Patients with severe hypernatraemia/hyperchloraemia were excluded, which may limit direct extrapolation to extreme electrolyte derangements.
  • Applicability to contemporary practice:
    • Control fluid was 0.9% saline; extrapolation to balanced crystalloids requires caution (both trial arms were saline-based, but total chloride exposure likely differed because of volume differences).
    • Conducted in high-resource ICUs with established research infrastructure; core findings remain applicable to similar systems, but the magnitude of effect could vary where practice patterns for RRT initiation or fluid strategies differ.

Conclusion on External Validity: Generalisability is moderate-to-high for adult mixed ICU populations requiring resuscitation in high-resource settings, but is limited for excluded groups (notably severe renal dysfunction, burns, post-cardiac surgery, and traumatic intracranial haemorrhage) and for settings where balanced crystalloids are the dominant comparator.

Strengths & Limitations

  • Strengths:
    • Large, adequately powered, multicentre randomised trial (n=7000) in a heterogeneous ICU population.
    • Double-blind design with indistinguishable study fluids and central randomisation, reducing performance and detection bias.
    • Patient-centred primary outcome (90-day all-cause mortality) with high follow-up completeness.
    • Pragmatic delivery reflecting real-world resuscitation practice, improving clinical relevance.
    • Renal outcomes assessed using both therapy-based endpoints (RRT) and structured AKI staging (RIFLE).
  • Limitations:
    • Comparator was 0.9% saline rather than a balanced crystalloid, limiting inference about starch relative to contemporary low-chloride crystalloid strategies.
    • Resuscitation triggers, bolus sizes, and co-interventions were clinician-directed (pragmatic), creating variability in exposure (“dose”) that may attenuate detectable treatment effects.
    • Daily HES cap (50 mL/kg) and observed mean daily study fluid volumes suggest that very high-dose starch exposure was uncommon; results may not fully capture extreme-dose toxicity.
    • Renal endpoints are challenging to interpret mechanistically because urine-output criteria and creatinine-based criteria can diverge when fluids alter haemodynamics and dilution.
    • Safety reporting required later correction of adverse-event denominators and table footnotes, although this did not affect primary outcome results.

Interpretation & Why It Matters

  • Clinical practice
    • HES 130/0.4 did not improve survival and was associated with higher use of renal-replacement therapy and more treatment-related adverse events.
    • Any potential fluid-sparing or haemodynamic advantages did not translate into patient-centred benefit and were accompanied by kidney safety concerns.
  • Renal endpoint interpretation
    • The trial highlights that composite AKI definitions (combining urine output and creatinine) may behave counterintuitively when an intervention changes urine output and volume status, underscoring the importance of therapy-based endpoints (e.g., RRT) and transparent component reporting.
  • Methodological legacy
    • CHEST demonstrated feasibility of large, blinded, pragmatic ICU fluid trials with robust follow-up, providing a template for subsequent foundational fluid and resuscitation trials.

Controversies & Subsequent Evidence

  • Renal endpoint discordance: Published correspondence highlighted the apparent mismatch between lower RIFLE-R/I rates yet higher renal-replacement therapy in the HES group; the authors’ reply emphasised creatinine-based signals, intention-to-treat inference, and the borderline nature of the RRT confidence interval (with rounding at the lower bound). 1
  • Safety reporting correction: A formal correction clarified adverse-event numerators/denominators and footnote interpretation for “days receiving” organ support measures in Table 2, without altering primary outcome conclusions. 2
  • Reanalysis and transparency: A subsequent reanalysis of the CHEST trial dataset was published as correspondence, reinforcing the absence of mortality benefit while re-examining renal safety signals. 3
  • Longer-term outcomes and value: Planned follow-up economic and quality-of-life analysis did not demonstrate benefit from HES and suggested unfavourable cost-effectiveness (i.e., higher costs without improvement in longer-term outcomes). 4
  • Concordant RCT and meta-analytic evidence: A contemporaneous large sepsis RCT (6S) and subsequent meta-analysis supported a signal of harm with HES in critical illness (mortality and renal outcomes), strengthening the inference that “modern” tetrastarch does not confer a safety advantage in this population. 56
  • Guideline synthesis: International guidelines incorporated CHEST and related evidence to recommend against HES use for resuscitation in sepsis and critical illness, favouring crystalloids as first-line resuscitation fluids. 78

Summary

  • In 7000 adult ICU patients requiring resuscitation, 6% HES 130/0.4 did not reduce 90-day mortality compared with 0.9% saline (18.0% vs 17.0%; RR 1.06; 95% CI 0.96 to 1.18).
  • HES was associated with a higher proportion of patients receiving renal-replacement therapy (7.0% vs 5.8%; RR 1.21; 95% CI 1.00 to 1.45).
  • Composite RIFLE-defined AKI rates (RIFLE-R/I) were lower with HES, illustrating the interpretive complexity of urine-output-and-creatinine composite kidney endpoints when fluids alter physiology.
  • HES exposure produced modest fluid-sparing and haemodynamic separation (e.g., CVP 11.3 ± 4.8 vs 10.4 ± 4.4 mm Hg during the first 4 days) without translating into better survival or shorter ICU stay.
  • CHEST, together with subsequent RCTs, meta-analyses, and guidelines, contributed to a major shift away from starch-based resuscitation in critical illness.

Overall Takeaway

CHEST is a landmark ICU fluid trial because it provided definitive, high-quality evidence that “modern” tetrastarch (HES 130/0.4) did not improve survival compared with saline and was associated with an increased need for renal-replacement therapy. Together with concordant RCTs, meta-analyses, and subsequent guidelines, it reshaped global resuscitation practice away from starch-based colloids in critical illness.

Overall Summary

  • In 7000 ICU patients, HES 130/0.4 did not reduce 90-day mortality and increased use of renal-replacement therapy compared with saline.
  • CHEST, alongside subsequent evidence, underpins guideline recommendations to avoid HES for resuscitation in critical illness.

Bibliography